From 444fa80095a0f786c1d7c09d393e8b28bc14976d Mon Sep 17 00:00:00 2001
From: pat-alt <altmeyerpat@gmail.com>
Date: Tue, 21 Mar 2023 08:13:17 +0100
Subject: [PATCH] fair to say I hate latex
---
.gitignore | 219 ++-
bib.bib | 2428 ------------------------
paper/paper.pdf | Bin 150794 -> 145800 bytes
paper/paper.tex | 4 +-
paper/references.bib => references.bib | 0
5 files changed, 214 insertions(+), 2437 deletions(-)
delete mode 100644 bib.bib
rename paper/references.bib => references.bib (100%)
diff --git a/.gitignore b/.gitignore
index b7d8c1d6..8f59f64d 100644
--- a/.gitignore
+++ b/.gitignore
@@ -2,9 +2,216 @@
/Manifest.toml
-**/paper.aux
-**/paper.fdb_latexmk
-**/paper.fls
-**/paper.log
-**/paper.out
-**/paper.synctex.gz
\ No newline at end of file
+# Tex
+
+## Core latex/pdflatex auxiliary files:
+*.aux
+*.lof
+*.log
+*.lot
+*.fls
+*.out
+*.toc
+*.fmt
+*.fot
+*.cb
+*.cb2
+.*.lb
+
+## Intermediate documents:
+*.dvi
+*.xdv
+*-converted-to.*
+# these rules might exclude image files for figures etc.
+# *.ps
+# *.eps
+# *.pdf
+
+## Generated if empty string is given at "Please type another file name for output:"
+.pdf
+
+## Bibliography auxiliary files (bibtex/biblatex/biber):
+*.bbl
+*.bcf
+*.blg
+*-blx.aux
+*-blx.bib
+*.run.xml
+
+## Build tool auxiliary files:
+*.fdb_latexmk
+*.synctex
+*.synctex(busy)
+*.synctex.gz
+*.synctex.gz(busy)
+*.pdfsync
+
+## Build tool directories for auxiliary files
+# latexrun
+latex.out/
+
+## Auxiliary and intermediate files from other packages:
+# algorithms
+*.alg
+*.loa
+
+# achemso
+acs-*.bib
+
+# amsthm
+*.thm
+
+# beamer
+*.nav
+*.pre
+*.snm
+*.vrb
+
+# changes
+*.soc
+
+# comment
+*.cut
+
+# cprotect
+*.cpt
+
+# elsarticle (documentclass of Elsevier journals)
+*.spl
+
+# endnotes
+*.ent
+
+# fixme
+*.lox
+
+# feynmf/feynmp
+*.mf
+*.mp
+*.t[1-9]
+*.t[1-9][0-9]
+*.tfm
+
+#(r)(e)ledmac/(r)(e)ledpar
+*.end
+*.?end
+*.[1-9]
+*.[1-9][0-9]
+*.[1-9][0-9][0-9]
+*.[1-9]R
+*.[1-9][0-9]R
+*.[1-9][0-9][0-9]R
+*.eledsec[1-9]
+*.eledsec[1-9]R
+*.eledsec[1-9][0-9]
+*.eledsec[1-9][0-9]R
+*.eledsec[1-9][0-9][0-9]
+*.eledsec[1-9][0-9][0-9]R
+
+# glossaries
+*.acn
+*.acr
+*.glg
+*.glo
+*.gls
+*.glsdefs
+*.lzo
+*.lzs
+*.slg
+*.slo
+*.sls
+
+# uncomment this for glossaries-extra (will ignore makeindex's style files!)
+# *.ist
+
+# gnuplot
+*.gnuplot
+*.table
+
+# gnuplottex
+*-gnuplottex-*
+
+# gregoriotex
+*.gaux
+*.glog
+*.gtex
+
+# htlatex
+*.4ct
+*.4tc
+*.idv
+*.lg
+*.trc
+*.xref
+
+# hyperref
+*.brf
+
+# knitr
+*-concordance.tex
+# TODO Uncomment the next line if you use knitr and want to ignore its generated tikz files
+# *.tikz
+*-tikzDictionary
+
+# listings
+*.lol
+
+# luatexja-ruby
+*.ltjruby
+
+# makeidx
+*.idx
+*.ilg
+*.ind
+
+# minitoc
+*.maf
+*.mlf
+*.mlt
+*.mtc[0-9]*
+*.slf[0-9]*
+*.slt[0-9]*
+*.stc[0-9]*
+
+# minted
+_minted*
+*.pyg
+
+# morewrites
+*.mw
+
+# newpax
+*.newpax
+
+# nomencl
+*.nlg
+*.nlo
+*.nls
+
+# pax
+*.pax
+
+# pdfpcnotes
+*.pdfpc
+
+# sagetex
+*.sagetex.sage
+*.sagetex.py
+*.sagetex.scmd
+
+# scrwfile
+*.wrt
+
+# svg
+svg-inkscape/
+
+# sympy
+*.sout
+*.sympy
+sympy-plots-for-*.tex/
+
+# pdfcomment
+*.upa
+*.upb
+
+# pythontex
\ No newline at end of file
diff --git a/bib.bib b/bib.bib
deleted file mode 100644
index 6aab437e..00000000
--- a/bib.bib
+++ /dev/null
@@ -1,2428 +0,0 @@
-@InProceedings{altmeyer2023endogenous,
- author = {Altmeyer, Patrick and Angela, Giovan and Buszydlik, Aleksander and Dobiczek, Karol and van Deursen, Arie and Liem, Cynthia},
- booktitle = {First {IEEE} {Conference} on {Secure} and {Trustworthy} {Machine} {Learning}},
- date = {2023},
- title = {Endogenous {Macrodynamics} in {Algorithmic} {Recourse}},
- file = {:altmeyerendogenous - Endogenous Macrodynamics in Algorithmic Recourse.pdf:PDF},
-}
-
-%% This BibTeX bibliography file was created using BibDesk.
-%% https://bibdesk.sourceforge.io/
-
-%% Created for Patrick Altmeyer at 2022-12-13 12:58:22 +0100
-
-
-%% Saved with string encoding Unicode (UTF-8)
-
-
-
-@article{abadie2002instrumental,
- author = {Abadie, Alberto and Angrist, Joshua and Imbens, Guido},
- date = {2002},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Econometrica : journal of the Econometric Society},
- number = {1},
- pages = {91--117},
- shortjournal = {Econometrica},
- title = {Instrumental Variables Estimates of the Effect of Subsidized Training on the Quantiles of Trainee Earnings},
- volume = {70}}
-
-@article{abadie2003economic,
- author = {Abadie, Alberto and Gardeazabal, Javier},
- date = {2003},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {American economic review},
- number = {1},
- pages = {113--132},
- title = {The Economic Costs of Conflict: {{A}} Case Study of the {{Basque Country}}},
- volume = {93}}
-
-@inproceedings{ackerman2021machine,
- author = {Ackerman, Samuel and Dube, Parijat and Farchi, Eitan and Raz, Orna and Zalmanovici, Marcel},
- booktitle = {2021 {{IEEE}}/{{ACM Third International Workshop}} on {{Deep Learning}} for {{Testing}} and {{Testing}} for {{Deep Learning}} ({{DeepTest}})},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {1--8},
- publisher = {{IEEE}},
- title = {Machine {{Learning Model Drift Detection Via Weak Data Slices}}}}
-
-@article{allen2017referencedependent,
- author = {Allen, Eric J and Dechow, Patricia M and Pope, Devin G and Wu, George},
- date = {2017},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Management Science},
- number = {6},
- pages = {1657--1672},
- title = {Reference-Dependent Preferences: {{Evidence}} from Marathon Runners},
- volume = {63}}
-
-@article{altmeyer2018option,
- author = {Altmeyer, Patrick and Grapendal, Jacob Daniel and Pravosud, Makar and Quintana, Gand Derry},
- date = {2018},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Option Pricing in the {{Heston}} Stochastic Volatility Model: An Empirical Evaluation}}
-
-@article{altmeyer2021deep,
- author = {Altmeyer, Patrick and Agusti, Marc and Vidal-Quadras Costa, Ignacio},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Deep {{Vector Autoregression}} for {{Macroeconomic Data}}},
- url = {https://thevoice.bse.eu/wp-content/uploads/2021/07/ds21-project-agusti-et-al.pdf},
- bdsk-url-1 = {https://thevoice.bse.eu/wp-content/uploads/2021/07/ds21-project-agusti-et-al.pdf}}
-
-@book{altmeyer2021deepvars,
- author = {Altmeyer, Patrick},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Deepvars: {{Deep Vector Autoregession}}}}
-
-@misc{altmeyer2022counterfactualexplanations,
- author = {Altmeyer, Patrick},
- date = {2022},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {{{CounterfactualExplanations}}.Jl - a {{Julia}} Package for {{Counterfactual Explanations}} and {{Algorithmic Recourse}}},
- url = {https://github.com/pat-alt/CounterfactualExplanations.jl},
- bdsk-url-1 = {https://github.com/pat-alt/CounterfactualExplanations.jl}}
-
-@software{altmeyerCounterfactualExplanationsJlJulia2022,
- author = {Altmeyer, Patrick},
- date = {2022},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {{{CounterfactualExplanations}}.Jl - a {{Julia}} Package for {{Counterfactual Explanations}} and {{Algorithmic Recourse}}},
- url = {https://github.com/pat-alt/CounterfactualExplanations.jl},
- version = {0.1.2},
- bdsk-url-1 = {https://github.com/pat-alt/CounterfactualExplanations.jl}}
-
-@unpublished{angelopoulos2021gentle,
- archiveprefix = {arXiv},
- author = {Angelopoulos, Anastasios N. and Bates, Stephen},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2107.07511},
- eprinttype = {arxiv},
- file = {/Users/FA31DU/Zotero/storage/RKSUMYZG/Angelopoulos and Bates - 2021 - A gentle introduction to conformal prediction and .pdf;/Users/FA31DU/Zotero/storage/PRUEKRR3/2107.html},
- title = {A Gentle Introduction to Conformal Prediction and Distribution-Free Uncertainty Quantification}}
-
-@Misc{angelopoulos2022uncertainty,
- author = {Angelopoulos, Anastasios and Bates, Stephen and Malik, Jitendra and Jordan, Michael I.},
- date = {2022-09-03},
- title = {Uncertainty {{Sets}} for {{Image Classifiers}} Using {{Conformal Prediction}}},
- eprint = {2009.14193},
- eprinttype = {arxiv},
- url = {http://arxiv.org/abs/2009.14193},
- urldate = {2022-12-07},
- abstract = {Convolutional image classifiers can achieve high predictive accuracy, but quantifying their uncertainty remains an unresolved challenge, hindering their deployment in consequential settings. Existing uncertainty quantification techniques, such as Platt scaling, attempt to calibrate the network's probability estimates, but they do not have formal guarantees. We present an algorithm that modifies any classifier to output a predictive set containing the true label with a user-specified probability, such as 90\%. The algorithm is simple and fast like Platt scaling, but provides a formal finite-sample coverage guarantee for every model and dataset. Our method modifies an existing conformal prediction algorithm to give more stable predictive sets by regularizing the small scores of unlikely classes after Platt scaling. In experiments on both Imagenet and Imagenet-V2 with ResNet-152 and other classifiers, our scheme outperforms existing approaches, achieving coverage with sets that are often factors of 5 to 10 smaller than a stand-alone Platt scaling baseline.},
- archiveprefix = {arXiv},
- bdsk-url-1 = {http://arxiv.org/abs/2009.14193},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- file = {/Users/FA31DU/Zotero/storage/5BYIRBR2/Angelopoulos et al. - 2022 - Uncertainty Sets for Image Classifiers using Confo.pdf;/Users/FA31DU/Zotero/storage/2QJAKFKV/2009.html},
- keywords = {Computer Science - Computer Vision and Pattern Recognition, Mathematics - Statistics Theory, Statistics - Machine Learning},
- number = {arXiv:2009.14193},
- primaryclass = {cs, math, stat},
- publisher = {{arXiv}},
-}
-
-@article{angelucci2009indirect,
- author = {Angelucci, Manuela and De Giorgi, Giacomo},
- date = {2009},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {American economic review},
- number = {1},
- pages = {486--508},
- title = {Indirect Effects of an Aid Program: How Do Cash Transfers Affect Ineligibles' Consumption?},
- volume = {99}}
-
-@article{angrist1990lifetime,
- author = {Angrist, Joshua D},
- date = {1990},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The American Economic Review},
- pages = {313--336},
- title = {Lifetime Earnings and the {{Vietnam}} Era Draft Lottery: Evidence from Social Security Administrative Records}}
-
-@unpublished{antoran2020getting,
- archiveprefix = {arXiv},
- author = {Antor{\'a}n, Javier and Bhatt, Umang and Adel, Tameem and Weller, Adrian and Hern{\'a}ndez-Lobato, Jos{\'e} Miguel},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2006.06848},
- eprinttype = {arxiv},
- title = {Getting a Clue: {{A}} Method for Explaining Uncertainty Estimates}}
-
-@article{arcones1992bootstrap,
- author = {Arcones, Miguel A and Gine, Evarist},
- date = {1992},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The Annals of Statistics},
- pages = {655--674},
- title = {On the Bootstrap of {{U}} and {{V}} Statistics}}
-
-@article{ariely2003coherent,
- author = {Ariely, Dan and Loewenstein, George and Prelec, Drazen},
- date = {2003},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The Quarterly journal of economics},
- number = {1},
- pages = {73--106},
- title = {``{{Coherent}} Arbitrariness'': {{Stable}} Demand Curves without Stable Preferences},
- volume = {118}}
-
-@article{ariely2006tom,
- author = {Ariely, Dan and Loewenstein, George and Prelec, Drazen},
- date = {2006},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Economic Behavior \& Organization},
- number = {1},
- pages = {1--10},
- title = {Tom {{Sawyer}} and the Construction of Value},
- volume = {60}}
-
-@article{arrieta2020explainable,
- author = {Arrieta, Alejandro Barredo and Diaz-Rodriguez, Natalia and Del Ser, Javier and Bennetot, Adrien and Tabik, Siham and Barbado, Alberto and Garcia, Salvador and Gil-Lopez, Sergio and Molina, Daniel and Benjamins, Richard and others},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Information Fusion},
- pages = {82--115},
- title = {Explainable {{Artificial Intelligence}} ({{XAI}}): {{Concepts}}, Taxonomies, Opportunities and Challenges toward Responsible {{AI}}},
- volume = {58}}
-
-@article{auer2002finitetime,
- author = {Auer, Peter and Cesa-Bianchi, Nicolo and Fischer, Paul},
- date = {2002},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Machine learning},
- number = {2},
- pages = {235--256},
- title = {Finite-Time Analysis of the Multiarmed Bandit Problem},
- volume = {47}}
-
-@article{barabasi2016network,
- author = {Barab{\'a}si, Albert-L{\'a}szl{\'o}},
- date = {2016},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Network Science},
- title = {Network {{Science}}}}
-
-@unpublished{bastounis2021mathematics,
- archiveprefix = {arXiv},
- author = {Bastounis, Alexander and Hansen, Anders C and Vla{\v c}i{\'c}, Verner},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2109.06098},
- eprinttype = {arxiv},
- title = {The Mathematics of Adversarial Attacks in {{AI}}--{{Why}} Deep Learning Is Unstable despite the Existence of Stable Neural Networks}}
-
-@article{bechara1997deciding,
- author = {Bechara, Antoine and Damasio, Hanna and Tranel, Daniel and Damasio, Antonio R},
- date = {1997},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Science (New York, N.Y.)},
- number = {5304},
- pages = {1293--1295},
- shortjournal = {Science},
- title = {Deciding Advantageously before Knowing the Advantageous Strategy},
- volume = {275}}
-
-@book{berlinet2011reproducing,
- author = {Berlinet, Alain and Thomas-Agnan, Christine},
- date = {2011},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Springer Science \& Business Media}},
- title = {Reproducing Kernel {{Hilbert}} Spaces in Probability and Statistics}}
-
-@misc{bernanke1990federal,
- author = {Bernanke, Ben S},
- date = {1990},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{National Bureau of Economic Research Cambridge, Mass., USA}},
- title = {The Federal Funds Rate and the Channels of Monetary Transnission}}
-
-@article{besbes2014stochastic,
- author = {Besbes, Omar and Gur, Yonatan and Zeevi, Assaf},
- date = {2014},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Advances in neural information processing systems},
- pages = {199--207},
- title = {Stochastic Multi-Armed-Bandit Problem with Non-Stationary Rewards},
- volume = {27}}
-
-@article{bholat2020impact,
- author = {Bholat, D and Gharbawi, M and Thew, O},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Bank of England Quarterly Bulletin, Q4},
- title = {The {{Impact}} of {{Covid}} on {{Machine Learning}} and {{Data Science}} in {{UK Banking}}}}
-
-@book{bishop2006pattern,
- author = {Bishop, Christopher M},
- date = {2006},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{springer}},
- title = {Pattern Recognition and Machine Learning}}
-
-@article{blaom2020mlj,
- abstract = {Blaom et al., (2020). MLJ: A Julia package for composable machine learning. Journal of Open Source Software, 5(55), 2704, https://doi.org/10.21105/joss.02704},
- author = {Blaom, Anthony D. and Kiraly, Franz and Lienart, Thibaut and Simillides, Yiannis and Arenas, Diego and Vollmer, Sebastian J.},
- date = {2020-11-07},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- doi = {10.21105/joss.02704},
- file = {/Users/FA31DU/Zotero/storage/7AY87FGP/Blaom et al. - 2020 - MLJ A Julia package for composable machine learni.pdf;/Users/FA31DU/Zotero/storage/D69YSMVF/joss.html},
- issn = {2475-9066},
- journaltitle = {Journal of Open Source Software},
- langid = {english},
- number = {55},
- pages = {2704},
- shorttitle = {{{MLJ}}},
- title = {{{MLJ}}: {{A Julia}} Package for Composable Machine Learning},
- url = {https://joss.theoj.org/papers/10.21105/joss.02704},
- urldate = {2022-10-27},
- volume = {5},
- bdsk-url-1 = {https://joss.theoj.org/papers/10.21105/joss.02704},
- bdsk-url-2 = {https://doi.org/10.21105/joss.02704}}
-
-@inproceedings{blundell2015weight,
- author = {Blundell, Charles and Cornebise, Julien and Kavukcuoglu, Koray and Wierstra, Daan},
- booktitle = {International Conference on Machine Learning},
- date = {2015},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {1613--1622},
- publisher = {{PMLR}},
- title = {Weight Uncertainty in Neural Network}}
-
-@article{borch2022machine,
- author = {Borch, Christian},
- date = {2022},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Technology in Society},
- pages = {101852},
- title = {Machine Learning, Knowledge Risk, and Principal-Agent Problems in Automated Trading}}
-
-@unpublished{borisov2021deep,
- archiveprefix = {arXiv},
- author = {Borisov, Vadim and Leemann, Tobias and Se{\ss}ler, Kathrin and Haug, Johannes and Pawelczyk, Martin and Kasneci, Gjergji},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2110.01889},
- eprinttype = {arxiv},
- title = {Deep Neural Networks and Tabular Data: {{A}} Survey}}
-
-@article{bramoulle2009identification,
- author = {Bramoull{\'e}, Yann and Djebbari, Habiba and Fortin, Bernard},
- date = {2009},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of econometrics},
- number = {1},
- pages = {41--55},
- title = {Identification of Peer Effects through Social Networks},
- volume = {150}}
-
-@article{bramoulle2020peer,
- author = {Bramoull{\'e}, Yann and Djebbari, Habiba and Fortin, Bernard},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Annual Review of Economics},
- pages = {603--629},
- title = {Peer Effects in Networks: {{A}} Survey},
- volume = {12}}
-
-@unpublished{branco2015survey,
- archiveprefix = {arXiv},
- author = {Branco, Paula and Torgo, Luis and Ribeiro, Rita},
- date = {2015},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1505.01658},
- eprinttype = {arxiv},
- title = {A Survey of Predictive Modelling under Imbalanced Distributions}}
-
-@book{brock1991nonlinear,
- author = {Brock, William Allen and Brock, William A and Hsieh, David Arthur and LeBaron, Blake Dean and Brock, William E},
- date = {1991},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{MIT press}},
- title = {Nonlinear Dynamics, Chaos, and Instability: Statistical Theory and Economic Evidence}}
-
-@inproceedings{buolamwini2018gender,
- author = {Buolamwini, Joy and Gebru, Timnit},
- booktitle = {Conference on Fairness, Accountability and Transparency},
- date = {2018},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {77--91},
- publisher = {{PMLR}},
- title = {Gender Shades: {{Intersectional}} Accuracy Disparities in Commercial Gender Classification}}
-
-@unpublished{bussmann2020neural,
- archiveprefix = {arXiv},
- author = {Bussmann, Bart and Nys, Jannes and Latr{\'e}, Steven},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2010.09429},
- eprinttype = {arxiv},
- title = {Neural {{Additive Vector Autoregression Models}} for {{Causal Discovery}} in {{Time Series Data}}}}
-
-@report{card1993minimum,
- author = {Card, David and Krueger, Alan B},
- date = {1993},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- institution = {{National Bureau of Economic Research}},
- title = {Minimum Wages and Employment: {{A}} Case Study of the Fast Food Industry in {{New Jersey}} and {{Pennsylvania}}}}
-
-@inproceedings{carlini2017evaluating,
- author = {Carlini, Nicholas and Wagner, David},
- booktitle = {2017 Ieee Symposium on Security and Privacy (Sp)},
- date = {2017},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {39--57},
- publisher = {{IEEE}},
- title = {Towards Evaluating the Robustness of Neural Networks}}
-
-@article{carlisle2019racist,
- author = {Carlisle, M.},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Racist Data Destruction? - a {{Boston}} Housing Dataset Controversy},
- url = {https://medium.com/@docintangible/racist-data-destruction-113e3eff54a8},
- bdsk-url-1 = {https://medium.com/@docintangible/racist-data-destruction-113e3eff54a8}}
-
-@article{carrell2009does,
- author = {Carrell, Scott E and Fullerton, Richard L and West, James E},
- date = {2009},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Labor Economics},
- number = {3},
- pages = {439--464},
- title = {Does Your Cohort Matter? {{Measuring}} Peer Effects in College Achievement},
- volume = {27}}
-
-@article{carrell2013natural,
- author = {Carrell, Scott E and Sacerdote, Bruce I and West, James E},
- date = {2013},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Econometrica : journal of the Econometric Society},
- number = {3},
- pages = {855--882},
- shortjournal = {Econometrica},
- title = {From Natural Variation to Optimal Policy? {{The}} Importance of Endogenous Peer Group Formation},
- volume = {81}}
-
-@article{carrizosa2021generating,
- author = {Carrizosa, Emilio and Ramırez-Ayerbe, Jasone and Romero, Dolores},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Generating {{Collective Counterfactual Explanations}} in {{Score-Based Classification}} via {{Mathematical Optimization}}}}
-
-@article{cascarino2022explainable,
- author = {Cascarino, Giuseppe and Moscatelli, Mirko and Parlapiano, Fabio},
- date = {2022},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Bank of Italy Occasional Paper},
- number = {674},
- title = {Explainable {{Artificial Intelligence}}: Interpreting Default Forecasting Models Based on {{Machine Learning}}}}
-
-@article{chandola2009anomaly,
- author = {Chandola, Varun and Banerjee, Arindam and Kumar, Vipin},
- date = {2009},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {ACM computing surveys (CSUR)},
- number = {3},
- pages = {1--58},
- title = {Anomaly Detection: {{A}} Survey},
- volume = {41}}
-
-@article{chapelle2011empirical,
- author = {Chapelle, Olivier and Li, Lihong},
- date = {2011},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Advances in neural information processing systems},
- pages = {2249--2257},
- title = {An Empirical Evaluation of Thompson Sampling},
- volume = {24}}
-
-@article{chetty2011adjustment,
- author = {Chetty, Raj and Friedman, John N and Olsen, Tore and Pistaferri, Luigi},
- date = {2011},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The quarterly journal of economics},
- number = {2},
- pages = {749--804},
- title = {Adjustment Costs, Firm Responses, and Micro vs. Macro Labor Supply Elasticities: {{Evidence}} from {{Danish}} Tax Records},
- volume = {126}}
-
-@article{cortes1995supportvector,
- author = {Cortes, Corinna and Vapnik, Vladimir},
- date = {1995},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Machine learning},
- number = {3},
- pages = {273--297},
- title = {Support-Vector Networks},
- volume = {20}}
-
-@article{crawford2019variable,
- author = {Crawford, Lorin and Flaxman, Seth R and Runcie, Daniel E and West, Mike},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The annals of applied statistics},
- number = {2},
- pages = {958},
- title = {Variable Prioritization in Nonlinear Black Box Methods: {{A}} Genetic Association Case Study},
- volume = {13}}
-
-@inproceedings{dai2022counterfactual,
- author = {Dai, Xinyue and Keane, Mark T and Shalloo, Laurence and Ruelle, Elodie and Byrne, Ruth MJ},
- date = {2022},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eventtitle = {Proceedings of the 2022 {{AAAI}}/{{ACM Conference}} on {{AI}}, {{Ethics}}, and {{Society}}},
- pages = {215--226},
- title = {Counterfactual Explanations for Prediction and Diagnosis in Xai}}
-
-@article{danielsson2021artificial,
- author = {Danielsson, Jon and Macrae, Robert and Uthemann, Andreas},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Banking \& Finance},
- pages = {106290},
- title = {Artificial Intelligence and Systemic Risk}}
-
-@article{daxberger2021laplace,
- author = {Daxberger, Erik and Kristiadi, Agustinus and Immer, Alexander and Eschenhagen, Runa and Bauer, Matthias and Hennig, Philipp},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Advances in Neural Information Processing Systems},
- title = {Laplace {{Redux-Effortless Bayesian Deep Learning}}},
- volume = {34}}
-
-@article{dehejia1999causal,
- author = {Dehejia, Rajeev H and Wahba, Sadek},
- date = {1999},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of the American statistical Association},
- number = {448},
- pages = {1053--1062},
- title = {Causal Effects in Nonexperimental Studies: {{Reevaluating}} the Evaluation of Training Programs},
- volume = {94}}
-
-@article{dell2010persistent,
- author = {Dell, Melissa},
- date = {2010},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Econometrica : journal of the Econometric Society},
- number = {6},
- pages = {1863--1903},
- shortjournal = {Econometrica},
- title = {The Persistent Effects of {{Peru}}'s Mining Mita},
- volume = {78}}
-
-@article{denhengst2020reinforcement,
- author = {den Hengst, Floris and Grua, Eoin Martino and el Hassouni, Ali and Hoogendoorn, Mark},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- issue = {Preprint},
- journaltitle = {Data Science},
- options = {useprefix=true},
- pages = {1--41},
- title = {Reinforcement Learning for Personalization: {{A}} Systematic Literature Review}}
-
-@article{deoliveira2021framework,
- author = {de Oliveira, Raphael Mazzine Barbosa and Martens, David},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Applied Sciences},
- number = {16},
- options = {useprefix=true},
- pages = {7274},
- title = {A Framework and Benchmarking Study for Counterfactual Generating Methods on Tabular Data},
- volume = {11}}
-
-@article{dhurandhar2018explanations,
- author = {Dhurandhar, Amit and Chen, Pin-Yu and Luss, Ronny and Tu, Chun-Chen and Ting, Paishun and Shanmugam, Karthikeyan and Das, Payel},
- date = {2018},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Advances in neural information processing systems},
- title = {Explanations Based on the Missing: {{Towards}} Contrastive Explanations with Pertinent Negatives},
- volume = {31}}
-
-@inproceedings{dombrowski2021diffeomorphic,
- author = {Dombrowski, Ann-Kathrin and Gerken, Jan E and Kessel, Pan},
- booktitle = {{{ICML Workshop}} on {{Invertible Neural Networks}}, {{Normalizing Flows}}, and {{Explicit Likelihood Models}}},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Diffeomorphic Explanations with Normalizing Flows}}
-
-@inproceedings{dorffner1996neural,
- author = {Dorffner, Georg},
- booktitle = {Neural Network World},
- date = {1996},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Citeseer}},
- title = {Neural Networks for Time Series Processing}}
-
-@article{epstein1979stability,
- author = {Epstein, Seymour},
- date = {1979},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of personality and social psychology},
- number = {7},
- pages = {1097},
- title = {The Stability of Behavior: {{I}}. {{On}} Predicting Most of the People Much of the Time.},
- volume = {37}}
-
-@Online{barocas2022fairness,
- author = {Solon Barocas and Moritz Hardt and Arvind Narayanan},
- date = {2022-12-23},
- title = {Fairness and Machine Learning},
- url = {https://fairmlbook.org/index.html},
- urldate = {2022-11-08},
- bdsk-url-1 = {https://fairmlbook.org/index.html},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
-}
-
-@article{falk2006clean,
- author = {Falk, Armin and Ichino, Andrea},
- date = {2006},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of labor economics},
- number = {1},
- pages = {39--57},
- title = {Clean Evidence on Peer Effects},
- volume = {24}}
-
-@unpublished{fan2020interpretability,
- archiveprefix = {arXiv},
- author = {Fan, Fenglei and Xiong, Jinjun and Wang, Ge},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2001.02522},
- eprinttype = {arxiv},
- title = {On Interpretability of Artificial Neural Networks}}
-
-@article{fang2011dynamic,
- author = {Fang, Hanming and Gavazza, Alessandro},
- date = {2011},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {American Economic Review},
- number = {7},
- pages = {3047--77},
- title = {Dynamic Inefficiencies in an Employment-Based Health Insurance System: {{Theory}} and Evidence},
- volume = {101}}
-
-@article{fehr2000cooperation,
- author = {Fehr, Ernst and Gachter, Simon},
- date = {2000},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {American Economic Review},
- number = {4},
- pages = {980--994},
- title = {Cooperation and Punishment in Public Goods Experiments},
- volume = {90}}
-
-@article{fix1951important,
- author = {Fix, E and Hodges, J},
- date = {1951},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {International Statistical Review},
- number = {57},
- pages = {233--238},
- title = {An Important Contribution to Nonparametric Discriminant Analysis and Density Estimation},
- volume = {3}}
-
-@book{friedman2008monetary,
- author = {Friedman, Milton and Schwartz, Anna Jacobson},
- date = {2008},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Princeton University Press}},
- title = {A Monetary History of the {{United States}}, 1867-1960},
- volume = {14}}
-
-@inproceedings{gal2016dropout,
- author = {Gal, Yarin and Ghahramani, Zoubin},
- booktitle = {International Conference on Machine Learning},
- date = {2016},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {1050--1059},
- publisher = {{PMLR}},
- title = {Dropout as a Bayesian Approximation: {{Representing}} Model Uncertainty in Deep Learning}}
-
-@inproceedings{gal2017deep,
- author = {Gal, Yarin and Islam, Riashat and Ghahramani, Zoubin},
- booktitle = {International {{Conference}} on {{Machine Learning}}},
- date = {2017},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {1183--1192},
- publisher = {{PMLR}},
- title = {Deep Bayesian Active Learning with Image Data}}
-
-@article{galizzi2019external,
- author = {Galizzi, Matteo M and Navarro-Martinez, Daniel},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Management Science},
- number = {3},
- pages = {976--1002},
- title = {On the External Validity of Social Preference Games: A Systematic Lab-Field Study},
- volume = {65}}
-
-@article{gama2014survey,
- author = {Gama, Jo{\~a}o and {\v Z}liobait{\.e}, Indr{\.e} and Bifet, Albert and Pechenizkiy, Mykola and Bouchachia, Abdelhamid},
- date = {2014},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {ACM computing surveys (CSUR)},
- number = {4},
- pages = {1--37},
- title = {A Survey on Concept Drift Adaptation},
- volume = {46}}
-
-@unpublished{garivier2008upperconfidence,
- archiveprefix = {arXiv},
- author = {Garivier, Aur{\'e}lien and Moulines, Eric},
- date = {2008},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {0805.3415},
- eprinttype = {arxiv},
- title = {On Upper-Confidence Bound Policies for Non-Stationary Bandit Problems}}
-
-@book{gelman2013bayesian,
- author = {Gelman, Andrew and Carlin, John B and Stern, Hal S and Dunson, David B and Vehtari, Aki and Rubin, Donald B},
- date = {2013},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{CRC press}},
- title = {Bayesian Data Analysis}}
-
-@article{gilbert1998immune,
- author = {Gilbert, Daniel T and Pinel, Elizabeth C and Wilson, Timothy D and Blumberg, Stephen J and Wheatley, Thalia P},
- date = {1998},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of personality and social psychology},
- number = {3},
- pages = {617},
- title = {Immune Neglect: A Source of Durability Bias in Affective Forecasting.},
- volume = {75}}
-
-@article{gneezy2006uncertainty,
- author = {Gneezy, Uri and List, John A and Wu, George},
- date = {2006},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The Quarterly Journal of Economics},
- number = {4},
- pages = {1283--1309},
- title = {The Uncertainty Effect: {{When}} a Risky Prospect Is Valued Less than Its Worst Possible Outcome},
- volume = {121}}
-
-@incollection{goan2020bayesian,
- author = {Goan, Ethan and Fookes, Clinton},
- booktitle = {Case {{Studies}} in {{Applied Bayesian Data Science}}},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {45--87},
- publisher = {{Springer}},
- title = {Bayesian {{Neural Networks}}: {{An Introduction}} and {{Survey}}}}
-
-@article{goldsmith-pinkham2013social,
- author = {Goldsmith-Pinkham, Paul and Imbens, Guido W},
- date = {2013},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Business \& Economic Statistics},
- number = {3},
- pages = {253--264},
- title = {Social Networks and the Identification of Peer Effects},
- volume = {31}}
-
-@unpublished{goodfellow2014explaining,
- archiveprefix = {arXiv},
- author = {Goodfellow, Ian J and Shlens, Jonathon and Szegedy, Christian},
- date = {2014},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1412.6572},
- eprinttype = {arxiv},
- title = {Explaining and Harnessing Adversarial Examples}}
-
-@book{goodfellow2016deep,
- author = {Goodfellow, Ian and Bengio, Yoshua and Courville, Aaron},
- date = {2016},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{MIT Press}},
- title = {Deep {{Learning}}}}
-
-@article{goodfriend2005incredible,
- author = {Goodfriend, Marvin and King, Robert G},
- date = {2005},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Monetary Economics},
- number = {5},
- pages = {981--1015},
- title = {The Incredible {{Volcker}} Disinflation},
- volume = {52}}
-
-@article{graham2017econometric,
- author = {Graham, Bryan S},
- date = {2017},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Econometrica : journal of the Econometric Society},
- number = {4},
- pages = {1033--1063},
- shortjournal = {Econometrica},
- title = {An Econometric Model of Network Formation with Degree Heterogeneity},
- volume = {85}}
-
-@article{greene2012econometric,
- author = {Greene, William H},
- date = {2012},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Stern School of Business, New York University},
- title = {Econometric Analysis, 71e}}
-
-@article{grether1979economic,
- author = {Grether, David M and Plott, Charles R},
- date = {1979},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The American Economic Review},
- number = {4},
- pages = {623--638},
- title = {Economic Theory of Choice and the Preference Reversal Phenomenon},
- volume = {69}}
-
-@article{gretton2012kernel,
- author = {Gretton, Arthur and Borgwardt, Karsten M and Rasch, Malte J and Sch{\"o}lkopf, Bernhard and Smola, Alexander},
- date = {2012},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The Journal of Machine Learning Research},
- number = {1},
- pages = {723--773},
- title = {A Kernel Two-Sample Test},
- volume = {13}}
-
-@unpublished{griffith2020name,
- author = {Griffith, Alan},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Name {{Your Friends}}, but {{Only Five}}? {{The Importance}} of {{Censoring}} in {{Peer Effects Estimates}} Using {{Social Network Data}}}}
-
-@unpublished{grinsztajn2022why,
- archiveprefix = {arXiv},
- author = {Grinsztajn, L{\'e}o and Oyallon, Edouard and Varoquaux, Ga{\"e}l},
- date = {2022},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2207.08815},
- eprinttype = {arxiv},
- title = {Why Do Tree-Based Models Still Outperform Deep Learning on Tabular Data?}}
-
-@misc{group2020detailed,
- author = {Group, Open COVID-19 Data Working},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Detailed {{Epidemiological Data}} from the {{COVID-19 Outbreak}}}}
-
-@inproceedings{gupta2011thompson,
- author = {Gupta, Neha and Granmo, Ole-Christoffer and Agrawala, Ashok},
- booktitle = {2011 10th {{International Conference}} on {{Machine Learning}} and {{Applications}} and {{Workshops}}},
- date = {2011},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {484--489},
- publisher = {{IEEE}},
- title = {Thompson Sampling for Dynamic Multi-Armed Bandits},
- volume = {1}}
-
-@book{hamilton2020time,
- author = {Hamilton, James Douglas},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Princeton university press}},
- title = {Time Series Analysis}}
-
-@article{hamon2020robustness,
- author = {Hamon, Ronan and Junklewitz, Henrik and Sanchez, Ignacio},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Publications Office of the European Union},
- title = {Robustness and Explainability of Artificial Intelligence}}
-
-@article{hamzacebi2008improving,
- author = {Hamza{\c c}ebi, Co{\c s}kun},
- date = {2008},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Information Sciences},
- number = {23},
- pages = {4550--4559},
- title = {Improving Artificial Neural Networks' Performance in Seasonal Time Series Forecasting},
- volume = {178}}
-
-@inproceedings{hanneke2007bound,
- author = {Hanneke, Steve},
- booktitle = {Proceedings of the 24th International Conference on {{Machine}} Learning},
- date = {2007},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {353--360},
- title = {A Bound on the Label Complexity of Agnostic Active Learning}}
-
-@article{hansen2020virtue,
- author = {Hansen, Kristian Bondo},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Big Data \& Society},
- number = {1},
- pages = {2053951720926558},
- title = {The Virtue of Simplicity: {{On}} Machine Learning Models in Algorithmic Trading},
- volume = {7}}
-
-@article{hartland2006multiarmed,
- author = {Hartland, C{\'e}dric and Gelly, Sylvain and Baskiotis, Nicolas and Teytaud, Olivier and Sebag, Michele},
- date = {2006},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Multi-Armed Bandit, Dynamic Environments and Meta-Bandits}}
-
-@article{heckman1985alternative,
- author = {Heckman, James J and Robb Jr, Richard},
- date = {1985},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of econometrics},
- number = {1-2},
- pages = {239--267},
- title = {Alternative Methods for Evaluating the Impact of Interventions: {{An}} Overview},
- volume = {30}}
-
-@article{hershfield2011increasing,
- author = {Hershfield, Hal E and Goldstein, Daniel G and Sharpe, William F and Fox, Jesse and Yeykelis, Leo and Carstensen, Laura L and Bailenson, Jeremy N},
- date = {2011},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- issue = {SPL},
- journaltitle = {Journal of Marketing Research},
- pages = {S23--S37},
- title = {Increasing Saving Behavior through Age-Progressed Renderings of the Future Self},
- volume = {48}}
-
-@inproceedings{ho1995random,
- author = {Ho, Tin Kam},
- booktitle = {Proceedings of 3rd International Conference on Document Analysis and Recognition},
- date = {1995},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {278--282},
- publisher = {{IEEE}},
- title = {Random Decision Forests},
- volume = {1}}
-
-@article{hochreiter1997long,
- author = {Hochreiter, Sepp and Schmidhuber, J{\"u}rgen},
- date = {1997},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Neural computation},
- number = {8},
- pages = {1735--1780},
- title = {Long Short-Term Memory},
- volume = {9}}
-
-@unpublished{hoff2021bayesoptimal,
- archiveprefix = {arXiv},
- author = {Hoff, Peter},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2105.14045},
- eprinttype = {arxiv},
- file = {/Users/FA31DU/Zotero/storage/IQK27WVA/Hoff - 2021 - Bayes-optimal prediction with frequentist coverage.pdf;/Users/FA31DU/Zotero/storage/K8EAZA25/2105.html},
- title = {Bayes-Optimal Prediction with Frequentist Coverage Control}}
-
-@misc{hoffman1994german,
- author = {Hoffman, Hans},
- date = {1994},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {German {{Credit Data}}},
- url = {https://archive.ics.uci.edu/ml/datasets/statlog+(german+credit+data)},
- bdsk-url-1 = {https://archive.ics.uci.edu/ml/datasets/statlog+(german+credit+data)}}
-
-@online{hoffmanGermanCreditData1994,
- author = {Hoffman, Hans},
- date = {1994},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {German {{Credit Data}}},
- url = {https://archive.ics.uci.edu/ml/datasets/statlog+(german+credit+data)},
- bdsk-url-1 = {https://archive.ics.uci.edu/ml/datasets/statlog+(german+credit+data)}}
-
-@unpublished{houlsby2011bayesian,
- archiveprefix = {arXiv},
- author = {Houlsby, Neil and Husz{\'a}r, Ferenc and Ghahramani, Zoubin and Lengyel, M{\'a}t{\'e}},
- date = {2011},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1112.5745},
- eprinttype = {arxiv},
- title = {Bayesian Active Learning for Classification and Preference Learning}}
-
-@article{hsee1996evaluability,
- author = {Hsee, Christopher K},
- date = {1996},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Organizational behavior and human decision processes},
- number = {3},
- pages = {247--257},
- title = {The Evaluability Hypothesis: {{An}} Explanation for Preference Reversals between Joint and Separate Evaluations of Alternatives},
- volume = {67}}
-
-@article{hsee2004music,
- author = {Hsee, Christopher K and Rottenstreich, Yuval},
- date = {2004},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Experimental Psychology: General},
- number = {1},
- pages = {23},
- title = {Music, Pandas, and Muggers: On the Affective Psychology of Value.},
- volume = {133}}
-
-@article{hsieh2016social,
- author = {Hsieh, Chih-Sheng and Lee, Lung Fei},
- date = {2016},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Applied Econometrics},
- number = {2},
- pages = {301--319},
- title = {A Social Interactions Model with Endogenous Friendship Formation and Selectivity},
- volume = {31}}
-
-@unpublished{immer2020improving,
- archiveprefix = {arXiv},
- author = {Immer, Alexander and Korzepa, Maciej and Bauer, Matthias},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2008.08400},
- eprinttype = {arxiv},
- title = {Improving Predictions of Bayesian Neural Networks via Local Linearization}}
-
-@unpublished{innes2018fashionable,
- archiveprefix = {arXiv},
- author = {Innes, Michael and Saba, Elliot and Fischer, Keno and Gandhi, Dhairya and Rudilosso, Marco Concetto and Joy, Neethu Mariya and Karmali, Tejan and Pal, Avik and Shah, Viral},
- date = {2018},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1811.01457},
- eprinttype = {arxiv},
- title = {Fashionable Modelling with Flux}}
-
-@article{innes2018flux,
- author = {Innes, Mike},
- date = {2018},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Open Source Software},
- number = {25},
- pages = {602},
- title = {Flux: {{Elegant}} Machine Learning with {{Julia}}},
- volume = {3}}
-
-@unpublished{ish-horowicz2019interpreting,
- archiveprefix = {arXiv},
- author = {Ish-Horowicz, Jonathan and Udwin, Dana and Flaxman, Seth and Filippi, Sarah and Crawford, Lorin},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1901.09839},
- eprinttype = {arxiv},
- title = {Interpreting Deep Neural Networks through Variable Importance}}
-
-@inproceedings{jabbari2017fairness,
- author = {Jabbari, Shahin and Joseph, Matthew and Kearns, Michael and Morgenstern, Jamie and Roth, Aaron},
- booktitle = {International {{Conference}} on {{Machine Learning}}},
- date = {2017},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {1617--1626},
- publisher = {{PMLR}},
- title = {Fairness in Reinforcement Learning}}
-
-@article{jackson2007meeting,
- author = {Jackson, Matthew O and Rogers, Brian W},
- date = {2007},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {American Economic Review},
- number = {3},
- pages = {890--915},
- title = {Meeting Strangers and Friends of Friends: {{How}} Random Are Social Networks?},
- volume = {97}}
-
-@unpublished{jeanneret2022diffusion,
- archiveprefix = {arXiv},
- author = {Jeanneret, Guillaume and Simon, Lo{\"\i}c and Jurie, Fr{\'e}d{\'e}ric},
- date = {2022},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2203.15636},
- eprinttype = {arxiv},
- title = {Diffusion {{Models}} for {{Counterfactual Explanations}}}}
-
-@article{johansson2005failure,
- author = {Johansson, Petter and Hall, Lars and Sikstr{\"o}m, Sverker and Olsson, Andreas},
- date = {2005},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Science (New York, N.Y.)},
- number = {5745},
- pages = {116--119},
- shortjournal = {Science},
- title = {Failure to Detect Mismatches between Intention and Outcome in a Simple Decision Task},
- volume = {310}}
-
-@article{johnsson2021estimation,
- author = {Johnsson, Ida and Moon, Hyungsik Roger},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Review of Economics and Statistics},
- number = {2},
- pages = {328--345},
- title = {Estimation of Peer Effects in Endogenous Social Networks: {{Control}} Function Approach},
- volume = {103}}
-
-@article{jolliffe2003modified,
- author = {Jolliffe, Ian T and Trendafilov, Nickolay T and Uddin, Mudassir},
- date = {2003},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of computational and Graphical Statistics},
- number = {3},
- pages = {531--547},
- title = {A Modified Principal Component Technique Based on the {{LASSO}}},
- volume = {12}}
-
-@article{joseph2021forecasting,
- author = {Joseph, Andreas and Kalamara, Eleni and Kapetanios, George and Potjagailo, Galina},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Forecasting Uk Inflation Bottom Up}}
-
-@unpublished{joshi2019realistic,
- archiveprefix = {arXiv},
- author = {Joshi, Shalmali and Koyejo, Oluwasanmi and Vijitbenjaronk, Warut and Kim, Been and Ghosh, Joydeep},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1907.09615},
- eprinttype = {arxiv},
- title = {Towards Realistic Individual Recourse and Actionable Explanations in Black-Box Decision Making Systems}}
-
-@unpublished{jospin2020handson,
- archiveprefix = {arXiv},
- author = {Jospin, Laurent Valentin and Buntine, Wray and Boussaid, Farid and Laga, Hamid and Bennamoun, Mohammed},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2007.06823},
- eprinttype = {arxiv},
- title = {Hands-on {{Bayesian Neural Networks}}--a {{Tutorial}} for {{Deep Learning Users}}}}
-
-@misc{kaggle2011give,
- author = {Kaggle},
- date = {2011},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Kaggle}},
- title = {Give Me Some Credit, {{Improve}} on the State of the Art in Credit Scoring by Predicting the Probability That Somebody Will Experience Financial Distress in the next Two Years.},
- url = {https://www.kaggle.com/c/GiveMeSomeCredit},
- bdsk-url-1 = {https://www.kaggle.com/c/GiveMeSomeCredit}}
-
-@online{kagglecompetitionGiveMeCredit,
- author = {Kaggle Competition},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Give Me Some Credit, {{Improve}} on the State of the Art in Credit Scoring by Predicting the Probability That Somebody Will Experience Financial Distress in the next Two Years.},
- url = {https://www.kaggle.com/c/GiveMeSomeCredit},
- bdsk-url-1 = {https://www.kaggle.com/c/GiveMeSomeCredit}}
-
-@article{kahneman1979prospect,
- author = {Kahneman, Daniel and Tversky, Amos},
- date = {1979},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Econometrica: Journal of the Econometric Society},
- pages = {263--291},
- title = {Prospect {{Theory}}: {{An Analysis}} of {{Decision}} under {{Risk}}}}
-
-@article{kahneman1990experimental,
- author = {Kahneman, Daniel and Knetsch, Jack L and Thaler, Richard H},
- date = {1990},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of political Economy},
- number = {6},
- pages = {1325--1348},
- title = {Experimental Tests of the Endowment Effect and the {{Coase}} Theorem},
- volume = {98}}
-
-@article{kahneman1992reference,
- author = {Kahneman, Daniel},
- date = {1992},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Organizational behavior and human decision processes},
- number = {2},
- pages = {296--312},
- title = {Reference Points, Anchors, Norms, and Mixed Feelings},
- volume = {51}}
-
-@unpublished{karimi2020algorithmic,
- archiveprefix = {arXiv},
- author = {Karimi, Amir-Hossein and Von K{\"u}gelgen, Julius and Sch{\"o}lkopf, Bernhard and Valera, Isabel},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2006.06831},
- eprinttype = {arxiv},
- title = {Algorithmic Recourse under Imperfect Causal Knowledge: A Probabilistic Approach}}
-
-@unpublished{karimi2020survey,
- archiveprefix = {arXiv},
- author = {Karimi, Amir-Hossein and Barthe, Gilles and Sch{\"o}lkopf, Bernhard and Valera, Isabel},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2010.04050},
- eprinttype = {arxiv},
- title = {A Survey of Algorithmic Recourse: Definitions, Formulations, Solutions, and Prospects}}
-
-@inproceedings{karimi2021algorithmic,
- author = {Karimi, Amir-Hossein and Sch{\"o}lkopf, Bernhard and Valera, Isabel},
- booktitle = {Proceedings of the 2021 {{ACM Conference}} on {{Fairness}}, {{Accountability}}, and {{Transparency}}},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {353--362},
- title = {Algorithmic Recourse: From Counterfactual Explanations to Interventions}}
-
-@inproceedings{kaur2020interpreting,
- author = {Kaur, Harmanpreet and Nori, Harsha and Jenkins, Samuel and Caruana, Rich and Wallach, Hanna and Wortman Vaughan, Jennifer},
- booktitle = {Proceedings of the 2020 {{CHI}} Conference on Human Factors in Computing Systems},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {1--14},
- title = {Interpreting Interpretability: Understanding Data Scientists' Use of Interpretability Tools for Machine Learning}}
-
-@article{kehoe2021defence,
- author = {Kehoe, Aidan and Wittek, Peter and Xue, Yanbo and Pozas-Kerstjens, Alejandro},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Machine Learning: Science and Technology},
- title = {Defence against Adversarial Attacks Using Classical and Quantum-Enhanced {{Boltzmann}} Machines}}
-
-@unpublished{kendall2017what,
- archiveprefix = {arXiv},
- author = {Kendall, Alex and Gal, Yarin},
- date = {2017},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1703.04977},
- eprinttype = {arxiv},
- title = {What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?}}
-
-@article{kihoro2004seasonal,
- author = {Kihoro, J and Otieno, RO and Wafula, C},
- date = {2004},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Seasonal Time Series Forecasting: {{A}} Comparative Study of {{ARIMA}} and {{ANN}} Models}}
-
-@book{kilian2017structural,
- author = {Kilian, Lutz and L{\"u}tkepohl, Helmut},
- date = {2017},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Cambridge University Press}},
- title = {Structural Vector Autoregressive Analysis}}
-
-@unpublished{kingma2014adam,
- archiveprefix = {arXiv},
- author = {Kingma, Diederik P and Ba, Jimmy},
- date = {2014},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1412.6980},
- eprinttype = {arxiv},
- title = {Adam: {{A}} Method for Stochastic Optimization}}
-
-@article{kirsch2019batchbald,
- author = {Kirsch, Andreas and Van Amersfoort, Joost and Gal, Yarin},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Advances in neural information processing systems},
- pages = {7026--7037},
- title = {Batchbald: {{Efficient}} and Diverse Batch Acquisition for Deep Bayesian Active Learning},
- volume = {32}}
-
-@unpublished{kuiper2021exploring,
- archiveprefix = {arXiv},
- author = {Kuiper, Ouren and van den Berg, Martin and van den Burgt, Joost and Leijnen, Stefan},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2111.02244},
- eprinttype = {arxiv},
- title = {Exploring {{Explainable AI}} in the {{Financial Sector}}: {{Perspectives}} of {{Banks}} and {{Supervisory Authorities}}}}
-
-@article{kydland1982time,
- author = {Kydland, Finn E and Prescott, Edward C},
- date = {1982},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Econometrica: Journal of the Econometric Society},
- pages = {1345--1370},
- title = {Time to Build and Aggregate Fluctuations}}
-
-@unpublished{lachapelle2019gradientbased,
- archiveprefix = {arXiv},
- author = {Lachapelle, S{\'e}bastien and Brouillard, Philippe and Deleu, Tristan and Lacoste-Julien, Simon},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1906.02226},
- eprinttype = {arxiv},
- title = {Gradient-Based Neural Dag Learning}}
-
-@inproceedings{lakkaraju2020how,
- author = {Lakkaraju, Himabindu and Bastani, Osbert},
- booktitle = {Proceedings of the {{AAAI}}/{{ACM Conference}} on {{AI}}, {{Ethics}}, and {{Society}}},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {79--85},
- title = {" {{How}} Do {{I}} Fool You?" {{Manipulating User Trust}} via {{Misleading Black Box Explanations}}}}
-
-@inproceedings{lakkaraju2020how,
- author = {Lakkaraju, Himabindu and Bastani, Osbert},
- booktitle = {Proceedings of the {{AAAI}}/{{ACM Conference}} on {{AI}}, {{Ethics}}, and {{Society}}},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {79--85},
- title = {" {{How Do I Fool You}}?" {{Manipulating User Trust}} via {{Misleading Black Box Explanations}}}}
-
-@unpublished{lakshminarayanan2016simple,
- archiveprefix = {arXiv},
- author = {Lakshminarayanan, Balaji and Pritzel, Alexander and Blundell, Charles},
- date = {2016},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1612.01474},
- eprinttype = {arxiv},
- title = {Simple and Scalable Predictive Uncertainty Estimation Using Deep Ensembles}}
-
-@unpublished{laugel2017inverse,
- archiveprefix = {arXiv},
- author = {Laugel, Thibault and Lesot, Marie-Jeanne and Marsala, Christophe and Renard, Xavier and Detyniecki, Marcin},
- date = {2017},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1712.08443},
- eprinttype = {arxiv},
- shortjournal = {arXiv preprint arXiv:1712.08443},
- title = {Inverse Classification for Comparison-Based Interpretability in Machine Learning}}
-
-@thesis{lawrence2001variational,
- author = {Lawrence, Neil David},
- date = {2001},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- institution = {{University of Cambridge}},
- title = {Variational Inference in Probabilistic Models},
- type = {phdthesis}}
-
-@article{lecun1998mnist,
- author = {LeCun, Yann},
- date = {1998},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- shortjournal = {http://yann. lecun. com/exdb/mnist/},
- title = {The {{MNIST}} Database of Handwritten Digits}}
-
-@article{lee2003best,
- author = {Lee, Lung-fei},
- date = {2003},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Econometric Reviews},
- number = {4},
- pages = {307--335},
- title = {Best Spatial Two-Stage Least Squares Estimators for a Spatial Autoregressive Model with Autoregressive Disturbances},
- volume = {22}}
-
-@article{lerner2013financial,
- author = {Lerner, Jennifer S and Li, Ye and Weber, Elke U},
- date = {2013},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Psychological science},
- number = {1},
- pages = {72--79},
- title = {The Financial Costs of Sadness},
- volume = {24}}
-
-@article{list2004neoclassical,
- author = {List, John A},
- date = {2004},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Econometrica : journal of the Econometric Society},
- number = {2},
- pages = {615--625},
- shortjournal = {Econometrica},
- title = {Neoclassical Theory versus Prospect Theory: {{Evidence}} from the Marketplace},
- volume = {72}}
-
-@article{lucas1976econometric,
- author = {Lucas, JR},
- date = {1976},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Econometric Policy Evaluation: A Critique `, in {{K}}. {{Brunner}} and {{A Meltzer}}, {{The Phillips}} Curve and Labor Markets, {{North Holland}}}}
-
-@inproceedings{lundberg2017unified,
- author = {Lundberg, Scott M and Lee, Su-In},
- booktitle = {Proceedings of the 31st International Conference on Neural Information Processing Systems},
- date = {2017},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {4768--4777},
- title = {A Unified Approach to Interpreting Model Predictions}}
-
-@book{lutkepohl2005new,
- author = {L{\"u}tkepohl, Helmut},
- date = {2005},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Springer Science \& Business Media}},
- title = {New Introduction to Multiple Time Series Analysis}}
-
-@article{madrian2001power,
- author = {Madrian, Brigitte C and Shea, Dennis F},
- date = {2001},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The Quarterly journal of economics},
- number = {4},
- pages = {1149--1187},
- title = {The Power of Suggestion: {{Inertia}} in 401 (k) Participation and Savings Behavior},
- volume = {116}}
-
-@book{manning2008introduction,
- author = {Manning, Christopher D and Sch{\"u}tze, Hinrich and Raghavan, Prabhakar},
- date = {2008},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Cambridge university press}},
- title = {Introduction to Information Retrieval}}
-
-@misc{manokhin2022awesome,
- author = {Manokhin, Valery},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Awesome Conformal Prediction}}
-
-@article{manski1993identification,
- author = {Manski, Charles F},
- date = {1993},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The review of economic studies},
- number = {3},
- pages = {531--542},
- title = {Identification of Endogenous Social Effects: {{The}} Reflection Problem},
- volume = {60}}
-
-@article{markle2018goals,
- author = {Markle, Alex and Wu, George and White, Rebecca and Sackett, Aaron},
- date = {2018},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Risk and Uncertainty},
- number = {1},
- pages = {19--50},
- title = {Goals as Reference Points in Marathon Running: {{A}} Novel Test of Reference Dependence},
- volume = {56}}
-
-@article{masini2021machine,
- author = {Masini, Ricardo P and Medeiros, Marcelo C and Mendes, Eduardo F},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Economic Surveys},
- title = {Machine Learning Advances for Time Series Forecasting}}
-
-@article{mccracken2016fredmd,
- author = {McCracken, Michael W and Ng, Serena},
- date = {2016},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Business \& Economic Statistics},
- number = {4},
- pages = {574--589},
- title = {{{FRED-MD}}: {{A}} Monthly Database for Macroeconomic Research},
- volume = {34}}
-
-@article{mcculloch1990logical,
- author = {McCulloch, Warren S and Pitts, Walter},
- date = {1990},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Bulletin of mathematical biology},
- number = {1},
- pages = {99--115},
- title = {A Logical Calculus of the Ideas Immanent in Nervous Activity},
- volume = {52}}
-
-@article{migut2015visualizing,
- author = {Migut, MA and Worring, Marcel and Veenman, Cor J},
- date = {2015},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Data Mining and Knowledge Discovery},
- number = {1},
- pages = {273--295},
- title = {Visualizing Multi-Dimensional Decision Boundaries in {{2D}}},
- volume = {29}}
-
-@article{miller2019explanation,
- author = {Miller, Tim},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Artificial intelligence},
- pages = {1--38},
- title = {Explanation in Artificial Intelligence: {{Insights}} from the Social Sciences},
- volume = {267}}
-
-@inproceedings{miller2020strategic,
- abstract = {Consequential decision-making incentivizes individuals to strategically adapt their behavior to the specifics of the decision rule. While a long line of work has viewed strategic adaptation as gaming and attempted to mitigate its effects, recent work has instead sought to design classifiers that incentivize individuals to improve a desired quality. Key to both accounts is a cost function that dictates which adaptations are rational to undertake. In this work, we develop a causal framework for strategic adaptation. Our causal perspective clearly distinguishes between gaming and improvement and reveals an important obstacle to incentive design. We prove any procedure for designing classifiers that incentivize improvement must inevitably solve a non-trivial causal inference problem. We show a similar result holds for designing cost functions that satisfy the requirements of previous work. With the benefit of hindsight, our results show much of the prior work on strategic classification is causal modeling in disguise.},
- author = {Miller, John and Milli, Smitha and Hardt, Moritz},
- booktitle = {Proceedings of the 37th {{International Conference}} on {{Machine Learning}}},
- date = {2020-11-21},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eventtitle = {International {{Conference}} on {{Machine Learning}}},
- file = {/Users/FA31DU/Zotero/storage/46I2QMPI/Miller et al. - 2020 - Strategic Classification is Causal Modeling in Dis.pdf;/Users/FA31DU/Zotero/storage/NWREET6B/Miller et al. - 2020 - Strategic Classification is Causal Modeling in Dis.pdf},
- issn = {2640-3498},
- langid = {english},
- pages = {6917--6926},
- publisher = {{PMLR}},
- title = {Strategic {{Classification}} Is {{Causal Modeling}} in {{Disguise}}},
- url = {https://proceedings.mlr.press/v119/miller20b.html},
- urldate = {2022-11-03},
- bdsk-url-1 = {https://proceedings.mlr.press/v119/miller20b.html}}
-
-@article{mischel1988nature,
- author = {Mischel, Walter and Shoda, Yuichi and Peake, Philip K},
- date = {1988},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of personality and social psychology},
- number = {4},
- pages = {687},
- title = {The Nature of Adolescent Competencies Predicted by Preschool Delay of Gratification.},
- volume = {54}}
-
-@inproceedings{mittelstadt2019explaining,
- author = {Mittelstadt, Brent and Russell, Chris and Wachter, Sandra},
- booktitle = {Proceedings of the Conference on Fairness, Accountability, and Transparency},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {279--288},
- title = {Explaining Explanations in {{AI}}}}
-
-@book{molnar2020interpretable,
- author = {Molnar, Christoph},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Lulu. com}},
- title = {Interpretable Machine Learning}}
-
-@book{morgan2015counterfactuals,
- author = {Morgan, Stephen L and Winship, Christopher},
- date = {2015},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Cambridge University Press}},
- title = {Counterfactuals and Causal Inference}}
-
-@article{mosteller1951experimental,
- author = {Mosteller, Frederick and Nogee, Philip},
- date = {1951},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Political Economy},
- number = {5},
- pages = {371--404},
- title = {An Experimental Measurement of Utility},
- volume = {59}}
-
-@inproceedings{mothilal2020explaining,
- author = {Mothilal, Ramaravind K and Sharma, Amit and Tan, Chenhao},
- booktitle = {Proceedings of the 2020 {{Conference}} on {{Fairness}}, {{Accountability}}, and {{Transparency}}},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {607--617},
- title = {Explaining Machine Learning Classifiers through Diverse Counterfactual Explanations}}
-
-@book{murphy2012machine,
- author = {Murphy, Kevin P},
- date = {2012},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{MIT press}},
- title = {Machine Learning: A Probabilistic Perspective}}
-
-@book{murphy2012machine,
- author = {Murphy, Kevin P},
- date = {2012},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{MIT press}},
- title = {Machine Learning: {{A}} Probabilistic Perspective}}
-
-@book{murphy2022probabilistic,
- author = {Murphy, Kevin P},
- date = {2022},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{MIT Press}},
- title = {Probabilistic {{Machine Learning}}: {{An}} Introduction}}
-
-@article{nagel1995unraveling,
- author = {Nagel, Rosemarie},
- date = {1995},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The American Economic Review},
- number = {5},
- pages = {1313--1326},
- title = {Unraveling in Guessing Games: {{An}} Experimental Study},
- volume = {85}}
-
-@unpublished{navarro-martinez2021bridging,
- author = {Navarro-Martinez, Daniel and Wang, Xinghua},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Bridging the Gap between the Lab and the Field: {{Dictator}} Games and Donations}}
-
-@inproceedings{nelson2015evaluating,
- author = {Nelson, Kevin and Corbin, George and Anania, Mark and Kovacs, Matthew and Tobias, Jeremy and Blowers, Misty},
- booktitle = {2015 {{IEEE Symposium}} on {{Computational Intelligence}} for {{Security}} and {{Defense Applications}} ({{CISDA}})},
- date = {2015},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {1--8},
- publisher = {{IEEE}},
- title = {Evaluating Model Drift in Machine Learning Algorithms}}
-
-@book{nocedal2006numerical,
- author = {Nocedal, Jorge and Wright, Stephen},
- date = {2006},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Springer Science \& Business Media}},
- title = {Numerical Optimization}}
-
-@misc{oecd2021artificial,
- author = {{OECD}},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- title = {Artificial {{Intelligence}}, {{Machine Learning}} and {{Big Data}} in {{Finance}}: {{Opportunities}}, {{Challenges}} and {{Implications}} for {{Policy Makers}}},
- url = {https://www.oecd.org/finance/financial-markets/Artificial-intelligence-machine-learning-big-data-in-finance.pdf},
- bdsk-url-1 = {https://www.oecd.org/finance/financial-markets/Artificial-intelligence-machine-learning-big-data-in-finance.pdf}}
-
-@online{oecdArtificialIntelligenceMachine2021,
- author = {{OECD}},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{OECD}},
- title = {Artificial {{Intelligence}}, {{Machine Learning}} and {{Big Data}} in {{Finance}}: {{Opportunities}}, {{Challenges}} and {{Implications}} for {{Policy Makers}}},
- url = {https://www.oecd.org/finance/financial-markets/Artificial-intelligence-machine-learning-big-data-in-finance.pdf},
- bdsk-url-1 = {https://www.oecd.org/finance/financial-markets/Artificial-intelligence-machine-learning-big-data-in-finance.pdf}}
-
-@book{oneil2016weapons,
- author = {O'Neil, Cathy},
- date = {2016},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Crown}},
- title = {Weapons of Math Destruction: {{How}} Big Data Increases Inequality and Threatens Democracy}}
-
-@article{pace1997sparse,
- author = {Pace, R Kelley and Barry, Ronald},
- date = {1997},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Statistics \& Probability Letters},
- number = {3},
- pages = {291--297},
- title = {Sparse Spatial Autoregressions},
- volume = {33}}
-
-@unpublished{parr2018matrix,
- archiveprefix = {arXiv},
- author = {Parr, Terence and Howard, Jeremy},
- date = {2018},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1802.01528},
- eprinttype = {arxiv},
- title = {The Matrix Calculus You Need for Deep Learning}}
-
-@unpublished{pawelczyk2021carla,
- archiveprefix = {arXiv},
- author = {Pawelczyk, Martin and Bielawski, Sascha and van den Heuvel, Johannes and Richter, Tobias and Kasneci, Gjergji},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2108.00783},
- eprinttype = {arxiv},
- title = {Carla: A Python Library to Benchmark Algorithmic Recourse and Counterfactual Explanation Algorithms}}
-
-@book{pearl2018book,
- author = {Pearl, Judea and Mackenzie, Dana},
- date = {2018},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Basic books}},
- title = {The Book of Why: The New Science of Cause and Effect}}
-
-@article{pearl2019seven,
- author = {Pearl, Judea},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Communications of the ACM},
- number = {3},
- pages = {54--60},
- title = {The Seven Tools of Causal Inference, with Reflections on Machine Learning},
- volume = {62}}
-
-@article{pedregosa2011scikitlearn,
- author = {Pedregosa, Fabian and Varoquaux, Ga{\"e}l and Gramfort, Alexandre and Michel, Vincent and Thirion, Bertrand and Grisel, Olivier and Blondel, Mathieu and Prettenhofer, Peter and Weiss, Ron and Dubourg, Vincent and others},
- date = {2011},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {the Journal of machine Learning research},
- pages = {2825--2830},
- title = {Scikit-Learn: {{Machine}} Learning in {{Python}}},
- volume = {12}}
-
-@book{perry2010economic,
- author = {Perry, George L and Tobin, James},
- date = {2010},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Brookings Institution Press}},
- title = {Economic {{Events}}, {{Ideas}}, and {{Policies}}: The 1960s and After}}
-
-@article{pfaff2008var,
- author = {Pfaff, Bernhard and others},
- date = {2008},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Statistical Software},
- number = {4},
- pages = {1--32},
- title = {{{VAR}}, {{SVAR}} and {{SVEC}} Models: {{Implementation}} within {{R}} Package Vars},
- volume = {27}}
-
-@book{pindyck2014microeconomics,
- author = {Pindyck, Robert S and Rubinfeld, Daniel L},
- date = {2014},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Pearson Education}},
- title = {Microeconomics}}
-
-@article{pope2011numbers,
- author = {Pope, Devin and Simonsohn, Uri},
- date = {2011},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Psychological science},
- number = {1},
- pages = {71--79},
- title = {Round Numbers as Goals: {{Evidence}} from Baseball, {{SAT}} Takers, and the Lab},
- volume = {22}}
-
-@inproceedings{poyiadzi2020face,
- author = {Poyiadzi, Rafael and Sokol, Kacper and Santos-Rodriguez, Raul and De Bie, Tijl and Flach, Peter},
- booktitle = {Proceedings of the {{AAAI}}/{{ACM Conference}} on {{AI}}, {{Ethics}}, and {{Society}}},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {344--350},
- title = {{{FACE}}: {{Feasible}} and Actionable Counterfactual Explanations}}
-
-@article{qu2015estimating,
- author = {Qu, Xi and Lee, Lung-fei},
- date = {2015},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of Econometrics},
- number = {2},
- pages = {209--232},
- title = {Estimating a Spatial Autoregressive Model with an Endogenous Spatial Weight Matrix},
- volume = {184}}
-
-@article{rabanser2019failing,
- author = {Rabanser, Stephan and G{\"u}nnemann, Stephan and Lipton, Zachary},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Advances in Neural Information Processing Systems},
- title = {Failing Loudly: {{An}} Empirical Study of Methods for Detecting Dataset Shift},
- volume = {32}}
-
-@unpublished{raghunathan2019adversarial,
- archiveprefix = {arXiv},
- author = {Raghunathan, Aditi and Xie, Sang Michael and Yang, Fanny and Duchi, John C and Liang, Percy},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1906.06032},
- eprinttype = {arxiv},
- title = {Adversarial Training Can Hurt Generalization}}
-
-@unpublished{raj2017taming,
- archiveprefix = {arXiv},
- author = {Raj, Vishnu and Kalyani, Sheetal},
- date = {2017},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1707.09727},
- eprinttype = {arxiv},
- title = {Taming Non-Stationary Bandits: {{A Bayesian}} Approach}}
-
-@inproceedings{rasmussen2003gaussian,
- author = {Rasmussen, Carl Edward},
- booktitle = {Summer School on Machine Learning},
- date = {2003},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {63--71},
- publisher = {{Springer}},
- title = {Gaussian Processes in Machine Learning}}
-
-@inproceedings{ribeiro2016why,
- author = {Ribeiro, Marco Tulio and Singh, Sameer and Guestrin, Carlos},
- booktitle = {Proceedings of the 22nd {{ACM SIGKDD}} International Conference on Knowledge Discovery and Data Mining},
- date = {2016},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {1135--1144},
- title = {"{{Why}} Should i Trust You?" {{Explaining}} the Predictions of Any Classifier}}
-
-@article{romer1989does,
- author = {Romer, Christina D and Romer, David H},
- date = {1989},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {NBER macroeconomics annual},
- pages = {121--170},
- title = {Does Monetary Policy Matter? {{A}} New Test in the Spirit of {{Friedman}} and {{Schwartz}}},
- volume = {4}}
-
-@article{rudin2019stop,
- author = {Rudin, Cynthia},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Nature Machine Intelligence},
- number = {5},
- pages = {206--215},
- title = {Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead},
- volume = {1}}
-
-@article{sacerdote2001peer,
- author = {Sacerdote, Bruce},
- date = {2001},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The Quarterly journal of economics},
- number = {2},
- pages = {681--704},
- title = {Peer Effects with Random Assignment: {{Results}} for {{Dartmouth}} Roommates},
- volume = {116}}
-
-@article{sadinle2019least,
- author = {Sadinle, Mauricio and Lei, Jing and Wasserman, Larry},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- file = {/Users/FA31DU/Zotero/storage/YXQ8N76A/Sadinle et al. - 2019 - Least ambiguous set-valued classifiers with bounde.pdf;/Users/FA31DU/Zotero/storage/ZHB56F3V/01621459.2017.html},
- journaltitle = {Journal of the American Statistical Association},
- number = {525},
- pages = {223--234},
- publisher = {{Taylor \& Francis}},
- title = {Least Ambiguous Set-Valued Classifiers with Bounded Error Levels},
- volume = {114}}
-
-@inproceedings{satopaa2011finding,
- author = {Satopaa, Ville and Albrecht, Jeannie and Irwin, David and Raghavan, Barath},
- booktitle = {2011 31st International Conference on Distributed Computing Systems Workshops},
- date = {2011},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {166--171},
- publisher = {{IEEE}},
- title = {Finding a" Kneedle" in a Haystack: {{Detecting}} Knee Points in System Behavior}}
-
-@inproceedings{schut2021generating,
- author = {Schut, Lisa and Key, Oscar and Mc Grath, Rory and Costabello, Luca and Sacaleanu, Bogdan and Gal, Yarin and others},
- booktitle = {International {{Conference}} on {{Artificial Intelligence}} and {{Statistics}}},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {1756--1764},
- publisher = {{PMLR}},
- title = {Generating {{Interpretable Counterfactual Explanations By Implicit Minimisation}} of {{Epistemic}} and {{Aleatoric Uncertainties}}}}
-
-@book{schutze2008introduction,
- author = {Sch{\"u}tze, Hinrich and Manning, Christopher D and Raghavan, Prabhakar},
- date = {2008},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Cambridge University Press Cambridge}},
- title = {Introduction to Information Retrieval},
- volume = {39}}
-
-@article{shafir1993reasonbased,
- author = {Shafir, Eldar and Simonson, Itamar and Tversky, Amos},
- date = {1993},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Cognition},
- number = {1-2},
- pages = {11--36},
- title = {Reason-Based Choice},
- volume = {49}}
-
-@article{simonson1989choice,
- author = {Simonson, Itamar},
- date = {1989},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of consumer research},
- number = {2},
- pages = {158--174},
- title = {Choice Based on Reasons: {{The}} Case of Attraction and Compromise Effects},
- volume = {16}}
-
-@article{sims1986are,
- author = {Sims, Christopher A and others},
- date = {1986},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- issue = {Win},
- journaltitle = {Quarterly Review},
- pages = {2--16},
- title = {Are Forecasting Models Usable for Policy Analysis?},
- volume = {10}}
-
-@inproceedings{slack2020fooling,
- author = {Slack, Dylan and Hilgard, Sophie and Jia, Emily and Singh, Sameer and Lakkaraju, Himabindu},
- booktitle = {Proceedings of the {{AAAI}}/{{ACM Conference}} on {{AI}}, {{Ethics}}, and {{Society}}},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {180--186},
- title = {Fooling Lime and Shap: {{Adversarial}} Attacks on Post Hoc Explanation Methods}}
-
-@article{slack2021counterfactual,
- author = {Slack, Dylan and Hilgard, Anna and Lakkaraju, Himabindu and Singh, Sameer},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Advances in Neural Information Processing Systems},
- title = {Counterfactual Explanations Can Be Manipulated},
- volume = {34}}
-
-@article{slovic1974who,
- author = {Slovic, Paul and Tversky, Amos},
- date = {1974},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Behavioral science},
- number = {6},
- pages = {368--373},
- title = {Who Accepts {{Savage}}'s Axiom?},
- volume = {19}}
-
-@unpublished{spooner2021counterfactual,
- archiveprefix = {arXiv},
- author = {Spooner, Thomas and Dervovic, Danial and Long, Jason and Shepard, Jon and Chen, Jiahao and Magazzeni, Daniele},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2106.15212},
- eprinttype = {arxiv},
- shortjournal = {arXiv preprint arXiv:2106.15212},
- title = {Counterfactual {{Explanations}} for {{Arbitrary Regression Models}}}}
-
-@article{srivastava2014dropout,
- author = {Srivastava, Nitish and Hinton, Geoffrey and Krizhevsky, Alex and Sutskever, Ilya and Salakhutdinov, Ruslan},
- date = {2014},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The journal of machine learning research},
- number = {1},
- pages = {1929--1958},
- title = {Dropout: A Simple Way to Prevent Neural Networks from Overfitting},
- volume = {15}}
-
-@unpublished{stanton2022bayesian,
- archiveprefix = {arXiv},
- author = {Stanton, Samuel and Maddox, Wesley and Wilson, Andrew Gordon},
- date = {2022},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2210.12496},
- eprinttype = {arxiv},
- file = {/Users/FA31DU/Zotero/storage/XFGZAB9J/Stanton et al. - 2022 - Bayesian Optimization with Conformal Coverage Guar.pdf;/Users/FA31DU/Zotero/storage/RPWYDPVW/2210.html},
- title = {Bayesian {{Optimization}} with {{Conformal Coverage Guarantees}}}}
-
-@article{sturm2014simple,
- author = {Sturm, Bob L},
- date = {2014},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {IEEE Transactions on Multimedia},
- number = {6},
- pages = {1636--1644},
- title = {A Simple Method to Determine If a Music Information Retrieval System Is a ``Horse''},
- volume = {16}}
-
-@article{sunstein2003libertarian,
- author = {Sunstein, Cass R and Thaler, Richard H},
- date = {2003},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {The University of Chicago Law Review},
- pages = {1159--1202},
- title = {Libertarian Paternalism Is Not an Oxymoron}}
-
-@book{sutton2018reinforcement,
- author = {Sutton, Richard S and Barto, Andrew G},
- date = {2018},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{MIT press}},
- title = {Reinforcement Learning: {{An}} Introduction}}
-
-@unpublished{szegedy2013intriguing,
- archiveprefix = {arXiv},
- author = {Szegedy, Christian and Zaremba, Wojciech and Sutskever, Ilya and Bruna, Joan and Erhan, Dumitru and Goodfellow, Ian and Fergus, Rob},
- date = {2013},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1312.6199},
- eprinttype = {arxiv},
- title = {Intriguing Properties of Neural Networks}}
-
-@article{thaler1981empirical,
- author = {Thaler, Richard},
- date = {1981},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Economics letters},
- number = {3},
- pages = {201--207},
- title = {Some Empirical Evidence on Dynamic Inconsistency},
- volume = {8}}
-
-@article{thaler2004more,
- author = {Thaler, Richard H and Benartzi, Shlomo},
- date = {2004},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of political Economy},
- number = {S1},
- pages = {S164--S187},
- title = {Save More Tomorrow{\texttrademark}: {{Using}} Behavioral Economics to Increase Employee Saving},
- volume = {112}}
-
-@article{tversky1981framing,
- author = {Tversky, Amos and Kahneman, Daniel},
- date = {1981},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Science (New York, N.Y.)},
- number = {4481},
- pages = {453--458},
- shortjournal = {science},
- title = {The Framing of Decisions and the Psychology of Choice},
- volume = {211}}
-
-@article{ungemach2011how,
- author = {Ungemach, Christoph and Stewart, Neil and Reimers, Stian},
- date = {2011},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Psychological Science},
- number = {2},
- pages = {253--260},
- title = {How Incidental Values from the Environment Affect Decisions about Money, Risk, and Delay},
- volume = {22}}
-
-@unpublished{upadhyay2021robust,
- archiveprefix = {arXiv},
- author = {Upadhyay, Sohini and Joshi, Shalmali and Lakkaraju, Himabindu},
- date = {2021},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2102.13620},
- eprinttype = {arxiv},
- title = {Towards {{Robust}} and {{Reliable Algorithmic Recourse}}}}
-
-@inproceedings{ustun2019actionable,
- author = {Ustun, Berk and Spangher, Alexander and Liu, Yang},
- booktitle = {Proceedings of the {{Conference}} on {{Fairness}}, {{Accountability}}, and {{Transparency}}},
- date = {2019},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- pages = {10--19},
- title = {Actionable Recourse in Linear Classification}}
-
-@article{vanboven2000egocentric,
- author = {Van Boven, Leaf and Dunning, David and Loewenstein, George},
- date = {2000},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of personality and social psychology},
- number = {1},
- pages = {66},
- title = {Egocentric Empathy Gaps between Owners and Buyers: Misperceptions of the Endowment Effect.},
- volume = {79}}
-
-@book{varshney2022trustworthy,
- author = {Varshney, Kush R.},
- date = {2022},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- location = {{Chappaqua, NY, USA}},
- publisher = {{Independently Published}},
- title = {Trustworthy {{Machine Learning}}}}
-
-@unpublished{verma2020counterfactual,
- archiveprefix = {arXiv},
- author = {Verma, Sahil and Dickerson, John and Hines, Keegan},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2010.10596},
- eprinttype = {arxiv},
- title = {Counterfactual Explanations for Machine Learning: {{A}} Review}}
-
-@article{verstyuk2020modeling,
- author = {Verstyuk, Sergiy},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Available at SSRN 3589337},
- title = {Modeling Multivariate Time Series in Economics: {{From}} Auto-Regressions to Recurrent Neural Networks}}
-
-@article{wachter2017counterfactual,
- author = {Wachter, Sandra and Mittelstadt, Brent and Russell, Chris},
- date = {2017},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Harv. JL \& Tech.},
- pages = {841},
- title = {Counterfactual Explanations without Opening the Black Box: {{Automated}} Decisions and the {{GDPR}}},
- volume = {31}}
-
-@article{wang2018optimal,
- author = {Wang, HaiYing and Zhu, Rong and Ma, Ping},
- date = {2018},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Journal of the American Statistical Association},
- number = {522},
- pages = {829--844},
- title = {Optimal Subsampling for Large Sample Logistic Regression},
- volume = {113}}
-
-@book{wasserman2006all,
- author = {Wasserman, Larry},
- date = {2006},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Springer Science \& Business Media}},
- title = {All of Nonparametric Statistics}}
-
-@book{wasserman2013all,
- author = {Wasserman, Larry},
- date = {2013},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- publisher = {{Springer Science \& Business Media}},
- title = {All of Statistics: A Concise Course in Statistical Inference}}
-
-@article{widmer1996learning,
- author = {Widmer, Gerhard and Kubat, Miroslav},
- date = {1996},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Machine learning},
- number = {1},
- pages = {69--101},
- title = {Learning in the Presence of Concept Drift and Hidden Contexts},
- volume = {23}}
-
-@unpublished{wilson2020case,
- archiveprefix = {arXiv},
- author = {Wilson, Andrew Gordon},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {2001.10995},
- eprinttype = {arxiv},
- title = {The Case for {{Bayesian}} Deep Learning}}
-
-@article{witten2009penalized,
- author = {Witten, Daniela M and Tibshirani, Robert and Hastie, Trevor},
- date = {2009},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Biostatistics (Oxford, England)},
- number = {3},
- pages = {515--534},
- shortjournal = {Biostatistics},
- title = {A Penalized Matrix Decomposition, with Applications to Sparse Principal Components and Canonical Correlation Analysis},
- volume = {10}}
-
-@article{xu2020epidemiological,
- author = {Xu, Bo and Gutierrez, Bernardo and Mekaru, Sumiko and Sewalk, Kara and Goodwin, Lauren and Loskill, Alyssa and Cohn, Emily and Hswen, Yulin and Hill, Sarah C. and Cobo, Maria M and Zarebski, Alexander and Li, Sabrina and Wu, Chieh-Hsi and Hulland, Erin and Morgan, Julia and Wang, Lin and O'Brien, Katelynn and Scarpino, Samuel V. and Brownstein, John S. and Pybus, Oliver G. and Pigott, David M. and Kraemer, Moritz U. G.},
- date = {2020},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- doi = {doi.org/10.1038/s41597-020-0448-0},
- journaltitle = {Scientific Data},
- number = {106},
- title = {Epidemiological Data from the {{COVID-19}} Outbreak, Real-Time Case Information},
- volume = {7},
- bdsk-url-1 = {https://doi.org/10.1038/s41597-020-0448-0}}
-
-@article{yeh2009comparisons,
- author = {Yeh, I-Cheng and Lien, Che-hui},
- date = {2009},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Expert systems with applications},
- number = {2},
- pages = {2473--2480},
- title = {The Comparisons of Data Mining Techniques for the Predictive Accuracy of Probability of Default of Credit Card Clients},
- volume = {36}}
-
-@article{zhang1998forecasting,
- author = {Zhang, Guoqiang and Patuwo, B Eddy and Hu, Michael Y},
- date = {1998},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {International journal of forecasting},
- number = {1},
- pages = {35--62},
- title = {Forecasting with Artificial Neural Networks:: {{The}} State of the Art},
- volume = {14}}
-
-@article{zhang2003time,
- author = {Zhang, G Peter},
- date = {2003},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {Neurocomputing},
- pages = {159--175},
- title = {Time Series Forecasting Using a Hybrid {{ARIMA}} and Neural Network Model},
- volume = {50}}
-
-@unpublished{zheng2018dags,
- archiveprefix = {arXiv},
- author = {Zheng, Xun and Aragam, Bryon and Ravikumar, Pradeep and Xing, Eric P},
- date = {2018},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- eprint = {1803.01422},
- eprinttype = {arxiv},
- title = {Dags with No Tears: {{Continuous}} Optimization for Structure Learning}}
-
-@article{zhu2015optimal,
- author = {Zhu, Rong and Ma, Ping and Mahoney, Michael W and Yu, Bin},
- date = {2015},
- date-added = {2022-12-13 12:58:01 +0100},
- date-modified = {2022-12-13 12:58:01 +0100},
- journaltitle = {arXiv},
- pages = {arXiv--1509},
- title = {Optimal Subsampling Approaches for Large Sample Linear Regression}}
-
-@Article{barber2021predictive,
- author = {Barber, Rina Foygel and Candès, Emmanuel J. and Ramdas, Aaditya and Tibshirani, Ryan J.},
- date = {2021-02},
- journaltitle = {The Annals of Statistics},
- title = {Predictive inference with the jackknife+},
- doi = {10.1214/20-AOS1965},
- issn = {0090-5364, 2168-8966},
- number = {1},
- pages = {486--507},
- url = {https://projecteuclid.org/journals/annals-of-statistics/volume-49/issue-1/Predictive-inference-with-the-jackknife/10.1214/20-AOS1965.full},
- urldate = {2022-12-13},
- volume = {49},
- abstract = {This paper introduces the jackknife+, which is a novel method for constructing predictive confidence intervals. Whereas the jackknife outputs an interval centered at the predicted response of a test point, with the width of the interval determined by the quantiles of leave-one-out residuals, the jackknife+ also uses the leave-one-out predictions at the test point to account for the variability in the fitted regression function. Assuming exchangeable training samples, we prove that this crucial modification permits rigorous coverage guarantees regardless of the distribution of the data points, for any algorithm that treats the training points symmetrically. Such guarantees are not possible for the original jackknife and we demonstrate examples where the coverage rate may actually vanish. Our theoretical and empirical analysis reveals that the jackknife and the jackknife+ intervals achieve nearly exact coverage and have similar lengths whenever the fitting algorithm obeys some form of stability. Further, we extend the jackknife+ to \$K\$-fold cross validation and similarly establish rigorous coverage properties. Our methods are related to cross-conformal prediction proposed by Vovk (Ann. Math. Artif. Intell. 74 (2015) 9–28) and we discuss connections.},
- file = {:Barber2021 - Predictive Inference with the Jackknife+.pdf:PDF},
- keywords = {62F40, 62G08, 62G09, conformal inference, cross-validation, distribution-free, jackknife, leave-one-out, stability},
- publisher = {Institute of Mathematical Statistics},
-}
-
-@TechReport{chouldechova2018frontiers,
- author = {Chouldechova, Alexandra and Roth, Aaron},
- date = {2018-10},
- institution = {arXiv},
- title = {The {Frontiers} of {Fairness} in {Machine} {Learning}},
- doi = {10.48550/arXiv.1810.08810},
- note = {arXiv:1810.08810 [cs, stat] type: article},
- url = {http://arxiv.org/abs/1810.08810},
- urldate = {2022-12-23},
- abstract = {The last few years have seen an explosion of academic and popular interest in algorithmic fairness. Despite this interest and the volume and velocity of work that has been produced recently, the fundamental science of fairness in machine learning is still in a nascent state. In March 2018, we convened a group of experts as part of a CCC visioning workshop to assess the state of the field, and distill the most promising research directions going forward. This report summarizes the findings of that workshop. Along the way, it surveys recent theoretical work in the field and points towards promising directions for research.},
- file = {:chouldechova2018frontiers - The Frontiers of Fairness in Machine Learning.pdf:PDF},
- keywords = {Computer Science - Machine Learning, Computer Science - Data Structures and Algorithms, Computer Science - Computer Science and Game Theory, Statistics - Machine Learning},
-}
-
-@Article{pawelczyk2022probabilistically,
- author = {Pawelczyk, Martin and Datta, Teresa and van-den-Heuvel, Johannes and Kasneci, Gjergji and Lakkaraju, Himabindu},
- date = {2022},
- journaltitle = {arXiv preprint arXiv:2203.06768},
- title = {Probabilistically {Robust} {Recourse}: {Navigating} the {Trade}-offs between {Costs} and {Robustness} in {Algorithmic} {Recourse}},
- file = {:pawelczyk2022probabilistically - Probabilistically Robust Recourse_ Navigating the Trade Offs between Costs and Robustness in Algorithmic Recourse.pdf:PDF},
- shorttitle = {Probabilistically {Robust} {Recourse}},
-}
-
-@InProceedings{stutz2022learning,
- author = {Stutz, David and Dvijotham, Krishnamurthy Dj and Cemgil, Ali Taylan and Doucet, Arnaud},
- date = {2022-05},
- title = {Learning {Optimal} {Conformal} {Classifiers}},
- language = {en},
- url = {https://openreview.net/forum?id=t8O-4LKFVx},
- urldate = {2023-02-13},
- abstract = {Modern deep learning based classifiers show very high accuracy on test data but this does not provide sufficient guarantees for safe deployment, especially in high-stake AI applications such as medical diagnosis. Usually, predictions are obtained without a reliable uncertainty estimate or a formal guarantee. Conformal prediction (CP) addresses these issues by using the classifier's predictions, e.g., its probability estimates, to predict confidence sets containing the true class with a user-specified probability. However, using CP as a separate processing step after training prevents the underlying model from adapting to the prediction of confidence sets. Thus, this paper explores strategies to differentiate through CP during training with the goal of training model with the conformal wrapper end-to-end. In our approach, conformal training (ConfTr), we specifically "simulate" conformalization on mini-batches during training. Compared to standard training, ConfTr reduces the average confidence set size (inefficiency) of state-of-the-art CP methods applied after training. Moreover, it allows to "shape" the confidence sets predicted at test time, which is difficult for standard CP. On experiments with several datasets, we show ConfTr can influence how inefficiency is distributed across classes, or guide the composition of confidence sets in terms of the included classes, while retaining the guarantees offered by CP.},
- file = {:stutz2022learning - Learning Optimal Conformal Classifiers.pdf:PDF},
-}
-
-@InProceedings{grathwohl2020your,
- author = {Grathwohl, Will and Wang, Kuan-Chieh and Jacobsen, Joern-Henrik and Duvenaud, David and Norouzi, Mohammad and Swersky, Kevin},
- date = {2020-03},
- title = {Your classifier is secretly an energy based model and you should treat it like one},
- language = {en},
- url = {https://openreview.net/forum?id=Hkxzx0NtDB},
- urldate = {2023-02-13},
- abstract = {We propose to reinterpret a standard discriminative classifier of p(y{\textbar}x) as an energy based model for the joint distribution p(x, y). In this setting, the standard class probabilities can be easily computed as well as unnormalized values of p(x) and p(x{\textbar}y). Within this framework, standard discriminative architectures may be used and the model can also be trained on unlabeled data. We demonstrate that energy based training of the joint distribution improves calibration, robustness, and out-of-distribution detection while also enabling our models to generate samples rivaling the quality of recent GAN approaches. We improve upon recently proposed techniques for scaling up the training of energy based models and present an approach which adds little overhead compared to standard classification training. Our approach is the first to achieve performance rivaling the state-of-the-art in both generative and discriminative learning within one hybrid model.},
- file = {:grathwohl2020your - Your Classifier Is Secretly an Energy Based Model and You Should Treat It like One.pdf:PDF},
-}
-
-@Comment{jabref-meta: databaseType:biblatex;}
diff --git a/paper/paper.pdf b/paper/paper.pdf
index a4d4aeba9bce899c4d0ab01e3870eb72c94eb6ff..ea433b79055450e09a4dd7ad932103053538d7c8 100644
GIT binary patch
delta 38190
zcmV(%K;plOnhA*W2(ard12Qx<lTaZjf6ZF!kK?ux|9*dkzf}fqxuQtw5ugaJc{Ro5
zQd~D^g3Eo7W!hFxmK;&a#SQYmcgP`W%In>=wpXBEY>MJzW;pYk2Nf^2i+J(Pi`f4E
z<?hAp*F~}@NRcM#;%>8`alXiCOk$QT?$(QUQA+6YhQ@ImeOLBN8b{l@Rm+=1f5cJU
z;{WUA`@6rHby8AfEZ6H~oTRK+BqUC8omYO)X){Iyr;D4@pgTQi!AX&G-JKM+|0~lk
z2ww!9Kk81;%I(*hbO8ipW^JH3C%nkur5Q~bKO0}<A>SlUpg&$vBZ9O55%9&`9%g>m
zsbP7;G8T0k{2z7_8+$F1Op64ce-C}PYUEzwQ`PP5xJ}>fvGsEtC)=@?cF3Vq^-y=M
zT;ucVi2u8?wLPP%+bzmYu#jL&El)r!aW<j&XJA$xWL1}qmHGp$@(dmrolNd}%Per&
zKPqBJLY~6`PAQ;F%#n6h5DH|NsUWpIaIeVnCgmdf=Q1lIEk;p9YTRz6e>n_zK*5;i
zu%yw>j<M4pkyuh>Bu%HAYKhft)r@Po?aRZiuGHg_V>xJPmP}W%$E+y=avnP~PB;_O
z$#&o0o;a}-x+$AZJ(Sglax2NK(+U@$EPeF23L-MDO^Ot8VePrmVDM4@lt&8)M&~w1
zd@x3$8&2i%>0`YfcGxWKf3y<+_G2^DhsKW6wl-O&soRH_a8Tq&%Ow@jZ{^;vw~|e_
zOhokYvnOB!dG)A~O@^QUpB*q0BuxOhV!}XxW(BQTqmuTB-W}GPLH_ps_bsTbJ@7x4
zHv$MF<ni;n!Nomc`Pt}8goFonLB-!`Now2qb4+smUdz}>_x<ntf8(ZZBmsK=m_=4G
z$L?uL1Wi!~{~@&n=Ls|_4T<`JMPMk{YakZ>5qsf{w^P%-S|)Kcw(G{eU3b;Em+erx
zr4rCX5kF#-j3=bvY)aH_7!IG^-tOzZ?|Mb5p=?Q~k8XRrFRQ*&w~q@a5uD`EcH-gj
zW2XkXS1>j`wx$Gqf4S}DFJs-y^{wF=2mr>Rz1fcS+8oj))BsCQc<%NF_URMEg2$ve
zJrPSEJ4>LkVE!-xQo)srydViJP<F4h;nAoa_bYHDrj15TPZmrAm#Lr1zE;wX*~u1L
zuRU|8_`-vEmOv1ZEd)59B1_%Q_oWe%n`9}8^L$D!#vXp%e>^7O9bt&xGs7VmhD=t<
z1z`Zkla+o%va7OfyTNmLkbQH+r`=LS#{=*OXKVS!KW%^D_ZUs0zc;dUqS)5UB#nO6
zkY;;cmv_qyQ0~V$kUt@k^5hA~zo&`;*~X|etnTV|uo!F|vV-9ZmPQl8;*&xG;M`o&
zTWyDh8|C~te<Huti8yLyC6y}s<1z*4d;-$BAhbBCH9KcK0n$hzL=tkIK={Zl=Wpc(
zRG8+GY+>6c9f+RqcjHeOQ62!INK%qf8$+2*aYJA^jW%7QU04cE*F0NaCn9hyE&8@b
zuD2TXlu@EA`T;d41!P1slxuv}KDk%J4Zx+>tOi%xf5~!f4^r>h+s$l<rQ5wH%xsfK
zuUfspah5KIoWcP;B_tL~i3*a(!iI~xuC5GW+`<cws(SD$Dk3!;8#|y@_+M6)JPhT^
z^<x+zqNYQ}T@R0r-Do>{JGSatRjUD{Y`9OOZ-?pPVCef=soJkTbhtW5e{`hU!{Z>~
zc9<o&e~@{~(<j_54gIZ%!5bA8Bvt)yO#)9feGj`u2n`8`bZk0zUp5b<7N-qUusS-O
zr(JM?n;tqq4&#%|qqS^kH6UEnKkjs#pX*$lMsK>ok-sa4$>{84I@qsVM4=znoiQkA
zAv!~&>c*Le5Hdx1HWx_;G*=Tm_iJ?)yB;7=e{?bLz8=UtyveUZHwJ-KW)tZzcCz}Q
z#XfJN;%Z=n;Glri1vH7JR&(!m){|9!bHlL(IS|fND)bz6_{2l@09t3`&lWb3lJ>94
zd1qK~Y?`Cn*M=x{3(bAt;oyIUW=xo*BAyWqn@nD~z|(37=%+vc;ebJ{`e;Ha-M5p5
ze|hNsWrU;{=ta|D(br3mj^PJh%Cg$IVapq;6QjtCeu?W;yRzNdg{=LEXe~EJHz6Z4
z3?P&J)^hvyn|o!WATx$YOh6nxVdveQ_MYE=`ThGV1;UfU6o)3OAW*%)5&hN<1!@-W
z4E^tZS{C4=)pQf{QbaEuCq#H&Of#7Of2O-Il}A=o-K=uRgmQWod+VLIX8((plZI+R
zIY`3E@y+e4`y>hmA&~j|X$e7Agg%QopT*$7d+=<i>xm9xy0Hh4S>s#uPb{+8p`Cl7
zUKf(a;sg+cPs#h&2&$!Hi^?0&Y5AiLHDF`AKI-ztK;3V`X+~&vCWDu_fUHe{f6BC1
zNBnmfVM}ILA62+K6|)0lkKE-!2l{<&<N7%8JV8n0XoaAW{vd6$b6tI>gQ;KZwU4H4
zePq?wa^uP!r)tPrs;aLKdK&|bv9^EV*9!_EpUuK|!7f%l&cwkpVuIzc;*licTt!73
zM_<-!&%a~CE8L{eQ^{7j-tO#ef7Q7OJ}X$412}>~ef}^yXt4k~P$1(HJisER{D6(%
z$dV_VFS74N#%G8ObWWEW9uQgR8#p6*nl4zHWAR&SsSze)1qxF)yfF0xR!N}Yg=u=b
zv$rTr-NJe^G^WCQ5H2Avz6h6kiMvOZ_gm|I?7LphG1iF~sCYDU5Q1bye=#*yv`{%*
zmpj@O{0XOXPniFI7)9W3oH^WcCD&?sHYZtQNkU?71EN#N#vu+z%(gWW9__oeEgt*y
zh@D~^n}xO3XLW63hmJ~~%{f6RdN^3x)avl9jr6#OVHa0nTq#^;p*<I77Iy79IvO%f
zMgEk)vQS_UtwYj<bH50Ke^?f}lf)%5ClNB6$T+WSN(DiR!2%K46eDpdhKZ=ZwI2KE
zy)-%~PMy;|_yc6%cp5p_Rfb)yTo3`l##rH*%@JA1HWkHOAG~V1>cg20I3<ar4_XdV
z9?tCMi)oH9@zy~d1eRQ(O8F8Tbh%=IzL0SD5q@Q_x3^Zg&Tb*De<@o{{c##!1wD)~
zP`sOjnp;7fN-iPKBudcfJT}H*rz_o0T}Ie_JaP0I;zU7GeX~1peX)qkd$nm}6~b2%
z7=SQcB=DcRTI`qInM&`=+Ny8Wl=WV@-nCUx_tsA+SD>k7HMn?m?GzY-<@H#d0@$*u
zAY2|~<D*OnahY^8f9-+^8q1%Sy5uxR^CxW4^wH=u19z01-NaQLP+a2$h=<C+3RKk{
z<mXrBaWRFX@t5Wgq-_tYYzC=*cDg}G*_CQRp>?ARze#pdi`9L#c*tpD5WKqYHTSj@
zs2&Qz4MKMPhbsk?#5%v{&lOOTLQ;4NCsIDkY_xpRlYFWffAXnSH7c(eSgyr<qtm5K
z{iT4dtLt~$rHDXm8mU6aQ(dJ4Cp)y>AHaG(nP7{IWP(0duuNp;O5ICV%+|a>vg`Yj
z4FhtAM}T8AxA!T3Ewusus!K1nnu5aT!WE~mXoC$o;MJub2jOqRAzG0^#b;8jf@d^l
zUCwFMh*f;Ee;igmg%dlpTuRAJrCQgrlrJL90i}>Pnx=C>2{_$fd-D3^o)0-OBjA2c
z2cOMxlVOOSTy&G)Zq7`pbYC)HpdQ|Nb-f(w%GpOR)z}R4U1|P_UFk!Zpax&2@PMf+
zqAnI1rj}hQHSP}=u++y64@ux!J;-BFz4X}#zzP76f0G!C{{{eNxSmPAR&~12;=~Gx
zNVoZ4NlQ>7#Iqd{WPqSRURqiFT&rQ%SxsNIVLb{ks>X}Tnbwhg>Od24J!WUXP8np%
zc~EFcc;Vo@i0UPc{F(E@fjX8uUSd;c-?nL?JD+Ao=2a^trl`EIHU0v$;9$Z}iIe=|
zHoWCUe;~n|Ifx!IMYGvarump=s;TbtE)kz<u@^qYQsV(^BAmYm;OTwH;d7Ab=U_j)
zeXURHT60K&R+>K_JW`VK(`QWCrw&{%gbx%#$&B+*%`z7?K;!f=T_!AQWUk1e8$Bol
zIv;piLs(Wf3LnggkS);)m?yT<hD$c}plmmOf6XELL)n*mfUF*0P;>RfO(0_Tn|rrE
zVT@+{302eAWfN6{blJh_u2Zs|p$4D@jI(FQ5)2eS4p32i;_zJGj~8cZkSEA_05+)u
zn@qz&*V}7qsXs!)=BhLm1sD(>w!TNlQiQoN*BUh7C`6a=J{@35f2^usDk<%@<z|p}
zM2CKCTYn8NbyzzeazfT<mOck^)TM7j2k;$ghE^1uaLS*b)W(FeKr60_FcS)y9@ig(
z79`K@)yPzzzPfwyABc#1VUrO=NC7sJP$4LP)md9_+sGAu*RPnz80Ze?er6WN0^TIv
z1lTQt*e$kA{2)uTBSaBZU25;Y?{_GP5-m!k9Tn|9Xoe!socYdm&LOo>!V9ZJ5XJ_P
zgdr`o!W|OY3hRu}PPmAPgx8p&2+;|n1!>naMno(#UPP=jV349P2^?%>u3881vX~Qp
zW(QshGgv!|U8B%GD&eqW(7wle1+Z}*o2ejN!0e51;71YLgI_dJc<_rJGx()oga^Nr
zFk@@2oCrXoZR+8XVWEP@97qI~k?|8|Fbkk}fdv6_NJ*d%U}`Y|E~uy!0dq*bh&iex
z0(2AsDPm3-BDAET4k*PMA4G&;HsODN5Z1Fsw6FoYXxt}-Ibn9XXRwYSy&)*VC;<U%
zWNm>yv6eueoDva=OrT3ph{7gNgfqaAtZ+^lkzgKgV-JdVfjlfA9Pws6dkml>Jk;Rn
zBI_I=Tx5<Os)Yr$V%-9&20O3u1m=U5wdt`sy(50)g%1j=H7go~8t&m!;zv+_ijMpU
zIyHv(#Y}jtMX*q;_n0wg7&X`wdYY}^12e%R0y5TMhzt{eNHJR-U=c{Ep&W8h6NDib
zn8XV50Hh6!=Gjf)faD|fG^C(ln1*lxTWS0u(!2wf6Ci0Al_>EAY#%L9;0Fi8R&Z53
zfnUK_DKTW2fgok?{9I&LBKxI(xGqHYM!cHmgT<(rNPQ-L`l<Kov!#F^eO5|c@v!9K
zmWOXVjChz=L;1JhVZy^_<y3X&ZuicVhkkXH^DwK}Y^WC1;?14ER@BRC$$K6?R0ubu
zhKKEyIop?Yd0UBlP~Z~7A3Ti9y<6CreKsE}%!>;1sJi#>N)zMbDw{Wdq1h$4s~B&e
z&Cd>6xB=-e)w)UDD*O2EDv-0v8Tu8%MFpi~biccDdwX8jNomnRefidxj!0tnC3#oj
zFF_3(@RYm%*sQ==9o~juT(??zH9qPPw#lzv1a4j}8q^gtd?CUb$dkb=_4TE$$QL18
zV{`ksURT@~b+3EsF+BW#rP;vu^%Rmf({K6TJ_)Nm(k+|2dgyG!X>c?ux8`t7#(0*{
z8&D7RMPKR`+&5Xice}jvSpIFn+>lN#D`-W<a8m!<Ngeh8>%*38Vc6hQ*9T2~*f;$#
zT$AR!dHU=Z_ECJOt7CiE&W_~ea?zVgi07UvTm6P6|NX6w;l4qC7yByO-1)Uyd|hGq
zeIG-AI?X4;(WAWB-~Xag#Izoj+qCw$$MF+NY9&63>~gfYVV39Tz3lprX)dx4{rkL^
zy)Gt;e6pCMv}3|v_A#Fq%h@2Gb3I+&`7Ixg`ac(s;uDhul$n&wO%DuaSSiqflw=o^
zNdZ2exO(R;Z+f*@?aIf!?B`-O%x5JheUrV*-e<4z2PNMPZ#`IuPsT;5P>&iFCCYu&
z=w2$+%tj|^%Z;y=w+ryhejQD|W*6t@<qp|JX}jzy``gF&{Co9fv6%jtWe*Pzlh8;P
zf60@KD1Rr*@%ZK-9ST>JzHKzlCmB)qx6-L3QO&oJnE<z>*h+_9L81?;(_O;Mf_>u>
z-3;pKl~16mC%3-#T)MBXO9{i~5Bf{mcoA3Vaed7-e%f4<h@}Iuk7XZb#o#K3E#YNv
zE=6{oKQ5m7!xQ&Qd+rzG%iJ%nq>C%Tf8t7}7>@Yl+Mh(}P|Yr?@kdoI>U@-*n=6IF
z1KM<OQICDmt4c)#x(ji<Xgv*#fQUd(UcsOs9XD_tVAvDFNF)<FzYPqpoaFYi3k-KU
z7}k*Fju$yXYJ{-0Rwfw50hS7(I}F5BOx@B#oi3vPh9i(}rL*Lu5A9^or0GCse}aOh
zeLJ0oN)XO%bUG@jQ#(Jj@X}g3m}+*?OD(1B=Ac^M*`}J8rV)CYMv{MF8d>2;E1b!F
z*O{g*$v{SUtKc6<`LMs`7iLaNdiII-Z8Vz|Gn{6N{zMkD`|MkPFe~QSxWC9BvjKE5
z8{KA;{^E8u0c1Xx)8XCO=|Vn9fA5SiPDyCaE4aPJ5$&zC+BSyTySt(&k*sU0)5JUy
zFXA1wf{}sWijtO=-veroTo77-m?u%6Lbs^jr85W>;#Y7fNM+nn+A;(wN*!Ct5Pe|V
z!p@-2asuftLi7%lQOgu;muNj$4D3ek_eU$UeiE+yCh|q`%CATn3cn#;f6Z<^uLR-O
zqpR7TvJ}t}K0*RXUnYL{nw*Tz@{3qWE;TyfqJKL+76lzjWE~~p#C14s6n3@AN&{0k
zIH5NFc?o5{u92&3GW~j=-4?}H;GCZ<V;Dp#7X@4@5%C**Bay7RWC>CKxAa5_=NN*{
z!v9UEDih<pG^KT(-%8})f9^*4cvxy`G*~{|8l_b>EFLD~qCcEx@87<C`vU^thpTBm
z_`5a7@@|KNLlG?ODhUuVD+w&0yynj!fkS@clx*Om$QZ^e=~y6(BjbO8|4pgYZkfJD
zXd<GHKSse6-XEvs%!UIAxu63fw-YACc0!AIPfxdQ&fwdI>`2qbf1m$!Q)WNd|Lq<O
zj<DQ)SN_*&FaK+LVgA?aJa6=R-sJo6U8S2viI(`g8LVl`?{Vwy25#U?I&twf%m1|;
z&GKP(KkHAwj0SUD+}$rn!~8UPWFy5Uj|5E!L;Giw0_C19zh@Kj_)02o{x5`hekpCr
z*II~e^l%Xeqzx!xZ22p(f`z)jO5l8<W5={sCv1M5wMXOw@|lY5gpRCqKD85?h&Cd2
zAtWzDLqobx=#u1b;O&G^bEMFL5a(8xoi$n_q7K^8(c=7Y9SBKDW2gh650byfw-cgW
zLq6+3sDtG1_!Z%Q01_3NrIXQ269F`r(JKNI12Hu+lkp2Ff2~+ekJ~s5z4ur6whqwI
zNVa5Kr=mr-MHdTnF-T8s51v?=XzSR9Wp_INekoD@%5-Mqw3nEYOj#r!$*0iZE}+5t
z7xd)$X8Yp$oy-DBWD+I8b{{aF5+>syWt7l34Yoz_Is8HeE#QZ3zHa{-l><s-mV!tF
z7O*5HsbrJhe-ZTF>SA*hi8y?ydlYoFEBiv(0S~)_!r89w@)E`TSoc+ja=pj*cJ;C8
zJO6H8`8_A?hej<GT4n0K-6@09y-{Ar8luP@c{$UZn{@q-(JT-|rYU;ETbE^oCh5{R
zmS!iq3rD7cq!ACUKt)oCe#G<9@cloZ%u5t?)v2~Zf9q)o&M`3Fa2opcx*b;PzCf{Q
z>kSj(cU`E~2IZAI1|GG&^Jl(s2a4vA&0@kOEV-k6-fmbHeti4!>%xeq904DeBw~C}
zXQ_NbZ{Y+Ulb8!|0w<Z|sLII_MW3$(rQrwf8jlnj=B+}ZH@dn*QCYixXw!`|xiP9U
zC@+Ewe;%s~<;5cJaxWfXwlogQOaJz5ZnwQ|RjKmnla<;;BD|k}jS%uIXs*1~gR3%n
zj;p}GNyeA1vKHl!^X`$yybbpoW)1o__)VGT-8Ohf7bm{o>CQXfCmBq&D(~vnoC!9_
zE|O-+rD%&8U^SgQ$j<a21L=xIka5xE!P}=ifB5#1;YU?<x~@)aFZE7U#t()Z9#_$q
zXD4Aw63#EqLY|Ox=9g#{vaDa&<d@xJzc2!hTNd;L&@2`OJ&rT<m=(r?ZU=d_VLW`m
z!I(L2#pkJ&l?LCngSvf6^roFKebbc6#=+J^J9EsTBI;pOx5I}1iH}b=f4^E&$rn~Z
ze`Q?g{%;h2R@>Kx((f^blK%?CpJ>jfp=1)inaq1LA9^zgg*Ln1*qXt4cz;mchJ`Lr
zc8qNo-a6EqC~*am+s+Jm;o3E<xbl-59Mrd6t}Er0_HBJcsZlk=*DD-q<3(GcaZ5PP
z9~;4-hp&p~+%%SnwVl~Dnd_kmoSDoCe<y;am*z6#lti3O2uM~Z^9%v%ECdJ%2Vi23
z046;G;P4XyVAdi4!_Rzvx>qnzla}<I8ib`2@qIvu1%ltRQLU+g?UWBV1_}cO_Y0*_
zep{{z_`(vXs%rzTt9{)<X&I~iKm?$^L)qFx!Pi#!ZhPm@1Hl$J@42NuP6$o2e~U{!
zV<dCc*ux{clyHgj2-m=pZLhCg=`I|L9!bMUmFlYKcir$dkZqN>yMyN-P~1^E^vB%$
z*7Udks@(<5N&aT$^ppN3gaBfj%~K*_-YXzX7Zw;`7J#tt%hFFFx$1D_$br)`T?;+}
zbQzqv12l1urpl|<_-yjF(>vYde^uvS#$pSC0onyfJk&_m5)tbCsw|ls$0S{wB{?)>
zy|cM_Cg^!axVXF|Mw|fI8{;p3H}Csd{9{HqUGN*0czy#yp&$nO*0K|pI)Wn{!GWcw
zw$%Wci5*U1#Ftxbz8UDDFD*rZbkZA(AY3bT_06E3g@Jv30L<vFx0o5Wf4or+G}<$%
zHG!X7#bmTL@Neq8Ev9SLhnqLkl?g!MZ=m*Ac{yHX_v89H&@cg!i^Rp1GRlBc%Lz5{
z?`OqKp2CKvM99Ve96TAgl$ezbn8mUIvkD??>8yYI0Oi$nLM}Xcy7L-Mjyno05<EFA
z60qVK*=+maZxopGf1|8Hf6Uw<Ci%ID@vbus@<7`seAQnh1U8N%yk%U|Jr^3#5oGiW
zO%$RmrY0&M5yL2eJ}U*hW4{I;j$=^ENGxO2dA$KMp<lkXU~l03h5qq;6Z5-_-%Or!
z<-8)nHk=Dr0bLY6HdqWSM&d}$?piQV_Nw)xYx7}k#%Jt=!!kI$e^i~<fUK?eHd$kv
zq(z~}QZsG`{7l=sm%&K~F{fHW2f_+~AH%0b-U^5<jAQay&NhVNq#HT+6I&f=xIJk9
z2ZvF+iOAjSvK**fr~KrtnlSmqJ$`xfbpHnUn<rTp@3V+n#023y3a(-q5tg9t5-ERf
zU;GDB2RfMwWo~41gVs#9)=U9+Qv^3NFf^A@paB*GF*7-nQ6VRPthi%zWo@@69NSLC
ztk||~+s=+{r()Zw*tS!#ZQC}#dfvBBpC0Gy{?mPatTEO-aZOzFo_mi)Mx>xjD`;Y8
zWGZH7>rBf?$G}A>XK7>P;-q9}BWK4cL#t$J?n3xi&I&_DChTZx=xk|cD`Mzu%0;Md
zYC<StYD~z)M99d0$;k;rMks7&@8M`^ZsANwp{k@#Nli`jpOSw}2#q}cL-W_u$<o}G
zko@n9t0}<F-p16{`7gx(wWG4BDWS84DWRDqz?4u}UO`h*PMnZJTuzlx+|<_8(GWnW
z;9>-@G$xd>G&Z$$GNmLmvvVW_{QE#?Y-ej?`A=z1bbqmb1f2*C37zasjV=GWnYtUB
z+W(`XA+$Gjw6Sz@`uk33=|pJmXlU#Fw+GI4gqF6(02h;g68I}Ov-?*<_KtRcEo}bk
z{(>pkIXOESJ6hU16aI~=AR_kfJe@5Jo&SmLWcgQ3XlM4<(!|c#<)4!NrTYu>SL<wO
zY3oGjZ0hcR{7)<+Q$iC<CwqXQ$KTk0!R#F^|CNS|lclZse`Y{K=xAzg=x72kb#nR(
z^B4P{?)+z+g#SzRhW7RVkAHdF{mbfq=3wdUWC}2&gJER)8`IeNZ)|f*TNwI(W<}E0
z%#M(e;oou-7yJLvxtcout3ee1%naq<Bn(aLYylpBgeInDF!XYE&VNG^QvBbsO!wbM
z@_z%x|1AXnx6u3lM(%&*=zqDy|F8G?ze0<-0044^Hh%};-#doz?~*aJCH%W*2xSQW
zSu_Ac$N!&?p^YWL<NxIKU#-<m|DEpt0Zh`_@NZiLZO#8u!$8OIZ=t1=n5Da^iGroG
zu?3-jnIYhB&;BJ>wKXwy1X$Xd{^j;xZ6%~-WMKF&Iu#2`V{6-gNMQZ9%GB26zgF;<
zu7BlEFQlL@rXfrHzge*VGF13G2+k@V_J4ExAJxd(nf%YgKQKZ<cJ72;w2W*VgtSZ?
zjDPRn--I}sSbY9BA^!!$_@8H4LuW@zcS7xdzjZS({;T@`@%-mW_rJi1+8W!L{4+Yr
z&W5%of2Zw#3jR?WyEr=jCHY?q;BVFcasJobo0__t8pEtE*%@;OTcu^DJJ(|74o!$?
z_xCb_4u#m4G^iB*_OG_<@k<Na;jA$Ds0pEKoZ|Xw^H@By|2&YQIvnZ60Q~GpF}=!v
z$Mqqh?E8(pN#4OueHciuR|21Yt{&xey&$s<rp3mfcDOsCsH9i?0f9Hw!R!ci{Y>dk
z;`-Ykjr7SL!PhvoJ{G+bsSyaJw&3iXJn`iD>F2=c@X%n-EqLDn4s}i}G#R%crf>2Y
zQIfNVf#XAi@fY};E4x$WVpfZKHFBbVs+%Xy`;rR*>0`_eDr5R3Grsw^An-nez>-m{
zwA<CMP}mrz!UR4Vf-&J~6)C;C&Lzq76}GjDdOof+>&z9-7)#ir#>ge{G?GHrtuJIX
z3IU0Mg-or`coPXV1`ecjMNPFx-f4E0x*z;xqI?LE0=535<w?5&MbUa4&OZ2mP`uqh
zRVrzJ>ax!jXoq+$L208IkK7_29C#Q61T?Kesm`OGAm5bASYr71cz%GSY&K#Dl}}n#
z&JC@CL-MCJ#dIOpxJh5_lQF+x|3a47v}q*ZijvGFyZ63N)@$j}?&?kxWetR%=Qo%>
z?r)SX-i7jL7}AvJz62_jhUSZZk<0UCN!B|5L|8DY8EkB%(T+UP*{Q(EY8Oh$?1^C8
zOX9}MBz(GN!Pp~r<yXy58G{dXkFZ9Vj~T2*ks1Yl)TcLR0GvuyYe3f%a(8G&VriSz
zzvM@V+JM9wt?c^Ad`xaY3crJ>`GF)=xNYRM*+-I-+3!PpySipB#^vRIR-u#Zci}I7
z2D|h&c_jm5YR(qv*f*rxZ%n&FWUZ&fSri&sj~G|usi%MeqpD2ubtyJU;gq@Oy1J`a
zZ#~ImDUREj%8;o?aZ7ts_f(0SkW^?4`bpFm$z+REvDzTn1ca9Ahq%`{sUbbNx(C^#
zF+ToQU={>YAZ+G*&_EJ@LA)2dEL4A~B1Q#|V73G{QNq#>&e>BGfB-SclUwrzYpuC_
zN|2|`t9lhVxMHzocG9<Xx5QxCjS}>@i(NP8>$9SU=DScVG(=+f-d*U+7q#W90>=R!
zmD9~Rl?l<g$7wwjIr??CSC+dHs<j~3Mk$OipV6D9C#Azdhfz*{Yx{?r)9%!Sk#s}D
zZwq>U#r*VoOR_&`Q{<DmR=V(P@-m>|c>cLJTTROZrIO(zCZIcvs0<WKK|K>>>dR0k
z#+%r%#MkMOgHEFw14y`N>!U;*p_n>uha9o|vCPv^r8b@l%X5md(rUXf%g2s&vLdmH
zNtKbEjr;e_M*F9Kz+3E_oA`NV6W$~+B~GssGdkn?#+gsOOa>^(Xt)(F%?$RBl}P59
zT2^XSKw_Hjsv$e!H1&R+$xvYd8|(vaF7h6kQmw<8>=LCv`2hGQZ!Z>!VJQGcAI3oz
zgq?(HesiSEogrNYk@3vy1rDjp#(fWI9^Cw0raJL+y;n(p$d)Hrm<!3+aF)VjIbx`d
z2g~o2Qz8&wVAdddE(|+MxyJe>P9M2GUvXTLiXt6Pw&M4ZLm!VuFJZ)do~&X5&#^Qi
zZ|Scobft)`{Glnf6URGHjRPQ&b~@EWqIW%+PrtcV8I5dA1jQo<-4z3gmV~2k)|VkZ
zpz)RfHrB&`r0+6|@R)Q@QjZR}%BHCH>rc9!un<>PL1CIK=3@FQX%Bvi<F%rmSzfw;
zKlfDMMg*ut0~i=fBqw}*QFxBCLeEQ}xrXS$dPMKEV8zGWMAtTk5%;x6DQNxsq3?~2
zz9V2H98zGmMlO7*Q@F6`v}U^!yMtj|Ir*9@Lna%4Q_udXQ)>PfWUD=E!VT#bb4{wI
zR13SZf?#I3sMMmt$2>%L<DwJ+?XIK4cUtG~-s6|xd<JQI{J)e}x@m_&;xXrPZwgWh
z3oE4Sr~a%Ah2n*;%31n(ERE=|>Pf~HK1d<pji*ij%ID%rbpV$HFN$;@n@OB%-@P5A
zpoAKKW7JR1FC`LPc7A7tnjzrKt!Ig4(JQjaC4Cidd#92#UpVi8o?~rfZ`lM!GOWAS
z&~=Er>zB{Sz+jhQv5EpYS3HGcxARg-r_cTjgilL18=KV^d@hH%>_e|nFJ^*^Whn*r
z$nfm*vZWV=qTyL-1)GUjVlu8ee8jmheBDrgK!3?Zg(}g)0ruROMTi9oh%sEIE!?Yd
zy~A{<Z-^HyDhWBbEa{45Nm0e7Awh74X7FbSrVR002N)=AogI<nJo78ucei^peo#g3
zbqhY?`Jk_g&GD~Vx!xa%08}rv_$u4@^<3<d3hwYR5)#OP7x^cA6}cBYarbl!GPs$4
zGq~MiXeB-vpsO41Bz+}i?3<MkG~YCTO%<*@TIeViTn}fig49#p6<{QKw#<(N-nuDN
z*!V#}F4$B@H@>cN*a$bNrsYa=#;$nve?=-dpw)PQVmxeqIHKqmdHW8G*x1c`mElz8
zO`p9HRRcIbF+8C<Igdy0Fz!K0Ng@S*=83bgz6}xXd6XuJWWm#@j@{|Td**_}^^#%d
zk5gflmRN7H5Pb|}TEzkjFzc3!O<zKgdFJ-ox>u3U`Gwxct`^!WvesxG6S+P?K2^#1
zgjOmgiUzK`P!<i%;F)ke8fmkA9$z`~zbOJdL7t%ww`-L87@LI3hh7Y!SOAHCq|zw(
zB+q*vW~jo3Kc~JG@2!H5TPzlVg1YrX9_jv6O??H&!YF`nqUM^L+GD2o{oFD!6w(em
z|H#vx`t0~25l*?4L?TcKQzv~mY+6_ZGj`+Dpia<ufFD@Ql_A?flq|HEC-0uenK8>6
zyCH|Fb-2bzNCrswNv8KfN|I`S9enAM$6!2xaOj=dC52e#n0@YsmGNM=CNH-0Wxevq
zhL9dRtY3jO{sKu6317+?ta_yRP*H<(u!{rg5z8*)N<hnB)k8T%!e<`aI~pMG1=Zok
zNk(F+!Y?e<YM-|~EMI%lTWWVO4L&dBxs`n;M9RN3V~0X!ylT*`cX5_~L~kL9BK#Kq
zr9ow0Tg!~Qf%k=cgNF{C_1^f093hZE^s_IU!;rb-KvGs~M9Vm7kK=8?9TTR7H#vwp
zp+9Y#ZrI*7*KH(ES}AMrv%`&L&hssgq|KYn`}!xqQQPThk1DK7AO-6^)q8if0RGBC
zMN;D#e6I_@U~oKMFcT(!?3B0!_U=G~>StZi=yf{*FDGD|cxUu35wX}uAzPJbuk4yr
z8uB1rIUn~a+4Qq%kS2BG&b9u=rM)6mM`Qmxkc6Q6(p&K$5$;9AHC?dXlAH1K<q)It
zHZN%zeyEQP4<rQlK<nK}$a@_ZaXD&C01zI@2ba}yMQi=r@V<tBFn~xywxLyZ!isd4
zIf-yNltn}z0ZYkOr;`OJUBS)fQlTQ^dKEcPS}3TWdBj<M5%5RY7AQ5iWy}68q5o3=
zf(T7iE#jBzrH^H!jD4k<d`jST{T*bff1l?_f(e$NIaMMX5IJkiA`q3qvN^TfmD;oZ
zye@h^hi7I6UdS?kEHWn(r&}r<D}F4U(83IBBnNDDycQN$Y|n6!o!Rt644!5f5GQA2
zROvP*I|J|v<=LMwg;{UxZFK<(Qm5278=s@_^qS8B2rhAx;k;rh+fT{isu+(i3_uVc
z_JZ!|3U956Q|%^ac2IgW6HH*ef$E`+RBXcq%i*|;5O58DVqKZ&vl-%I75-9r6NXa<
z<gfIbC6^E_P=+mmxxflpYZzJbMYjGwML*`yK0OO7B-<Xb9+cmGpiWz-W>lGyiPui8
zXDr?qn`7V^eTy)>B}|&D`B)weJvedAI>}SjOnZIrd3^W3-|NAK7Bh90{BwU(!T&*~
za}O&_7X>zdYGLOYa~`j?96RdNM|L-@VGe(-m%<||o3zHkH~UZ@UaTeQ6{qn0HLp6P
z1C|OFfP5R+Z<6}teAknBW8p2X;-yJg6pGhTjr1Uc=4UQl?s~JD?B`op1BnA#?BfH)
zd+P1lXNvYE(_Mn<`%DCDK)XYpGm*v9>E1x<>(qmP^B=#zVH|AWvWsiXK2eZQ%}ei;
z`@Yw1L&$?E*Dr+^Zh~KiD#b_0H8Bup3k9XA?rwR{Zm)As$<dOLhBlwza93VdH!nf!
z6iDHyeNhZVFQKFw$u_XFwFQ?EM81L9@e{3;B~yCOW@F{alJFZ|(&Soak956)&w6)S
zBR@WW@1H22Q{Wq;7Mi&Dlf<ymsdS=W14O1Fp`{+iXu8#VKsV=@u^kSZklM$L<||d$
z{7tYhT=@V*aKbuXxSHT%iX{2NDJi%3k5t9uO)4C)%)j?Pvv>9+?3cKJ^H?8zK?dT!
z3l3g4eYvSFa=^_wPuA-#6_x6uO=aS?0Un`$aSKE#FE@DDI*WV~S`&cPFy4?iWPnt}
z9BauqspYbWE7v`Ju1z@hCK_|)97BM`Z7?1jx?)-b<+~Vf1Zt<~JC+U_fSXLBAV8iK
z7cjPHxb|op!g2G2-yVN!K@+b2Y|TcF3tfJbOULR@px$hHNvw5(GH9CxY*S09xX3Dh
zIi2IpcMn0sdKOODD-F7sA9_8c(pReQT?o4D2Q}@HxRZGarhBG=`Mt6WP6=){K}ij1
zs778#Vkh5v%D^n3^`V!Q!wm$#BtkiBA14e|03E&O@0R)d)7Wgz3b>aSi-_PVaL@`~
zr#fgz*=l9I4J#DiAELPv9&5AbErg+e)~bA)`j<Ezdo;BC2uLLfdG+SO6X3UQUo4GC
zaka(%eow-@3Rc1H!AVPkd@oy#I_Zq%fVBr_e(JlZm<k0q|2~DoIk%z@$t{Yt!}px+
znhJUB102!yrKP9`Qy`&Hj_EOTq|kQ>rtUpXSmFSdGN$FYv`}3{evIf-EIFBfylWj%
z1JjgbVDXN{M^g%%1xxGnNU22RcnwE5+LMs_(=!9$^HM{c+~1Z=m`15xGufP{l_m@c
z7B*y^GcOstbe)sRoBPcoy12jRuy)Zh+J8M^f1h%CV6Or7D8yO1s?ADoB0<l#6e&m1
zZjrA|SCh^uF;W&xCvkJ>BI|vBI{c&L;Yq>m0@5V?AW=&t={0blT4p3N!??m1=RWFy
zFcW_YA6GJL<MYPp#p8rhFSj}<s-vwL<$0gIcn#C76TQz%UtO5yjlAgOWinc(i!9k5
zfs4Bm=14{&EcC+IGRfPw%A{G!aYUC}qmXPHN7JE<J#$culxnxTY$L3H*S!Oyt#x<w
zB{yvJ;)-RMW9W|O;a)21GB}CSq7h#97oP7irOO3;z3pc~ads$?lmj~dtFOKzcNE4S
z@QH@PlUfcYCJP8(3?xrKbFWXbGC@8U6l&Odc^_3ge}CS33?SSF)*?~MqHN<>(PXsM
zIIx=WpRx6LoW5iZu=Njr0^<;`Ne~s4s1>~y$F8gt;iqtLfeZEIH0Gn|obnL3Pm>en
zu~^zw)4VOpl;kvZ%j+%;y)L;8FgOT|_)U#n6G-W`60)w_VmDn}ni_%-Pm%iIq*cYh
z&ZGjs*i_U4xNLtGe_^8CFt>}Vm942`nsnp@s9jmVrE#KIB!nq{Po2NkOf}Ddqe4)S
zBmBOWhlQ<dlF<hqPVX<eHol%E>3(%!R8jQNC--vUHyhLT)m`cnz8A^LN$7*(Z|nO5
z<9dFPqpHu+JEzIR9u=vw1UzMEbbBD`qCcbK1hdgTLDMN}D3ZiJGCpK}15RgGbk}&g
zok8tcqh04*F)i1B146;_An>aWf!hjFz+ZU$jS_Pr-p4JbnRZ5EAWSkQL<ImfbY<Cz
zPqK<Zu7C8t>gYX+@a@v4OE`JXe{n0ymyF~olDsFCY>Rkl3~UlTVS{j&M9&Tx^$0Rh
z7MX(A$RmpRzV5#-E!x6%yZ0sv*YMILe5%pCOAp6Zuie{!VIOy2WCQU&@Juy^PLF@=
z@2Jcl4m81qW!Wvkpi!u839XI7BbA3w*)e%{=tW*MLu$i#2RR-K2I#F~{rFseZfx%F
zw!~Pw3NG6?xE>DQFsf>$hEeTOi_^X8<k+npZt5dKGKHR@vAd=KC;@<zQJ~002Vv50
zG`8#u`s|{A4iKH6$EY=C{o6igya&Ucl5LAPU*>0+ezV-iTrsbdMRZl|j)2O3jl%O`
z3k?9BQmQ&46MsDov=CK}IDb%2JtS(9%7VOnpw?E{a%{rI{FZ5`4vUGI*wdl73u+w?
z>ezzgN!Rm|&l1;84ILD+*iAQTrmJ!CCdPM)83YP{stng2AX0%S^$N;$kZCn)p?~xB
zQrnI0q-fjPP3>R$YTXOLHa%3~^e@2hG_%_Epo198dUomw6G}{yDevbWnwigI*(daf
zzfC{@=3bwABTOTTa&1Y_|8?2io#|3X1(iRD1~<xqH}93Psbi-oBE^<BCR|Xa)SteF
z<fY7i$dfnZ&gD2qC}0?J^{We2xmZVQ=c?K>pCx5;Ku3U!4Mj#N8d2^P+BXnyvJ^84
zF&dAd-*6H_iW7a{!(;T(bEwZCEPg%@;a6RI_{9w^>c-++c1{@w;_T5~#53M&*+yAu
z?Sap%gj{tBMU2TI$h98~gb07u=~=7w9x{P{dc}?ZYr|KN`Rd}2(#VyAEY(gNiy>H-
zy?bD|KD`)RUUdfq;@j6S(wcj23ze*Wsm(jp{ofCc?e!1UlQ!5f?c@w<P1Tp52j`Gu
zlRzqZ?21O)YTovD8+(Vp2fx(?hGj*~?NcVcR}Hy(2kdyk{hD6YzRjw3Jylz5Kx03D
zOco}~=7wn>`=qpuM`7zQsGdO9Xm#bem?=1^_1R0*leAB<%tAHzOIw)Lvro!Tphj|&
zHHa;97$1@K+*N+lizWY1yb~pVtC^}HK?1?9o8}^bUNt|URM9f+NLlUFmDcAZpA>(Y
z<R3?wnFzH*L+jwycc>=XC<D#Tqek|Bq5ipq;878%qs*x&Dt}z_J8Sbg<y5h}to5ql
zwX=w-?9y_n%z3GuXdr>xEAf%)T-S@|GmRFDF0^6Xc+^4|D%lw)ZGusO5!AQxbRKc$
zM*<>+7eoL?M%d0sZ-s7KYxkH5Ls{3NV28%HzRxpVTb7c8(liI|+!k*On8vDqx+55j
zE?KV$l!*1Avd||nEh~Punr&Wu`sa&gC|Y*N3a%9btfoHg$0@oxMI+ZiM^Rcm@M9Hd
zhfgL;_r&_0q|etTxs`O?F#*kJ|L5;032wZRNgkO*s^rWmnyOL6Q>M7hMqd*Gx>W=`
z7363Kk5#O5@K>3(9%<|odKi#@<AH>(<lAxJjMEvemrWLW>)B^Z4q;K(SSj~C8?W0+
zsIHD6#o@=k!Mw@&e8BK?xU3YWJ)*33*=%@c%!Pft`vb7@Lm1YkYr!4DSI|8<@N|)!
zj(>HFKNQZ*<2XDlB6S7RX13hf2t^Cyu#klc(G;(`sQliyR>iJYL6rS}z=7~B6_Nf%
zOOL%@aB@O#*e{g?u=&+Z;(K4E)Gxfy+fF9JaIal)WsP=REm?@#ySp+Y&SQL->BKB%
zvBUjoB^kLIF$1|eZGsYlNqU3CPkTnb12XB2<<qBK=X|S9fTY1xziW9+TgzwoA*$_^
zfG#c?5%i6<dCV>hG!Er|d|8n!G3gZ?!ORAl%7S6N#g9r1Oa3R?9MB4`J-|eOOukF0
zVTb-pi`?FP*qc)>;HP?LxrO}Ja2GILAQv<9ICg^%mYG+o_KVZQ)zRK31gnz4m}yNs
z6=T6^xh-jZkpq+<Z^s|d`#HBpSL24rH(IH2wCs6fJg5ZWI&fHj7L`VSvuaW(;3UVP
z(C{IRpB4tg1$~JWck&QeHVUJ(^Iuyxd^QL?<m06qjrDEusT$ZSS)t4n%++=k%b^iJ
zifAic&`49awYM>Z2n|FY>|?_S+KDtLz!wZpJS;p4=n1yQ;FGW=ct%5}=-qWiaM)D?
zfR-n_7qbnI2Y&T`laP9wzNJ@|3wj_kgJjxQ!F{vsll<<jhF3G<M?*ewMP=(}s5tuO
zvZ_?M>b*irKt4oKVS4IrYzmlXuWd&QjK$QgmlnOGg_WMIbw@k6EwEFvP&mb)?h=_F
zh5f)1gk)0xb5*myu05~^3*a=+M^5`bd;#vgW5Q5xQFfDmtN#_KO_ox53H&Zi!k$i>
z9OKK`oFi63?VrSoY?x!_o#cW>xK03UGvYv5vZA{zaQcLaHc~Kw+;59!!LJVoj0k8!
zvbsP%Wti_UmfU|0|CC%W02L7SPO(7~6_CppF~sTH1Rn~zaQ13N4e<*L{(CV_5c+z~
ztuY)uMadU`L-BxvAChg_R=cmsdO(ESQ>UMQa`Gi@^4@^P*H~K7Y5W*WPL}UQze8GV
zES=12R1(OFdLkhB@iaAHqlRhAM@za@N%<pp#fc5;hS8_>i+7n6n)DC?{yk}SsuR_V
zNHyR&a<M<i3EbWMVj%@A-{b0+3=iS(EB+sCtb2EV09-M%qVdRzu+5LIKj*ihuu2>V
zGVBq*rjHe@j|(-<ZPA1`UA@BhRw8GVAIEwE{UjAc`fy0jPM-S75WCRP1Z>I3gN9m|
zoymXP#PSJOQT1(MSn$p9%D%#F-D_>lJ>PVJX$%`b4MCpJ5IU)+R%jRhQLhRf)R`O~
znH4mDmVa~M`cd$@LLi(Ez>iT!Kh5>JCRS}3d)Mvm8BN$hU)ySP);FZWmapW#0KD%i
zO~V0^=cVbNg~u8-V@o*(7ROh48BgM$vE4p(y~Etq4KHU80ki8}m|uue(&Q>Vr44Pf
zPNaQSdFO$di*vb(|I!YbYS$SwvgAx-!a4AN-Gs*(j;a><MA8@fo#3ZZq0)OrTfdDM
z8b|EnwPUuKz8@FwxZyN@qd8ZKh#5r*z5qI@cA?AWFggW|oje~biDP=)X8~OJC%82u
zevuDuC3LtOB4Mg_nkqt<=JE7?6nrB<GO&(R^bdn)cug)7i8nxoV;|c`;EN8FC(ZGH
zlUq37eZ(#UyjKH%nz(8zl5;8z=~Z$%LP3^m{ESVAG&NWk3QWrl26o|14hGZ`s;1>N
zv5B1L@Tbh3!WEvalmkH~4#pYqYB-6!W~wHEkokvVp#adx<Cu$f2p5_t&T2sue6CVr
z2TT^ph8|(`tZ~5@o|PYu$V3yQDLUeRS|JI67ISO2`h6f@&j0d%wJXnSs6pQWC7#mP
zk9k%TQp(D%ke&f4&*eUBOg2%)ijq@H+pJKJTX)V{mDoyR_IZ$WBs2QdIli@E@8-t?
zLB1?fMYxI6c=8h3Hd5L*#fd!vaZd~M1Gv$ArnNgSC2CDXozE;RSNYu^Oi(y~7sE?`
zL}eC|VZ(ERz%VCpY4V(oIQ<~{%$3XWzAzGKR`3C`D3<F>nix?#*ySXJ`#<$!gE7E|
zI-Rb_pQ3&@?@~{Sirj()7y*7;0s~>tW<=(T9gTey^oKBye0+ln*m>Xgr-(og#P$OT
z8^+2{{o|S%Lkd00OELuLPtvb{p(-{@Y}qEnDAcWk8P^4TYCX*qk&gL(XA7&)swHk=
zKa$j=1fotJAaX{AiuRR;pm3ejhp2H9BoH)>K}KYB#Q(?!={6=-PaB6z3b}q3CMf4j
z|31;N7Nz<|W{Ray2-W8|^CN)O-03dKhxTJYxFw=eq^zKGC<Jg7uQtVhjr~bP^_rrS
zRfft+5UZL7cX7v->lJjZQ!&aYfxD2-5dDFiF1&Zq$RO#qscL59IsweLxbMWOOb?e~
zS*cVUR=hZEQil$iY0@?v7ESBUn~nx23<e*yh3kYG^y@Hd+jjXO@aajoU%CG_+Hffr
z6lsisAUq!!FymS9K4z7F$My<bXtF5*Lr)TP4unrJe+<53B^x*Q@lo<q!3OLcbN6Cj
zipM71azfx&Y*<kMwjfte#*|u}dKKs<mGma@3*D3kLqm5M`+c)Q$p@)Sa@^hfH2V!%
z@c<MzEWQ}73RC&W993w0eV75%?VYg<dn9~TeK3BBkucT7gGPvdJCF|!;O(A|#NR)8
zHPkI5(VEr0Cou}2FXJpYSOF>zW_JiC%3MX}`ZgOW;9l-SdAu3A;M~kgN6d{ABiFvX
zfOgm7f#sTN>Mj#g$fuAl99Ky{Jpdx}-5e;d{5D9pVEme9fK*S>Nc*5}*158Qo0g!~
z$qQ_==kPo7ad8}f+c5bl03(+4Rc2Ok5}T^_)sy6RMdhBM5lor7DmfD#&MQ44(tWMl
zPp2M&Wx)4dh_`qNM9s$>8S9hmk&R7_KS)YI){2m8G2SdVLvVIAI@QyTgl%PhT+yEq
zp;}8<c?c2T0@h}t4U3=7NPf<qy_OAC`e~&V*P=ny0P|&kjZiR|3RWA@DU4<frt*Mi
zu$5aSTUhd)@_N}*dy`*pi1S0U61O7ssG%#O!do|}A(njBE<L?}c2<jh)Bytlki$|<
zpofb3pa>GE=iAN~Xx3XFP69TkPr&c-{L`vGiEOGya|k%Qw&Y>9g+5=TO2NVKq$q%T
zfI!0a`{=oUd!7|Qev4y}=|1#|r90P;_!TKVre|P(b{Lyv&*wa_jTa!saJQWf$q*QJ
zO`zA;`avE<NF2y?9LDnvHxCRKKckZtd^}XAsbKnV{8Tw#kc(phzdF0Q41{tRviy<y
zYV1<*C?4T9%QOCm2Gf@F)9wr6rZ%xoW-ZjKZKZvG?UzB7$UYi|NuSQ;yqVs2#d<Dp
zDR=Vdy!F9Gt@Um?2b0T#H3DW5E^EGdzh2~@S9fC3%7#vGv%Nd@Sdk%N2@SF4WrSMH
z_pQO!)AvUNaN)yr&Xkf3mf;Fga6|RzQt|WmaLs3{geQTDIqQ|7MG60#{W$cB*UqL%
zmrId<-4Ayb?407GX_E&u&sw-ftZZkkjrl@Z_jN%JzzO`(tdc5Kr^$j!CN2|`v;Qu4
zowh^J8cI;J`K?m7+jh3_Y)72vpzHTKN}z=L7RbY%5T)x7O^xoa{!W%iq2gj;152|r
zsT`a_9IHL4dJt;!IQF%Pb~4EjLgA(|Zn)@wi@`ImKJ;C2Q<qDeVPt*~12}qG(jQMS
z<s=utXX%gnRx6l4zK7XpS0kjl0vIVf?0wTY0S{RCmB7tvB!!gQGv(JW3zNn}IynSH
z1bO=7x;}GLc07+uPl*|QX14cp%SeNx{uxSNLtkg+05qF-=J@;W`Xr8=*N^I3i*2cY
z51(4+2!#2X(2v1N9{w?v0a<<p)a7*xbxDUV2D~Y8e-pL#7fLCuSXl`@t)Q)64clF5
z*b2v3`)RB&;&7~btA=U}EkH5(GJ@wNx`M-PsYftt$cJolSWxF0!B#lEkU1EpGOou|
z_(&u9@tEN#4;5t<w^hekK&*(YW8&L?GBSXD@S04llQ9kumsS2)Gy7e9ZDLG}d!$`9
z`@_};+Bjh;iT#Ik-fjKu#L+#h!R*33z|g!<R{dUR3rF`)>^)A405m|$zXi?yNzKXR
zukALZW_C2<$r(puAe}|~j4s(tZO@VJi4-DmbzMUX4x`0`{^^+MK((#5jDaE8M$1YX
zf0_+^L3J^3O;kMq4tOD7zW*543nQH{jV|X6R(dSak*kSeh|yY?PwVl6|BMoPKpcgH
zNyd_=%8ED&IQq3z%Oq~JFz)Z5FI2TR(O)SB#x@;qobU<bZzAe&&lNd4#Ae)qdn!_P
z0IY<Up}Ktx=n=X+bI>L0{3i!w)>xrTe}ys6?=+ar^^Y3%b~69gv-RXH8Cw&o<=H5H
zHKCIk;FghjoQP%WU2&*4tcf<>ZFVp&Yn{@AKXuNK(!(ucsvSZB^nd!>3M(d=Q5_R;
z8D_aVR?>Il7o0m|(1?I>*Y+on<;<e+r2Fng#xKP-ujqbpTxHPEH^$yf5dZl=f0nPG
zc`?M6g;%NRnRv+ZF&*8oV0qc6=^74-NJ2WUSMe}@v%+P1D`LgQK3vW+=F*{e%75DS
zp&tNZN0zu-<CcVCIhGiI-|=N-&tW<2<(nmN?a%xl-|%zJNx{B8Z6O(8XB(+jW}~(Z
zrKfpgNLhq@wwYpghY~b`FXzBAe>UhuH@R<~ejETLrLEI^{0`YKwcx5=3UNo|*NpO$
zSlCn8_iK^VchL8Tqrqp8@d}c@DsbYoCTOj!M%udypkeFVYYZXF$#eNWdkClim6U$g
zv)`nci6;@jp4%f0AhGkxQRH;mPMV-)x0F!0IF#|9oATg_P9D(2Mj~PNe`I(>*jpVL
zAegpLj>>bFxv6W5L93fM;Ek_izluxoKX7c`4beawpD8{=g=oi^9$fIY*)yq4)U3lA
zo!HJ;ANj#gri-1yDlbd1p|gk+;-Fn+uLQWUcjq3cK!;yyR}x?TtvD22#E<T^gjnS)
ziOxZC`s@~<(d@TWkcnUBe-Mp(>u-&>6QG1$GXht1HFRNP&lZ&NJXxG#{)o~~@i6^o
zCEostk5u=Rz&=Q3#cAH0G~K>yQ#NRupq>bQCj1MKR}-NKr@6~^4fR^KaU4y(0Xel2
zd^plW;<>H;jiq11WKJyW;(hU`!pZMvtV9(J@mrBqL1!-6%1ZD1e|1^6QM!v(@&{%o
z-YcAmXwc8XSoX|e{9m$Lv#3H0rZ97QHc6|5CSV(Vpd81$*}05IEb32(<7{`?iYTC>
z#|~jXe+X^Ctt3GlFny|h-4fMoVqfaax0{{eg2<16nVr9{BZ4(LL3{R~n^e#f9y{rL
zN~89dj(00&#HtTje>n1SG`AI{M_b=(Vy~0`{9*s8O0cGf7e<eNs(==-$Xx%GIitH6
zBWP${`WcY?B!6|Eb+`SH!<9iMY0hzmz!mV5(jVe%WmmFFn?gNCRo-WveAy$otTA`L
zG!y~+kk!eygiJSJASjk+eB;Zo847Cs1N~|)!q@_OmPP7*e{sOqns+WqK`LQxqcJxq
z%}7_g=d$}pNZJSA3=vv%x8q2`DjUCMqK+#%{M!cW0OVUg0Qv2M*tnG`;ascX(eJzV
z#sVmR=Q$z3)+qfGw6Vh5vU>+2<W%!GctYMch`iNyvq5#qj;lbDmo<v1v>Ineh-3pY
z0=;GT@RoD1e~+zWYeA7(l2_)`fqG@;giarP_SJxq^7dSxsDdK!i3{`YKJ3*&KZ4RT
zL8TP&b-JXTI2}?hg#Wsniy$FE5tPNhl8xpnibBF!N4nBcYVB1hXdjVS4zRFG|Ea*z
zs2h=?&o!PpBSdkhs^&RJCQP4#kJFMFK}VrAW&BZFf5&d2hWxo!fAUW5EhCO&0YvT%
zdb*HQod$K~k`QInU_!rwY|va(C>BXg7IJkBJQ+r4eD;`hFUxqemltq^b<;_y)`829
zX(%*ypGakok5P(!^>#MA1ISuPf8ZrO$uAq%^R*LbEbiWKPRkOP#S$)bP`8nY12(YG
znvHaof9P7qgzxbgIz<!fV#q=UBWEY^bn|1LiQe4I^R4ATXj7(aI^h&jJH9q(7vbO8
z-v7g;^kO|}oud#Y?WjroOm`Bj;v{t#v=(=&=2Px3c5xvc-v=n&a|kjlS{ccDZqDSX
z973f^m>}*T6WXcl?zyS_N!cmg0Ur@3oR58;f1jkt#=slI18#3-XdsaAY6XkR9BCu=
z$(ZNGiP&8tZU3(F!HbLk?0cub;4ik}zK8D9Umb6Ra>ul#@}x<mObuEh<*+SK<7=i*
z=X;J4$4H%egJ7mD4mC8HQWu|}?tGBB!>-<5AS&5~!lkZzC!Al(sp+R=LgswX;phU-
ze-XbtFXD{CAnV$2TxDl$Z{$@@%qO6p#EPk<kqxTp`KK-toG{H7{A|y*R`a%@RCnS*
zr8WZ>tOQXnP9WNJ4huly)WoFs&Ead<<a6ju3-;~-pWCH>oE>Z29Z>NA@38qyZCczB
z>t%@O<0aj6Yg2;1JDWZz^<rk_*Nn}Ee<;2Ba1kz68eo;IGq67}6<zh1r7^V+p6zQN
z+P+5ClFvh9PCi7AFwq%veK^OjG>(%s)w4C`If?9FcWbpAB01d9T7h{`l{NZX2)2(+
z?iemjl8gQ^bdj{xeDRNB+G)uwNqV--OkB)$Hv|lgU5O9jSw@6N0hcX=R391ae-0gL
zf1M!kOa+%%3x$gTk-qC&9o+NJJ+wDt^;4Jaoso6{-Y_BG8)uF%6nRZ(8^_3%h4)(~
zxL|$$imCnTHZuI5%H61(2001&XqCQb$Zrv}PX)~uxW=6Z*`Q~Pi$ZWRH^R}fIMo}z
zIZ_iKh<AM`e#*!E<Yh1Rz#s<~f6oKW>Fh7K9v*sRZl-bZn>h^vo|4m_S<P2=yb|{>
z>kK7m`dyk6HNS2Q3m^MpKaO~J*5JTD=ZNf!m<7G3^obK2u_Xx5*ecH-f&`Wn8O;X6
zS`8%NFsgTno&~GCroRZj^%!E%<n&p%%v|B;g)&SR3Th|KNw>^rjeEECe_*>oVQ0mR
z@ikIRLL4yUJ|U%5CnQ0k_Lw?<9oOA!<ZHtud4{Qkd!pdBEh1hPg;chdI^M3<K_F5m
z%Az6siaK}pG||z26oH$rwQ(}0%=g;q9LKCb5O%LUyKLwXwblEXYCG=~(L;uC1um~O
zDy{euanoekyVbv0Y7ndCf1bUaCRdQd9@|$FFzTuG6#2*NuJQ^kS`|7nMQs4`!#t>!
zRT6br_~L4;5D!R5i_5PY2&*}yV9rMw@pCXRdf1?Ww5@~t+t5j>sqgWR>~m*=y*OP{
zA&;YLAM9$N9$MIkEM#gk)jdPyu{3@c#A`Qv1&{RkORfntO2*Fbe`+ANNfA{Agb4Bn
z5w-aDf>kWecH69^Y=z{tH*~x%*2Um&pvoU;wfMuTm7g=0u+sEq?t@@qP{LM^litAW
zkrKb(<D@gXrTqLhVG<(tgc-9hhjy_ZMB<Q#q3ImFry5}$20Kf%c59HNV=T5m+*T{l
zlX_XyixUIpsb^yee=&vH=v`Y^6sBN@s!hUtiDZQZp{=Z2yfgP>FYMG_y`~}N&JJA?
zc50;#)`6`(-l0KIB<;|X4Z!waR4(ZrsM4YxIr!ZX%X+Duc7{n6u@M3OADSmULxs?(
z4I3;$YS%(yk_nvZ?3G@2%UUBeitN@k@J5tbvBiUItKJ3vf8*3s75s=<8H<--wnT(}
zn=0ix0o#%iaz>K?d!Kn1L40XidRm3+RlW^koHa-OlU+(*-sdEtxoCHqCN;<7vcy+O
z^aLqn+>*W^-i|HNU{!CxETY&d1eX;pD;np8#A(EOfQ|2ySx)m|osWyfA<!8S=XFFp
zc^9zqNiW<Re+yM=7)#Q$)2x)&RjswhX)El_=E5%Z%^G8Z13@Z3{OQLlB2<vEhx6No
z&(&#yl(Oh%j$mo4^`C8>KRSb%O-NqaF=c~LfNT(x4$LZ_1kMrH0uHU^re7r|htPPZ
z&>I=?f69Nh?ZFgl%eHAl)lvBtl8(m4#OY*N)$-)jfAnT=E=cnehc({}>TKXi!P>eG
zc#`>c5)?Nks{-xL6f<}pE?-(^I2@1!jVwBy#j<vTOYYH4!i)t3-fY1_O_yzUId5kU
zqN_;A0)1-j4?aiv;KuWM1Z;|C970kd<3t##6U|FU6U%Z<CFSd)+4(cLid*X!*&u`@
zFGj94e`qb^LsvnaNSc9U$H7Q)H8_YK%!;?AoC}m;4*h87=sP+TS&Uli4dFAoQ7+L+
zK?KaL_p>IRK0}^%V)dLMkJ5Pk;IB-Cq2Zx+VZmWoBOvf*yW@;a2=Z9FZcprkevL1R
zRsj5%v&S+P;4XzZ8|1yeKk#auf{L>L-C_+ke^o@!^vB}C+?f4UCMHF>rRR(6VY-%b
zOWURKfE-nTMMTG=AJO_qFtdw#cti`&*l5Wp^@X6X!qChms8<BsH5BcDE`f)6JV$2E
z#kY=Fbh$HiIXa3tiI@Xba?S*Jd<B2xX`A9VX}QNlQ=jh^wjqlo7bXBbH-uf>2wmby
ze{IgmG37%N`%Z!{%|gmK>{yYD8mlF=(8D|d0a_RvrtuA39|WX1O2J#^h!=}$0-Dt6
z)aP8B*{2<MLD;K$KNE#r<=KS{a_i3dV`%H>j+YV)h<t;%AoBr4xFBXYnpN3I{tFfr
z8y-wx)=M6$>`x9|fC%ip<BN|gs{TV&e<AHBfN?*fGr?(=vM_CS(q5hdwnyfAc!o~@
z$j`=%`pgA09sNuvvKe>B_hGaTyTOI{$m^ncP{(0lFSzyVOh}I?K5+{3+)*(aNJ~>i
z^c_Xg_zh>;sBapX8Ch+Eb8Rg4P$ES*XFltQEa=M33g5mGz02VMq*riGZU!4{e-;Cm
zCutq)l4?26waZFNfb)7D7@UFRl-I#e=jN<YGA3jPeFfOaobFXhl^%FjpxcXYdiWq!
zfLH)>WzGuM`}hNSM9dcO9_ij<Zzv6qX<)rVYf=3ZF>yL*!>#5J=b^3|xbKt=2JCmr
ziMu%{08+b5FyPT8?~4YcMpkUPf7Lm99I6lXd)3@Zj3(}4lTxgLtbY#x%l2y`+-lpi
z-B3YJqNN3IfwQa}pnXl(kh67hBJO(=zE2~Ws9<XUVZ6fhY8lT1<6^)V%XD(cB#T||
z@#S9b)$<aIq|%S0Rs1Uf()L_I;1#C6JRJCXwoZ(_b1d_=z@#yJ^X~Y=e-+%>S)<A;
zD?)_`n!+{5MVTGCQXyQ&L*zOc#P74ieBsaYz4nQhZj`h6S_NeF9B41`g{Kdhi>W>j
z=K(I<kVBW#SYnc@N!XqZQ(ipa_YNr0zCgCqGm@6HR@RI167l6(tFr<J@aN%c)_J$O
zKKOBI+8q;238YkT4Z<Y^f61=sWmp=)J*E3nGC**>^MQ>0T$GCCiI#Hb*|<&>gn^i~
zN`Ay?7*XHGT<E)>?k8Q20dPtOzOwv;tric1fg_z=NRp$k0ZiI(6-I{5jsdbr^L`C6
zIL0}eN}A168JJ;)x@lNl;f{8uImqGkCSQ*0Ior;ONi!DFOo1ygf9ywg7maufhKl%Y
zb0Qi-fUj_zi%Gh(mwDLmk^tVw_75(S++T6{sEK>r3ol<;$V2Xd<0pzFz(t#^0(ZZ-
zWyzX;``+?V2s-^jiNjZY7XcLGKI!*|7@1Lbir0*)CC6Hwx-6791Tp2%tQjtA<TMc-
zNs`S6blC3(k5ATSf4epDnn4ZyY=1<=3yb?~9efcBOYV|@i)Z|%*+$EZNnjabeF5h+
zN&Z~by^f?+=L6`&ze)@{l?hX=qzYQ$pE~Q5C%vX97WvxG>O<dEh~|D;HpyHw1QIT~
zp4Z%X-PK{chtY?XGF~VutO6y(;y|rm7*D;r6Q(GprzdJ(f5`+S&Od)oHY`)-hZs>F
zaMBlGv`4|c|0)&rMVNU7QOCW>t!byl?RKC7v!eWYUbMJ4!MPOMC|4cI0}P*;Wg{PT
zCBQQi-IYjv4k0%U(#~#pROJf9>G@(<mr`!4HaUToxkv~GthH0}z>=OsgAQ1aCz-l~
zlX48G<S>g4e+w+MwM01?5ty;Lm65=*OZ1UJyE!?9R_|JALKv1IJV@Q9Z|(;i5I^}9
zo7|k=4@nW`(qa*LcG9I+)m~B;9W1sJ3g~*_N%mYuGEiOBZz@QP<)2{tc?SQOkioXc
z1CFfdyyxL7bb27~MyDgaO7WBuuuqk#CQgs$4BCTCe-&MkU+K-mb7-Z0%vl--qvknO
z^>8LmLXpx6zB|6S+a=8^Ls>3326LU(=En{^e54;%(gk#cg+DImqfRr7dwkTN%Q}|N
zym_`$Xr1lt5;E7ObI9`*9%*8(3D>q1WY}mKds5^BXA$~B8=v@?xaba_`h9YugBfp8
zM)Vn3e-(wHI0ma(1$f#Nz=%j(803~y(#^X9qiT@5?!`r9$wADZv(l!>GUSp9eFyz)
z`m|xYyYl~JV4=W6IhjKzOW8;f--D}1sqnned|w~G#BBfR3CkYwd(!a+dzB@A8gEFg
zS$EBj;ODD2S?1gjfZ_*3Z<|R&6wji-p>D-ze?#P==#P<2^N4_-6JdduDp>eeoW008
zF%(UGYZ!9<PqOoS(f*%<qmJsn@k{hfW6^`HpSx>H5SkeJW?htiMz{&(8#aad<pG?u
z_Q3sw&l0fQsa)7N=%rA$%kj&aqiv@X20j^awspa@nzs)|nRhscs$i>Ygp)BTT-`0f
ze|sdM9y56IO8!16V<R>=@QGhl?c3DIT5vbDGbsw~Ct3k{@?|$}tFn6R;}u+-YL|Mb
z+!rLh8k1FgO|6e<9T>MTI~G2hB_UVPDMzW6R^kkV-BSa}D&2;llRvgaq+?9M;YYC5
z4JM2r(_CvJneD#8#}!{*U>i8rM;MeRf3bEbdS5zU9p+jfsg`7AvR}rekHIcvTaZ0A
z>iCEK2)MU$FxV@VJ-wkElHR8Y$Sx#bldGN$qsg!^bLg5JZK_zqZ|dk^BaDfBWvuQw
zxTk_Fy!FzJ#e}(K3e8y6#Su!FPCnmLQRMAN(=8c@CYXt({rtGW!|cz4plxZvf2Sb(
z?Hv1>dFkhPu)ZEhdfrP~!oB%-voQ=GC@hCjHs}6~Mls^*C)ctbS8h$`o(rT7-#KFU
z#bIu@zp!t#;utW5r|SXmB<Kn-5?5>$ECS^!DJWvMJoI074~2B53Tq6OIkegQirPUb
z7r7tRYCs-vh*z<RRO%z(c*|x;e+3b#>uuS(hokjDfTcj6VD~8@_eY4!HqE;B@{Ocw
zGxO@@R(U%iMuPEUz*+C|dWl<7fTn+vu7|hdl^}J`C++o^=r;$Hz0`c&PKq8!t-&yL
z@oDv-BOUjwvn4wW$YI)-m4V&uy_aJyxNo*U?WCfTnyMHbdR-ML)UodBe_?RRu-3*F
zIMx-Z#wT~UeXHD{O$4thmR*(QYXk8Qco1&2V^$*Efo56FVY;1^kK1Bsl1P#7hCo#;
zG=7n3OTKX3LsQ=kN^etS{{ZIuO^rpio{Mwbo#@PI?$Ip|I@_C2T++CjPN0O7E=I{|
zZbp<DuJz<VATy7~&{Rj$f4ClzE5ZEm{0TF3f!`Bo7Uvqin*By{F$>?h@>Qgdg^34F
z*HC7tToBf?ZSKR8$SyQ>yM#8Xk^dPpmcoIFrdK0n+Oyp=OO!VN9t9D?mM(XmSk}5(
zgYyWC;FE*VjslLKLp<Br=~-}~Vq*Wgf;s2mR`8KQyqqYS+?CElfBl)t`9A4VWra@k
z)r6ZSDR#PZRPiALIX*rPWjk`<j8Db}A=D!`IYV9@r>*}Zr(i?aH_o+UnUh`Z1Xy6O
z7-eSc>1VZr3Ji5ztJ8R2YW!V1#W|J-M*LUVCp|lp;UEI%`M6tFeL=PQSr<d-zFLd=
zrjOkR6EdnIR<FUWe^QU}@+O{K8A;&6EPLdgfX{<WD5(*SdmwcDu|T&BkJHGNqSUdc
z5nwG|v9r+hn^Zb67`0||`W_LT6$ZDudG8XEv=zk-vDu<U;7y<A{EkOH_UP@$LtWmh
z9Z$w5vYf(|v7K4x>yfqz+lt*EYZam1HE=mu?g_m_-e0~+e;(npv=-$6j1*!;NP@Qe
z60YuYoQ0S5?t3-+p6<kfwmNcER7|<!gF<fq3Z^i%KTlZRxq4^dM4`^apM*FVpj3XQ
zAiyb};$LK3?u~(LONCqft@XrR9TaiDeLrH8Hd^XO(;=^!A~q7|KYlUs0u3@WGW8tR
zIxyPr#2_q`f9jufrkYL3y<t**`ytLWCDJ(7*0!q?LQI7m5Wc$xNCkrRxN#Z>xZOE4
z;4UO?IC$^{cgXRU9bD3#rV8@X--8}Cd7Wd>^=)HXr2{&?<a0%SpUYKk%bx64mON|+
zg1L}ynP`gooVDXN4J<q=tn~rMZc&PA`%)$hHbMdXf5`Pkc?qf@(G(nl$|J59nSe&s
zYn`O3#rj#mJDa~#>@(b~?WB4)R&qIjzc;67fKVRmlU?13c0pRZ^nmjDSD~juLZs8>
zUl*+IQ-HC(E&;%!!FGa5LWO3L0c=1_OVij@L;*o2Gars7#fHjqWMFT#mcnPUCuTMm
zH7%2se>J;;rqtDu3wf1GvRL6b?>Q!xth?~SyI5SnB#=+ahl6}!z?j;~b|ND<UFn{v
z4lls2xv7?F8hielI9%7L$U8}Gip8T@DdKf=#>A<P3Nt9%zD|v3kL49|RYLxxMQ-}`
z@L{Sw#vGf1>)>D4!+o{NovIzS=)-IYOcQZee^tO_Vbvq~I5p*H&)Vb4a+i6LZ<!RY
z5$J(a5Q-JBLDol5Mym)M@Z6$V{E<<+<_S-1<pk$DFVip{7LSAKRIuQu7>u}=@5zh^
zK`0_^uQI=nZSm=Fdcacf5vNe4!T=_E98fEL_yn?kw6{8j7!Dr}@Jp%J&Wd|N2NWLc
ze?1Phd=XB~`7uZ{W}KDZnZG)>Xd=}8f%iF2QtazlhY{7Z3uN77NG4uGr%{&iysn<O
zP+4H1aOyEx!(AR>F<{tX(&?|?AeZdyfk<R34Q0&+Zb_WzU$Afv=<0ZH)%i(zGxfA9
zn%GJL#=*uTE`LX}E;9u#n8sY5K&`w^f5aDxrfCrjwX@&&5kqR{(LT9Hu%$owkR?py
zn-dG-7(50MTAou5ZN0a*iWtvtY+N?xoTFX#ohb|wHx<&H{N(6BtN=Y4W<9&n3i(c6
z$??ZdoKZOt^sC#q6YdAK!S$WdAA(fcuRMKIfG56pZV-4UF|j=F3r;ncw~}@Vf0E%a
zGcI=?8D$I<bvb!Gb72d<eL<C?42X#?IohdEcz5C$74+lVg8dnm!s-4JxerFr!avpd
z11>L3R_3{*8Dq6e{M+s;Oy*4@j&;vnXPH&h;CJi6Rg*OQVEl3H?!u*wFhUrJvxX&V
zP2-Xw#^iwLL>`<m<!r@T*IoC+e~;=A$1k?&L#X;ICAz~r?@tTR1XXn6I$3uZlW&Yp
z0}M$+m}>n>vgvzm{qjc^k~x57<4CcShH*#@$Zy#ls&UMCne)!ssnvVc55<7_)Fp&{
zz?oM6{=xpr{48mA9R>p8ahsFY2~d8gr-yzU-T}cx1J&mAf~gkN4;ZW^f7z=e`>vgz
zkv%OuV|U2M`dvs>Ry;=$`F*JFV54Nvbf}f0-z6kI8J`wCwcQ?OTP62wIH@*HQG&)Q
zfj=?9HtKcyFfK4FIJV~xX@S`MP+xy}FV%1u`M=|PxB5nDpP~Hz!)9wj-{|QzCkYca
zJRq@C#)rvPvx5?e_5TIce+4S~1@r*{8oJ_dU5Z`t332Pvgw_LDcXF5uB!s)THn{Cx
zR$Wn3vhaF<sJ)G}s_>gw*mQrD^E1^X$_u!T^bMi#ud*=iB}y6cMMU`Tk<jTD-*H`r
zw+dc^x+5?8S6xhFc%`8c0bdnZ*miXhpx?Hg>8SsT%e^~#r4xl>e}vcqP!x)i$gwOU
zr<=`R>th_fGu4rQ(XI~4Jp(&hy=(EV*BtD~@Lb+(K5yJur|r;?L(Y%2(8{-zE`^{l
zX0J`nnOCikMzIICX2`pz?{*295tGM<QZfhG$1QBrtO3%o7p_-)AN~lILNjbdc1HCT
zgVL%*luLk&<fbVIe>Ru1-S4DGo<;9_Gqq2VG~s+mQDADdZaeD(@rVi9Ri5e0B3+Zu
zIwa=_T#D$ca@X48g;A*7G$mVx6I<xO1}Kg7y4kcLg(h>Q9eb9xf1T#OjT#H_UrRgN
z=|GWKZ(<+sp~h0x<xV|0B&E_c+5$9aV+!zD8o^lgD|rxLf6r15(DB6E7J?WZ-s$@Y
zf{fUr4mwf`&y?0dggc7Av*@50NHM}rTYP=9&t9tmP=VES?yV+`30-ABh6yW$|Lxh@
z#cnfTK?RdVxGY<ja{N4@dLgL|^nBmWwF^q78jui=tv~te3+L!F{YL7RjmGx6PJ1F`
z_?~FAq+{<pf0rvdpm!|whNrO3yjG;YwOHwUt~ZWpvM=|4ktEQ#skv#pPJ-9<ba`+j
zC|(w`9F*xoGn#>$B`43G7uRN!O`)}aCaaoH?^8xzJN|2=`wb3Nb$!pf2{4_tU$ORb
zwe&n3R|08PKi-f3ksUiHKl{qYo4dE{BD^;e_bK}uf7xdK`06a+R)BYM^~KP53$)qM
z#Jrw#dfz@ouZjBMd8U?j9@p3JC9P)<QA13?T87y;t8mm7%{gI%iVtv>##{1AHuXB%
zo}jT)Tq&HHmo~-RaL@OYEov&`GiZ5%G~(Ytbu0Nwn2?ty@}Z-rHg==fema0b#lOV(
zh><uFe+Dc_c5&7-e#s4k8vkll8P|w_Do4U6k1N%^CKRQqBvPb=2V#{6+S2drEq+%v
z!Wy;?$X8GenayOl$#ST|p&jJH9E40?t5W|Is<`tOpLsUCOn>Q9k<TVA_GR8Z+9Xzy
zdO}+*%woeIH-j;-XaS79<5k|_S>Sj1{zXhkf847zY0j1Fk}a%`ef!0C1s*v~J#;qp
zC4r0FY77e|-?39tY+WDfhyS{VqQx*Yfw&^1g1w(vEdo5u2+jj4J;d5PnWRW^z_=0&
z(4LSGGY|dQW%#W%Xj`Vf+4uZ9<?bOrpIDS0z-R1C14=;v+O}c9KOWAvg|wU%`YGg|
ze^{RNoYU@vJTVr+OM55LeDi0#^A4L%*RS8T+Y?yx)uP0`@niheN`_oRtyG&>_0s9h
z%&g>MeExt=9W=?k{1(GIT#x|!IN6rpiD3QM2jvJgIj!W_9^o%<q@u(w5Gy(lQlGRU
z#vd{Q02!a+5RQ9@5vce_efWQ;N2FQdf1n_1qvmbfVy4`mgD`o;03N!h5d7AC>!?wB
zpr&c?+70*~cTRiWg)T7i;ittd)np)+^cL#HAnfs46E_I|>ko4wJE5_Jbab^DLC$^N
zSfeMGJ{Nc1DR#74BSLaJ!F9u<yBU$@7n}mAnSH>P^6EPeEUUS8Uw44-`0Rb^e--^>
zD2BJ`E$&3gaKsa++GS>elyK-eSDc1Mbe)$_^rx0Re-^4LsdDxcr!vxB+`hW?@Id14
zabZe2lhh$h&LD2#zcZjozuOo%aT%5kj4>*Pw0iVRXiEM_&YBz|s>qhprOv#|2~dfb
z(0MTnpPeG{Za^5entMFm+RVT8XW(Z{_x>@7s%kTJD2OsW#;~J16IKx#9-8t2C~sF;
zGZN~Xo}nGhlEPDSn&*nsy`&W7HQv+@3rrqL^BhRe{@o2%p&o9qQdOeIR`S8{3`D)p
zL>35ojnIpNbQk+BsiF`vg*N;)lhI5Qmy+rMBLp`&G&GYzY7_%FH#n0|At`^YR%KLN
z-IB#Zupq&`gS$H+xVr@n)-=#aH_~`;2|<H<u;A|Q8Z5X=kOX%RHYDG?_hx-F^JnJ$
zxOd%C`_!(gv+LBlw6D}OS;ftvW*`|T1jfqA#vuey0o$25J83}eRG@-NtQsIoXL*>Z
zE#L{`MWdyabOf2gz)*;kDGYxk1keGQ1EfGe02dd4Q&3P4jTRsYwRd*}TUx;Y4B8qx
zjLgg||FZnv0x)y`2lC|U1h#|#=$<}YK(<hOI}il+MDc%i)C7S5Fe?zi0&EKcNUEyo
z$*af$7-Utn0kR+n$nmKNHD@zhFc6>w27(|?AVz=%)Dd9&*8%{9Ld<``ze{ssd*Tvz
z0+<4v>_I^AlN-nl2(teTu>kBrj&@)tr>AcK*a=|iXbOQnodFCA07HPb&gQ=pc(PkS
z|0HDZ2z}aN_k=x>sX?7!PC!SnJq+*^RZU9fuRLK^rm)|!oxo2p0BZ5H(;NzP{$0`^
z?1|<Hg_(jOP5>Cl4fcOKmKg|O4tBD)HFbZA{X}N(2>z3XvlAF%`L7IE0FEF_Q%7@K
zkdxCB%@g<U)A?7Ofd5jxslC0e`yX%UpI!gV0St2j*;=rnadJJy1j3$TTY@2I?7#a(
z9%2CnaB}=*H+Q!G2j&8D{BuAIzk7!9DG5_^D8$wsU=FfCV^@EH!k$6`82)c6v;F%-
z{x?YcZxQ(4BJclC+`nt|UoP?g_dfrfTE^MdR>jorsR90)Gk~W-V+sL0%^H9b;P<ex
zHFX619ynm&e{(gp1KYa)Z_a<;tpoZi>;L5ZJN#D&afs!ULady;Y`lM&!A>$@H;}m+
z7zVTgSeV*A9o~N*yEeoe<Y)_qfSyGDbG}bb92|ecw5-5D8^~`pc>jVx5c9v+@TA+H
z3fR@ur1fMJnE#h)`)9M-Q&+&W-0eYt|BR%g3^o7f;y0Cq1k?@S$;!zm2w>&n=Xkn|
zPuEe9i^uDK2>XYK^IvOaQ<x*z4Pd~*#=*e}czXZi`PYBy-QQ@WAwa14?>^FmnL^B;
zTJN8R-!RbG(eX*~Khxl;@_(%VTmlft4Fp7+pMe5}1Fe%%lVKHv8GXZ223;MTh<!o!
z`PEuqvVF^-?LJ8%-vx_J?#hGMYDR?~8r`RN?e8}gn0^d&5ZKbUCx8yK$h@c-JF{_?
z>6-bOfB1j18|A-9-qQ*9JepEkMbhWvc>7~*SY5*??+%%Kw3*ux<>-phm)a%U7mw!B
z9@86`SQUlef?WgmvLa`9k1lrm>XN?aYkyyF`^mG;O;YC1QK+=Sri9+{yRYJ4?k0}s
z)xd{mzg+m8(OqA7;wIlaQ~N=vlAw~69uqfQz6^iO`DX#6h7yTR)$>+093Dq~=T?#;
z)PN6UL3b(q@BL&ds3U>Fr^ncvUBVYw9Axn;%nI5mtuDI07u9nKFW$?rGjONu%e$SL
zX6l&g@>tai%PcTxERv_9YhFq?tJiJ&kl@8+rO(2K`?AK`BPh$O6g#VPA=32G+qW^u
z*<XKG30$}3^4=@=YmK0A<Jbn$zVwLB__afIx$h{UQKo6Z{${swF`FZ=*m^$v)Z?zo
zieJuqR0FRKHw$;d%cw`b1pa;h3Hwz&XI)vgl-XJPmgGt+QQ_wWzOkTEK7_;A&##%p
zZsXO@f9xRXL`EE!<*lJ&PXi32J*6wg&Xs?PlHcUru@UY{HoP$_t|la`{x%yhp%sjx
zR77UZq?}0n$?Bu}3vZKwCUm*RDFmuXYj~0`19)M%J_H8!MEE?bXnLL_<Tp_>Qul9!
zP_s!Zh}wbPW|;CSdx5WY43-?=&G6RJUx*;0hkh<A2%$PmoNIb)W!GR8>N=CBPoRGw
zO2r^}&1cjs4i~bcZ!iJ(lX<#!Tc-8AlO8@)D?e=1u3xIBv#bNf>)d{pO@_?Z{bW1$
zrMt503qcZuDTcejBGsPJH%H%I)@sh(>FI{W(1MDyJKd|&M@MsMm1ji-!WtF4zns`&
zX3|IOW>GT#uq&;pkZgCOti>l$rip)U&z=%QVGF;?G;9eJ6e?sVM=hrSpMU|N)vDP3
z*=B7Ta9gP&o_i^&^AC-3B?4A{1h0i8pb_P9O?Ok;qWDf%%nrPiomdts@*sw&d4{`H
z&ErI$CJ4QX3sd_Sacra#zw>Euram^#xAV>ZGtHKIFO>wG+6WqhuvW2Vz5RbztzyMf
zjnZifrBOxY63^OW_XdTbYq=Ly(18SN;n@WF*3dwR&7@}5>5yJG)<hz@6a}kFDPkZi
zJIv37Ll{@7ZH9qaO!X^{D#LWW4AiB7FCr>lQW-WlLCKsG+><`l5Y~4-mC=4!EXtPg
zv#{lwN=Zld<PI%ley#UeB%FT(w||jua`^r_OZVu7Tt}`)2tG3R*$w}W378(%xFS~j
ztT`_Zzg>nPI@q4cin`#Jh2-7rg^^=i1Z%JUof^R{ix}D-TKTt(blPIrl%7n_6lz9}
z_D(}X+n_l|Ve7Pbv+kp@*A-i1nudesKA>=}Z0uvX>v8W~0#&Hcbs>L{-FW$8<*aO8
zq$WT_d7C=e(zcA1sLf*`d}!mc8hN~-F)W+R$x@1i3f#<Km4vdDY=(W4zHPMbM-}eG
z_uYq(LA!@#Fk56xv%f;as-$L4Bj)9ZM65Ax(^;f&Q6?}Doo??kcZv@Ik35+oTlmIM
zLFgmb)szF=W2l$n3Tc0T<=jH@jwZq5IUQw9G3T?;E>D|lGGB<o8o~u*+Qm8OL!ai4
z&x9xGn&62Vhejy!)FMlNR0Yjk!&hNEA~$UZnx$U<N=Y;)T_4KnC`{)v@f7Ae<NiGs
zyK5&xN$J5<+Vg0B5?r+)LtlH5Z{zJNE!|nK@P=bvn3yaA$C-Z%H=!?w320R{vZ0$k
zoA^kR+v=GX#sy;~NuapR?o@Sp_23flvil^@00p^ikWUnYWpjv0HJ@xa?@w%exhUu$
zt{nM^Bdgj0C3s)B*8J*f-NBd#pB#NcjPNLHWkv_5*8O`OmduYt6!%Nyg%G67>|_)>
z{53+k7U{?#23CJjREmtO4f%(}p!cbEo2(%f?eHQl0J5*e8ig!3dV}|DmY<5HFJs<&
z$5YRBCblc8<<oEAU;9e3!ZNUB6VHZKTrZoKcv+VrLT}bTJkr-Yp-|5Il!7yBkLaxw
zfU~y=yV7&<Y42>h85$Sub_bjl)<2i&54aF?bXjTlr)Gb3;x^Tl2+Vz##0fv)diQ$8
zo4}#zd@&!MLaIu`_xrSQE%bbyCx6O=>J1^59S<q<%Jh%j%@l#+ZyrK<NePxGK26wN
zN5q|$Y5FB!ZVoGraATZ+bIm1Xic}_c=hxnS!m~;{YSKI$z<O-<Bh;>Uf$3gl(FpoV
zb^?op7easWMQ-KI?*`%`kgT@QUOWuO;-!Q7!mDm))g6KHhaDOEXEWvXH=sjSoIC2X
zl~bhc_=Zc{W0+yFnAOl4gxS4_k4;QN6`cD<v%$n6;{*u_>oi+4xhZ_Lz=$H%4_L(`
zChZvlBsi%-$*sejIq0?0IesUO!!9wDtUYt!S+##kAyjTCzPB-hOMrMpGTR*e<;-}0
zq-{p?iX({GoWijMzE3dQaN@0bh@H$mXQrnh*6W`bBQ%h)wZU<UuQ5ODodu<$H^o{g
zODtimcXC0LWWt9u(h}_xf$8$$)G_?~OI19&-4P`$wu12r6ERyUz|X#@x)qF0F0}6v
z!=Qh??XGrJGFHdudeV-&8?PVT$zrR4Xwn}ZU85}UKbf<VY7Pa@T4WYDY5e>wX)ncU
zTVRLshG_me{94k|)m89L;*gJKFj1?guAo}DL;tl?8rS)G*^#}1Q4}O)97y?@LjrN)
z(^U4|?p;kC?{Ip^+!1m|;J``_!{PZ#(X)TO)!F0P$HKQ67^f@7QitiuB3<Lz=_pZ9
z<RH%`XZDHwnHG*z&Xb_8obn8A&mEDKfG2gM!o<5t{(^8&D*e!n*4st@_=-d`3RUq;
z@y15~QqQ-GX3xc%tBGiCIPaVSH0nhul$PaRR^t{g2WA_=DA76c*niqtX~+@8UQ&O2
zsuM-MBKAJw#|%>@;rBvp;pQUeYbQa&9L%0woI~{VR?z)90$-cvlqJOW;Qqnkp7lcm
z(RjOhIzV~!6MRBelAc@51i5Qls$I|IoAxu<gAW#CUWgsFyK`~U6g!-}PAPfXtT+;P
zPSg2A9#+mvhcJ|Y&^GH{YjfsjkqUo>OfTvyz6U2pR;y2w{*2ndu~K0gSsIRZp0PlF
zngtWya)!Jg-hC(Ft&;<<>Dy0gQKfz<)jYBH*nV~SzF-`qT!^?6zKW{I%vz61;H(*f
z>JpK^sc^9;FjQcm+vEB6$Y0<=)G%b(Daln7p>~S0^&o0bH`9baKaGdxhc$o0vRimK
z^^ZKj`Mu(@IU#;7y4LGVx?$&^q70&VG%%6DVRWT~)t5{P*^_XCl+2hYI(5dXS{^cy
zUc#?aU;g;@;hCe*JNYsBa{wI_QPJ^G*xuupTkPV=Q3^AUUGg$PtyFEi!rXA+#EkEb
zsQ;C$2NBlU%(ct(2C`xUg~flPMjO%bD7G+@HYisqh|j~Bcti5r$mUJV4Mr?$nVDq=
zapMZf%j+11I#Vm7k6+@>SluiYaXu6ZOC7yJD34Dv#F6U1-l2mVYbvBK?hMEkpru%!
z5xCjhdtDyczrgmkb_`2UbbZ*41#mI@lSB-CY22u^ZAHLY<Mrwz7LR|T3#!eJp?2L@
za+!wKwCnJ`-#yf>XUA5}t33+hq)qZh&|~XLw~-FmplUS5F4=fB@Rw_=Y*<zpc6Tdi
zJEsQqdJU3|Z1|y<mOxB}JIY1^38JbEcwTeY!1+a4;h0(!dAx5mMQhpt&!u=!=(ZC}
zlrN<<UdkgX+1H<9s!x9gzX;=|#Z;}b&*$M!(rWxhk<${FMyMQydT<wu8O*7!IGxx1
ziH2(|vQ~v6pUurX4b#f%{*LGcoF#Z{;-Zb<A+K_eFdhd#q4d!5sQ>0=DB*~dz3k7$
z()zL#Pk{EAN|n2f5z?$qN!E&YPHij>N1fbOhueI*>g;SfUYvikqgZzp@#<ugnyP-@
zSLWotUoy4sekk6;Ki#t9sdLsp*c|zGY|!VgPCUj+`-b<GDN!KWi)(z!yWEIKH4S3C
z8i~xT_HVM}(FhYE_#`Zd)XHL{NdhIs@KP!;7ajA8g7DL0U6imM<=$24S&cYh9Moo@
zrXmn}0aXv`zEpp_G{iygU7FyN7e7T73c`Q2x3dSe-|kW-jNdiR1Di_Y#j<^N&PtIH
z>(B9G6UQ(<tvo;jj32$9ZR+kC`To_YWM*wq^vnmS-1zR+MgmtorzbbVJztC>Yq~s?
zvS!BzC>5vMB(s{OO0rP|-)D!iYHy)+NTMUWOWc?ym!p5PEfTvvup-*;*iG3-Bk+Ur
zr%dpXN;|Z6ptI@)$=E%cwUQ?K;kiRi#jED^-g#vNcYgt5f2&#eTFKHLcY_`zWwV)k
z2v#Ce!AIi?hSI{NB#y-&YF;d~e6pXt%nzG%(Hbej-8LID&r=>Xx!3*cug3)lHFQA?
zro%F$bOnDe{BkP}I0Y>m*9ukRTwI!;JAaLkftb5g*Ay=q`s^!;Z?1(k_IGX9hB_2R
z40Xu9cx5TN>KC|-Y$Z^<_lAu`7QbaB(mBFs)-4l>1;TSryS_(vD83-qX}CjiY2H1L
zEaBO4jXYx?dguL)!-%s2VRh<dW^Bi!s)nqMMGb$gyTKivgt2Jga;|*p@<Y#Bm9(ak
zVk!+^y8pC8@m1@13$xS(@h8ILK=QB7J<o#)nLnfTA8|xIpSa?yw$IruQ)j-tT`PG9
z{mNBdZz9fzS~AN__OdX(u}TV=@XabW5okoI_-Zpu`CK{5rbQE1rg=LKgFO|`vcrU5
zACiBV-W$g(8<36-0;`wxTcsrKDZ52KQ)14)q-%JcAb)ho)xD~?WmsFVauNKievam*
zBHR<PXYuo(7%4_C4IvrbTOr?|jhSZMo<Ut+uQ9bo|7cky8w<$s0A=j^GuYv`LWJ$K
zTnR>Toc93)6`AO0O#!BZa&XltRSa7pQ^|iVjOtw%uSjPL32zP;oh$~&SM*6<5B*pa
z18m%Srwxt!l!^G!Y-xcTsj{*vZbtnF`YKb2acb&BckvusRC&k`1X86X{3TuTrzSpw
z&^UE)aLUZW@oZg^nRKfR<!0RO#0xR7@D2~McDaAK>#<|c#qtU+n7Zsxuq;FBXhna^
z0RBx7S4sur&;>4Wi&RxOkH{^=^CZ|gd{@^(xD;O4N16W2PEcWSJyjTa-S}g%UJ0Fc
zMECuyQ!0+50iI~=J5km9YRV<Cj&}mbME%XZ^$s+YGw<L9=#@%m%HrPKAOSsdS1}p7
zmtM>9Oh0%i0W{+DJW&O@8Xj>YbOL{>uNILf4rW5WvgEC6Hi|e5+SpW?nV0fVY00Vy
zEN;3l5jxU51H*W;s!K7o2EPKWBjCs;L-D2DkiMW5m^O>Iz=7gEvE1Hun9*+Keh9Hp
z&?Y2RuBtk*+g%YQj~5Iux75(NG+5gM622|tMu_lp=n-`p#W(NwRjlWRS7m=HNN?w-
zL|HyR>yAF+UL!fOU|DHN9q{F4ikTtD8cZj+#8MrPUNx+G^>`#4gMXIr#pq0@ZhPJ{
zNR^3^9#xc-!8=Y?Li{3@@LYksA-r@nx}LnD^5X}J!(PUhC2s&zJrHfR%NHxOGB|wA
z`k39A8t0m67ZjxKwxjH68s&c$E&&-wJw#^R@AFmFlt|0iCm@a+R$N}8Bq<pSJXB!o
z+bPrn&t}X3K%FuZ{AI);?}Q&Pb!2x!;6uT}7i45M_Lt;(m3WzHhbqI0#LDyh+-e)&
zgf2S1kdQCAy{0dTZi6GjCB#3@S-<HfCh^ebjmy2Su`*<wEmk?|()@oG+WM%rc04|i
zo0C$-t4$+?b{w;>W-&Cs!2a%4)@VKC5&506jlOFk4McDmVK7+r5k+nAG;vU!Oq?)-
z&Z!LFPKP5kI1n*EWmQU}Il0h9BCcvc>hnyweciz2GHMbpL!WMph0+)0wCvzUI}3u;
zCG*gtBk~i<@@YghJ6C@v*0T8;a@qKLrPG*YWk2%yrrxBk#Vmo9>z<0@>Po(7K0~Yu
zkG$P>1g+bPdH(u@@MygJDzjQqtAc9=>ra${$5UgIMEtjIBXDe>7jsuN=O!bVkKPFN
zF%jmW)KR2YNsOZ33kQ4<UuJrgi^X+}D2Vd2nWOXL(nYh_&DwvKnOUsGbII*MFw5Wb
z&4+IBv-01(`hXD0pZ8+q*O*q&N0W4$3=v#urXRdQN+CDfi#h!q+CTj@eVku`Ju&^_
zBfJ76TQkHu^y_d(0|-%30?al!r`<Ml8!g=Lik20<7NiZ!2oz}OCUr8!U-Wv;ykSho
z*=6T>%h-d^%*=l~)*vp)$CWKLF}hLvb4AF_>86`Q5P?1Q+j|8g@6v}@W}QdNjA>sa
z=7X>-B0&{Il4>-#++B=Z7S>_)7O&7C%kl36DwMr=4T7tfM#ZFv*k{(ZOFWJjb~ava
zZZtEBTnmwu9{Urgqk?_DdRi0po6ig4ai52b5n@|v-k^WYy#D@y!#tk2t|zGsYs$D!
zwf-#)-~41~KhN5b-<h<yJa(u~bTf|B*kGxo-;9*#y@S+NgMt=bSJRsnv~|BXIJlBf
zL>0G)&&2hBoF9NO=Y5ujU1^Vzb-AzP)!P#k`<)oU)hO=^LsB`5LZ%2vrPK>czNTYW
zaj{9DB*cF%ppXx^6fVc4%^2biX)G)6B^!F-8b+0=3~)s%%ujdA8l+fH2D&tTm0OJ3
z%6x6O`mhfdXA@YLl?a>pZiivv`t+CM;ga-7S(5N#5^g)0H@1GxTJQeFfCfM`X-Rkm
zi`|9IE#6W_`1m{7)dX8a?Z<u9CIkPHxzE|!)+K+h>=usVSX_Cn<e=uXjUa{-0!3r<
zg=rtA1Cm*jke?iB6=*rS{ho;li<&6Jl(zxx=;hx0B1HtRXjvC00~W6N4isS?yWQ2*
zfm!4?JG5lav>Yl~`sB9)nkfRc<09;ne!jgqn-=*HvSL9^r2Rvio1bCROiXQ9^yXXk
z%X5Dg058rrJZ~jztU)G%L4e?WvbGTqP*LM}^EA2H0DVV1jzQ(Y?Rjt}DEOF7qv}P_
zy{g}klf7tDspT%tpgJk+YJV2xYrBE;BQAa3iUOuoYzIl5SKq^Wi-MQ19ZSGLX5qNT
z`B1<@AnB&zC*JL_=f%Cg0cCcN^ypPLv%P<KH;3@|s%&=mj3!KytV?<RF5|%HpH53v
z?ZPA51Xa95gy+_3dB<1SX*R33f@G7KjS~ovdtn_J=xIy2cgPwqofdB8kl95&3Lms4
zPw3ulLW?Bsvu+Tp`S^CFgS2vM!bc=UhpPaQ0?pAvqNTQ%rb|W%Z+ppp+0XmNA@F~^
zSIc?{**IMnq2X^UpW#|*@{^O_MIlCRX?}%XlQkd{dxDnNRtcXV0rw2Lf${t`<4U(3
zwsKCY5_@8p)22X}<^ksT&O4Nr`&rV7TB%)`@a&f#zi05$UkNDOKPb4i_5dw?KAxsQ
z{81}07NQC2%rFT7=lUJomLlAfQA2;{tB}tH5_7Ukpq(L1-mKT2St7vfRTHUmdmm8=
zL06UDiCkXcCQEwtxjfw-qvws<k*<)qebne`qVeD&_%0Ha$CDg%u8?Vzf(3d8htsKj
zvOZg^Qm?xTBf=o<bxtY!R&Vot4`5{9{^y1EMY8gXTYtn+VOmEVt0gj{mNtLsew+Eu
zgY^0^FZmFgdCSIi!}q@go2eMn$vdyOXI)#X8fIO-y<6qKb7CL$!$7O^jN-H<i!eW6
zWY^ZyYVU1CLcJ=-=xv2F=8LOm<H-Hi<nn8&`4`^yY8n?it4`I3!U)O<eVL`M(MuMC
z#Bg?FvURlAMNIaX7zqaW&zyhD26ygE8~AYmyTy;#>0wnVERGqS3wg4l(a(lMUdX(E
zL0Dw9{s<MHw!uBP7#>10d`r@+%e<WucwAZNpF8>iv~ERV-Y}=|epXFIcj&`IMNL^X
z?ioVnrQ9AI?k$tM8GNEGh(mR_Ku%Ed;ocGtwYkbAEwmMroV8j#s$GAZdNzh}r-_$M
zJobyqmh_bg?_3s5kaP7iV*DMC%{0v?9#u?3YG11Z(Cmq|y`&5h-viboBH21BN0-y?
zJNMD_@A_PXvEhVwUHL}}V#;x=qi8M>o4CloME%wix5{!fFT{TuO^zI(*s#6^>yNLz
zi8Z1osnh94s`*vAG2DN=<n_J5RH0_DHWb-d|9bEt5`F3X-tx^|ct+1N)`nT*u`8)8
zSu_vJqPJ4Jzow;f*xe8t0dSJ*DP~1L(=vU*PDO%m1Xv4OF|3L0=%WjJt$A&(=7YWd
zqXkP=p4f-kCllwqLgSu@jBpTzbL+sc%H!(AZX1#ux3&DJ{@#D*P1V{4blUbj@dB1C
zdtc}dl^1T*<>gE{&c;v3@~;hm+?{FByM$}|J0?lW&3Ny<*|~6ohbv>PpjJ2u#hO@)
zgmw`xPvV94Y{g_Q>Pw-$eP3Z=-iHx`)O?!YVzleTj-;t*348(>u4kVl#YU;!W1Z?{
zf5Zxrm@@6cJ&%9c&e6^biwPRKrM*N(+ycv^exCx(`<6<@8lP@&)S7s(-bG_`NN~`v
zd}w0I#N?8#IEag@Zw?5lAC1SmEo6`AlxkgNSbL9`9FX+Tkg*yOdBxv-$0MLy*b|qR
z{QgMmYt4&p+0z-1M_)_5CbVz5w-3#UVYfw-R8w>x^xuEwnO)fTAyIx)Z@`|OjZAIk
zHYp6}NOd$Cq;KVrD6)Mp=MPW_J`(bRm{?PnCL!mFh}7RQ>0G}VRiFw@Z}w>MHv5%x
z%S?0%6b-i&vI5L?aJ*|J%7+bBxS<%FM&aDj;)1~6w3g=w&IQYpnDd*gh8H$O6#?B8
zBJ1p;S@*Pzbe6P#@lp=--$4cs?fh-9X<SX_?3n2Zjjf}6UNhsbz20T_l`OZjVM}_q
z<Uc0+!1p!ZF>}DMqUw`wAbHZO*_-v59>NxkDh;u>{UfP4)-I1U%&I*Tk>wWh(>Q(!
z!)2Tv6+C49)sovu%df;*f{y8eg;MWZ+ZpxjFI}{w%iQ{Z(|GbpHR|GvSEN@+%<a;8
z9qdzZ-M+ou+>TDcPOqz^cs72vKV(Dn>bi;p8~f0-YMOSgG=M8WbqldM_kr}i0mxf~
zuIYoKp^ftQX9>MJ5w5hQSZb58CQI%jX*HF0Jeo_nOWTYY#e$hdbe+%Q3hVXkHz%VC
zYf018m<J4h0z8zA(azpB=-)S7mF?fdGZ5-ITXgn?=ih@!PHr!M-icld*|bEKpnB*>
zS`WNwG|lp9Z57|92#weWAIic>bG=Qwj0kPlL%R07+9n?ApKXnp*QhlnEAS%+Av_mi
zz5k)(Z)6?hTh}K9Py4Z%-Hw;ajfA0vJ-gDZCHiE4EgL;h_g>1~?-zHbhQrhURdVKs
zXEJ6w>H&2|8A50nj=IUnJc6YdoA-du>}4M5A4zq~WvitgcSG2HKgQAVa;CB$)`i8-
zC3CKK6dkKul_|I4_R}2f-Z4mse0w|5bzzg823`_r)QJke<|1l`>z?TLENB`eQx)Zr
zw7LI(+zQ{O6NJ({8iwW}cy~WB);-TKogMNb*iEJ}+x|Kc|D^OY4~VoiSM4T*_?0KI
zQQxKxHO+g2BaG63t-0E?GH&7z45{(#a70FkG;9n+?Qv(%zid|HYsVKGWuh54!O!9I
zHE6lh!!WJqxJ4P#mv&L?5=N{Vx2<@UQ;9ZzSM{6nFABcfq?hdDithR$ops!YdbUB{
zZcdG0cG*?~5wAwMSreGoWKn(*4L6Fc@(Y=*Jvz)cG`W7qxA@#yRHch|j3$9GFedgq
zJcVsC{bn2xTQ6BrjBl;Jb1NQhiysag^44{@G)<a0$6TLhr3V7&8YwrpVD#yNc}I?a
z6j|e}xFWvarITF|Lb_p1;rhsBPNyY@`WsUrq`z=hrfDP4I^i8GPh*1XJ|5xayJ^d0
z!IF%s*TYzP?!m;2)fua_qQ<?JO*REY#1Z;bu3z!I&g^zi=K7wQaB&EmPF_(|-itZ-
zBz!qaebcZtVM291vWFEuMdlmy&OMHQrI6Bw*hkIxtx%%{a=`IJ4Y9d;$Dn7QWg|Yk
z>hLwGafU86UfU~8o6V-~x|F%|FeC>SyiI>4LTFUyn3#i6%djFT|6r#Mc!qX^n>T3S
z*MrJ4L6!&345wr!#~Ua;wXLGVUjD+QnQ;l%R$G!KobhdQnnv;(J~LP~wQZ??;0own
zi8ueD^4zE43z4U#BSIUIk!9X~nu+6&Q21xy6q-ByV;aFhY@9OfI^obbCiUXS&oV>N
zxgNDvz&DWZ^z@CPKN$hR%IUI9f}h%GT4io#`bnIP#FHJ;K(5PgLW>I`u`)9!U4m=?
ze84SPgM30Dm`XcTY#)<}EQlk2rVUbvgDiDgUR1VsPgWFNH$6RkPPfeoQe_POIcl(+
zKcttc%s+R8#|Qsh@O@>4r;xGW$LnguPkB4C$TK?3aj$)hRp^=>YdEuyjy+=gsBU06
z9q{Qp_d+9`i*8M;TCRR?@$h0(4;Mw;dbfM|4t<fNIA*<Jt6X7{+(X$ILyk*7X}*vp
z$;cFRJ!hbvXFd53G2+D8lhI5Qx7Hd0Lm>q?F)}qVmmWL<9tAitGBq)mK|BJE0y8(4
z(L4et4mk=hOl59obZ8(mGC4AnP$4LP?OIuL6FU}u_pi_+87kJkRZ+JpAp~+EkV#kq
zR303slX!6K;DrnG>*pLv-BvGlV&*>Hf=;WYqjSEqN@_QkhA_gpB?@lR!^H_#oCu-e
zE{HG&?vh9*;jV~;iz`i}!*N5Dz;T#vY>z8PG;{FmiH3`7MzjaMniGS3wIIfS!X=62
zz)m`1HL%l?I0?_t_5!DA;w7-tz#4gT4VRE0P{XJIG~6Ip6k!DR0&id{P!KnY2+Ocx
zhzwJWBT`x*wv0%_c#i`J=@^K%3KZwS*`g{OX&mSxVaPF}oQEMNh?X3C6wwatq8%{|
z6*Hiif~i1jtaUvQV!Qx+oD<7`j3vO$3eG@JVkPJW`aoO2sDb~0(1I--8bSdRFp9Or
zapwpKb<(&V2Lw)mrl6M7%z$9H8%@AC&H!hysj~naZ0{@>0{r6~(1Smn7pO4Yp<fE(
znesj1h(`qk8XKGnBwnC14H(^nmO!U~Kae3#pwfVqdl)tb?Bgx*4w!L&4;0>MKnT46
z-2y8x2@_x@f%az#d@dM}G7SPy7-yh7v_K34Y0yHGaEUX}*&@sUmvPCE&SA@_o?sqZ
zAnn<+9z7smId>4Zr-Yt={sO!JitIB`eX<;nFM5Cc5$F!5lLdM9jL<_b(9mJRAyQkD
z9(=HTY(p4=o3IV@>B($=Iy@^D<SU^k2Zw~dFMcja#rplP`vL|AH${&=2QkHDF^9Nk
zIK4+ti}`dp8y52r<Du)gxEc-ir$5P8q%mM+>%|3#8O&fVx*N$7cT^CplPL(1q;20p
zwAVCj6Jorvhlz|=eGk$4FroE*<{m?6OrJaZ2tqCU7;q00O4i4J^naL;>Ao@QVM6VA
zUyDa5qYan`vh!g=#rs&N9wy|rFV!QoR!;Y|Ss%$+z}6$V3h~X+`_&532iou#on|2#
zK?$fl5Mb%AbgmCI9P&9nM>?-mU#FR=Kc9*H7rn3A_`Zg_VWJO36-tVd7UpTFzF-Bo
z^M34ux~f=T!WyA}k(bbjMdPReUK;AK38UEKgab)9UmAhbTJ^EAgpyPHBLbTE7d6=C
z%IDgOLui3Uut@b~>KB=vn$}8Vf%-ulz_;WtG$MFBYLE~v17)kcupi4lcqU%B2nkU=
zV;e7=r=VcsxW=^v`DC=^J;;Lb@;T69KlI@xp*A39yepu8DRGq;QAT;sb;CAZWxq{x
zYx+4~@h^EGM!ds?v6fIga3Q5o=%auB{`2?Yzgc8)h&P#HKE!#-xfJC_k(g^3lh}ha
zY|TF$$89C%dB{Ju$_LCVoAIx52)DC|e%ty`<{0uQFU09Z%5Tsu&-pUX%ld%1SLJ<b
zhdeFoL8={pd0l}uqH1G*X_KI@@l}1u>q2?Y>qdn=wQYyqCC1z)c^wIL0_}}$xWi(d
zd7`8Ui$=~ec3^FT2h=<FVX8euYStAcaPNo`8}@vuay=$Wg7(SlTI&j4fabCgsZ3_s
zZ-+_6$FkLh%ZOx^n=6+Z8{M5;g65A4p9^D^`>k7l(WQ{d`2<CBrg@yeHm$ynz~m4o
z>7Msjb5h%-b~#?n!L=kC%i`oB+tK4j>!6qqXQTVYbQU*WZwGfU^5N6L=cAWTpC2C|
zaRvk9!OfhgNU<L__PcN@+C}5T$p=riz?kmMhq%3m?KYef2KTRu(akM9ItiP<B985H
z96VZo49273-sEOnKo4}BEsDF30JBHW(=0en0h8Oo3^#U9XmLLuji(cOGb;v*Vn*+l
zI}y#{2zYR#^Ap8G+Pye<d3f>(v^!m=9Gq>Ea&Q*=Ka^voYR0cs?Rxa1pwme~?~56-
zq5myr)4Fzc8|_X`Uw(Xjj@sF7<%CVjaSX|Szd<>>Ub*?`XI&M)jVgynU(TNIgDPjA
z*LeyKIJ+KSkZ?HDTdP;ffM&lJhKg-a=Wa9}%xWrGzKu%nUjO^~mp4z3&-RbksRSi(
zSHZ4UNF{N5yHZJ+hJXA@CAP*nHEoQ{wRuk8($l&wavNPfe0llm%jc(OhwtBGiCg1;
zAXxWzC3;BWYai*daO1qD4n9^M5<;xeNZ0(s>KYx;7xa)GfoAXM2|cA}^gaDZ2lSF&
z4u;?6<H7uvUd{%?Vq9D=Vrzz+Li1a(sJfvY2ZnSw#nN*{3mSyd;`?$irp3?UcyLFr
z=?%T5x4-UhiwPakZ*)xW=!8yVN>@pL4foOXihidvI;RoFf-Y`nML`#TP3e*@Cs)O6
zKAg@9`h)&Of6`y=LSVNlb8q+ejy@bdeLK3lT+Yu1lQ+Ak#m%zYOZaYZH@q2Rj1#YA
zmPcIVUJ8#ZFLAlXOHQx6q(v?6q7_5kyPhs*bTXR63OFx*fDxf)WR__gVXJC?`1<_i
z!TGyngwyHW+o^xEi|?n_*uh}5DjwHtC@gt#tk)BKT-ky7HFn_j1gw>ly((Z|(pM0)
zZ|HF?a6i%W7`hL|=5Ts<H;55hM&<P=vv4g$|5F8NG<+GU_;N0Y2(2g}g>490bW1H{
z*T?BrRet~E)8YAX!$liH6<s8MT<IdAG8b7_bCIqU6mA|7s!FrEhxQvj`m^Pw55Z5L
zH+$-GIUW~X4$C|>{(GGDoqk^y^Ti0nd|yr%#nt7wYJ@4}L!1?!Vl%k8xEn=`t8DD5
zSC#YI?sr?Qh$km6J|F(s;ePbhx)6X>Us&9q@n#6%PAdxm<7=zFY_9r$y4u9JEC^IB
zO$Xq%=&GAqh0l^>ku377ZitTM^u=U2y&6q!0M_XGx&Y_ltMjiSguxiL?PUs=ULw-x
zKI-y308SB=;wB-BenbC+utCQ~7lt?sK^NQ`93882h_cfEDH4d0RY*ZnbtMI{W|bGj
zi83{avoSk}=Xr9#d8?LxAH>^`KPyAu)`B05oz~7p9((v}+@{!$S$FX$4^);X(w56M
zl!Hf_lxRcKV&`bG8jm{#C`+Zeh?VpRLakvZ;tc)Ruux;avNDv;whEnQ!(UJ~8xDV9
zT<wD3NoPBui+18djw~-1;SoC_IrazhBD|+>9TjVN9)6b$KM5Uwj%M=(rWC<_-J3yr
zF!1ncbhWshhhIVi>HF!22?X91CeWG<&n7^P(_IsfYvPeM!=q(}$F%X7?eWylV+}Z#
zHQ{KL!%-O=)rMoYhtoV$H35m1zj8n#2V~oT-1dOlr-&w`RtQO%L4xRAVKNDRnd9MC
z>lRACJ<8^}t%*^8H8GkztXv-EvIWSq?E$Ub-8Yf70$#a1!fa+d)-;}WZLEBIteZCB
z4Y;nA5TfR5p21a*fh)F$yJ<t(gex|}RX&3&9|KqZMuB-?1J#7w3RPv+nOOVBU{-C+
z__4qozZO{Ho~%b-kFH>ejW4oD%s)6?g&Y3fAMWWZhx8+Vabp>Jw_J>2Ycfx`Nm2q~
zlQIyZG+Rv^!9q&Xt3kR)%G!{=Lcotr;`6NdL3F*c$Em!WT56r9b`h7=5Qx<#ISIsD
z{Egb=t{`RlMV(l#(=Klala(;GL72>ib%9pC<fBZDPdPKNQL1BRWmY9N%w!n=<vQ3@
zCYEkJ8)<TXHXUeWWqheR*~FQwV`gPZxyh0uS48JYUx5p5OFEWrQ*~jYGPkIkTXg6o
zDhrFcg+*Q>iEx>C;WkB$sEp0s23>h>ZCTS*<Y_0=CSkmC3Ez|tDw8ET{KG5%@Vb9g
z#-t;#DodNLSwS2Z9a&sS>&6@><{J`{+dr`^O(IHv=6KSD`uF5uI@~>53}%ZR5;t1p
z$*b$>bTNTLOP}xHD~)Eaz+N7EkKm*l$YN}i9SnM6+?PE_BXumJ^v&pQv}j1L=q{QN
z;{ZyHfnAcyo$!fd)}wa9$C6o(-U**gW<6#nd_bA?*q!hxWdm9zwW3-B%<Y6PFf)c8
zdtBCkgV;rmU7&$tsVjTXK$_(}I7{<xTyAGY)s0J_d30_b><~FnG?&P6qP0Yh3XLUm
zP-rXJjR^&oMBR|kQZh$`hLXu&gZukpay9zd-`gQ3^<9r{ma`CsKB2mYRUDr!&z`|h
zz8e#oDUq}wg?_;7UB(Jc!g5)`BrNL)@sp%~&x;|xf%q39a?L(OC7P&8p*dL#Sv3|s
zCL4T%*Ho<&<=4qV$mx0)a`W{r;MMkKE<U4N2NO-C>&l~%bSZs#S)mM`>*z+Xywxv}
zyi$MFuY$UI4Qzzdq;5mc$3eRE9a2Dj)g>K_rrKH<M5Qx6HAR9dY0_eFIgYh8E+933
zB^tmg;!1_MZ{_7y)u*j$iejD;MWnJV7~fQLyP_}(zz>Hb1DWn>W1OtPiC-AU>1Gp-
z)@(Q)0FhuFuM?gR;1Dsr>BSv=`RQ)89nV(A;B5T0FD;xm#W5TY=!bc++PlAls<StY
z?-8r*=>Gs>_bjyvWo~41baG{3Z3<<VjXMGemP!I$QZP3)FhesqH9<o+FgZ9uH$ycx
zHA6QxH90geH#9dhGCm+YFgG<YLo+xvK|?n%IXFQ#Lp3%vLpL@xIW#agG&eIcK3xhg
zOl59obZ8(oGBY5PP$4LP9n-r~6j2by@$TaV1QA4Zk%fhY6<iPzR9HYoTm|upcvlqE
zz`)qhGVUq^U%=cl^BFX^z|ug<%)}Qk#q<A?UwzXxbLPzH9y7D;WR}cr=DU%OWa0wJ
zA;BQEdD0Hp2}9J*lXk&wI6-ZW)BuezPVF3N4>Z99wX>vV*b66rshuIUz&<!lZI;vu
zZ7@k~hO{5rf#qYGbO1VlZA6W95DvjLYEz^Fbi#FNESe5O7u=x6%Tp1$;U2Y9q#h`7
z$*zKpr(WoTGW0{$&02M)15n}S<1hvzFbX=gVK@pLY*OOr68EKJz|DyTl0H@6G3?g-
zOYdOOr2Xk|bZTmUW6iDbS-0qPb?oY!8hQ~f!2&G8O{lw-r}WqrxXOmAq|30xrFL}-
z?r^DjS#}%!p~vpR3fzYK@BkjdBX|s}@C2U18mz-Jcn&XE-aJ*uSvi=4NN)A_x@0DU
znG7Z|n8aZIg82)kE||Jt)`D3JCM=k+V1|Mj3MMC*oM2vmLcak+6O2me4`5z`DTyw(
zrA@uO==QQvsoU+{hEk8)$5o}0+n1k8y>8#OmHOOQMwH6#>sw0w?jH(D1MYubDOEf*
z)bd9+&1YI(XWBkxI^JeFCo;vanbJn4{5w<mtH<<j)q&Oqtqocmv^Hp7sCjfX)K&cz
q`VBPqbU&62Jq`WKi%!0=|I8w1ZkORq0v!xAH83(U3MC~)Peuw8<yNKu
delta 43252
zcmV(xK<K}S^9YKX39#!e0yZ;~5lSh4y;@n5+_n*Z*RPnTgj1!(5*H6urCe)iEz6N)
zmv@ucCF_ApVrQ7;G3Fs@?DBt40}XIUYIb+1IC+@?K^)zU?ynEfUu{<Y>g5yP{(pV_
z<ouhMtzr@fELdINuBacaLh2KrhpX#s_09_jU7u0k_q{hsv!=ec$*XpK#-#6m<rV(F
zUcbNo&a4xVIOLIDCln;$@rn^Yz;!N$L5Iyak%F$y4uh`spp_tTBy=~6ZU3d|mqe^a
zoj>W$ftB-bH0cru2+i6+b3{ZO!b>xnaxohpM`ON;&!FF5Q6qvN1`#meq=cz&>b6^-
z@sNA<E&g{~g^g0nB-Ap&=Ur2O-xR8B@hPoKJMOlrOKkn(`)t!U$`0ApZQkW|r80cJ
z+2j8<sccW*=G6v;C;6CSORiub7eAbm{0p$#?o^s5g_Zj~%!&{mIGxU}8_O>6**qv^
zM?@6C8O~^+TrQD#7!yikoC<;~K>#gTp9MmC|5}H!rzI(4ukAM*WzGYC9#A%>*{!L!
zwPWn`2ZYArkOaYWQ!TT+N{c>In<m+9^R#_fc02-24cT-Rd(xUBAm^cD6GU)1oox60
z^vIb7&`tSt>Mlv|l8qvZj%&OC<-vo;6%!c>ZB``qV{6le2856LC)rz2Fj{xJ#|L94
zy5Y1QpFZYUx5Z{+rzQA*SN27h?+QCko7-fWBCqaW!okrnTCJ({J|*R_-i<2iHIv@Q
zXOBP!@;V?#I2k_wpB*rlBw&DEJ`t4R3R<)JR@ozZx65uj_38cZ8&Fw$;D4;oBoIc(
z!{>K~iw}fFv(d*086VgY6@RBCscq+PK8f^uEn_3y_rI6>+q_VJ1nB)^5n1sPyQcvW
zp@obqr7bvO&`32Z>Id$Dp<u5;S@=izg*Stpn(mu5^S!>x3i~#z)4o(ym$;=E=%Mr<
zFiI{Mi3Oh$we7myv-9&ZZ<@MkN!le9srAuq&dVfi>h}C$;V>yk1Z^iC;Xib0pnEar
zcx)7B6bY!yd82-R?ej)u=Z0z6ANYp$?WWH&b4JHN0}P#s$ZZYW(?@6pr%5Avpp`y!
zlt5z4hqDAk1#d1Rz!WRF3vG7PtNQW={D^6zNz;S<(vi>9Pf3%v%8uEp3R|xS-VPDQ
z0OcWrFd|$@@IA4PTFOzhaUZoT42T~^6KXN|7?zE|fJ4oH4{XWp5R5}63q_J}z~b>j
zKOxa+QdM;~@VHY=vB#(FT6+5(a0dr#xyC<jd*JgpW!?{kN}M1z`I-gZ2MuSo=XG+u
z4gukQS_1cD8mVB9p#7kQ4m?Iw8++1-y3MQ3!mstk4uOvt8D)g~2W14TIXkAa+5}@a
z%DHj`eX9_EzE`PKwQbVu*8!mBBWNxpq47bb`J8bEq>)%kHs+i`^cY&se^j@izclhx
z1=~LAJM{cvH}Qx8MI$VfEC3v}Dr`2z4T0sryR8fDy#nyJ=GnSA>48_N(5)3J+i1XR
zjShw0KTv@J07bMxmEp5?$4%Rv0a6Z%Rp4q{m1OpRAoZS&-ONT;y4?(fnQbEPqS6cO
z7pdZyQ#hc<gv7^EqLPG8kJD|Pr-m?Lu|;@k-VG`$y|&vIc0k_Zf0CwZ*CjWuA0r6q
z6*V&MdW5&Hd)v|2v1whkdE0@M4flcfs+%qjc3$Rfn-8mZHLecQ?;WZ3@B~P>>J|xZ
z%sdr;!6TlQj$N(v!4dg3+-vjSGzlWm^xf|fAu1#Sva#vhGAZuKD-Ih5V0CmlPkY}O
zH{I2M9L67+M{C*8DnPiXf5PdwnCqM$cyH^@k-ts4$>{K4I@sU3U_w7+wJ|7YAvi-L
zt^0+CFlLI1a4C{@Xs#x1ZnAb3w(cNCbm4A)nRjFv-b9}h69F-SI1+maaGk;z$j3YS
zUW=ANRBr?4Dg)HxvG)`e{X24DgipP{;J`QaZ<7IU$a6bDA2f|)i3hC5UatdyPYA^U
zNWklz-GFn;@KsW5>L%~DCToUg13J0^rdH{rgDG{Do171No8H8C&_Xi1CNakgTZ>SC
zc(1A@CpnP-74Z_`1uzu?9y9eN(I*yJHe1bP-_#%%7BKGxxf))Fkyk?=)ctmcm}$DL
zgEQ}$NB(X{zgFn(A|p=)>9CgPQ#3qE!{f9uh=X_>OzFGGJ|xbSL^e~3ytk@vG#QD(
z{upEjvM&ehjJzM3wE$ezzSjDL$h+Eq1NxPArEoxjqqGQ!IK+$;BNT{;c8gNg$r4oh
z4<3{?Ib(#jisi?@Fc#>zUhAswJx|J;CeJqNDVo}wi)Xg+>KYqAG{E&T<CAoW{_qg1
zY{vK)7$1ux?`wbt;e7<V5<c1$1H*em@a<phAf+HEvTtA5cN3d?L`5XTSka7sglmI8
zhJ^YKTgIWrvG?-w?d8?0t7q7I`Ds^x!5V#l#~)Q<FgwM^iB7@XH+80p7T<l$P0SB1
z8zG*2`eQ<=oQJM6Mq?6CJbjJ_^YXG%BSrVUx0n6o`RY~K7o)Hib(|biZqC8UNq6LL
z6(YHr_4As7X4TEzGt{x`%h%t3U!7m=w5(u%NKUm5{u4$~50Zgo^O$zQF$GK>)Qc}s
zFK{?PB&~Y!_XtJbCT**V5|Jqc#J(h^8F*Lewo*l5pDwjh1>7hcO#3=%?vi}+j?F+`
zkAAVQlG3E(`n}N}_=iNMsZd?u39$|ujY@SCft?*LfQWA5x4PL6<N5}Fy!j#zJ>i~8
z@sQwVK{$_NViq-#4sf(<4Q?`SX@>?xojO%BIl;zgK6I9Kc2FXC&h39=xP!(v-|gx&
zDR)Ng^c%zPW|;B(f|FyWJED+-PUw7dLgjZ_3Bj#YZeb&O_&3G~5l-Evxrv8d4gTob
zg2zM#GYL?6#0tifL`1lMl+#ICHA1Zy&m<G^8ASVR5!%lvkc%_6=UbKDX;@l@-@(br
zT1bLe*NY+oh%g3Fo%j!%8(6T38^FXg(xdZCN=%ae1`@H^_}Rjy@}u>zbp;1lurG?e
z+t=1y@~K+~h*}gQO_-!ITu|KOfZL4O@M>E-tP~QHB8V-Ls{LqxYD2njs|@qd{cEpU
zO)rM~-1}w?(lPwNOO>QsH*9@IHEMWf^iy0X-6qw>E@X28>1FEH#M^y=I)O|!8_VsF
zFF(5p8x{H?q>_S9*?D)Zt^fNM-@iXmAcDo_zH71yCSOl*M6<C&ftuAjL;t&<*RhT$
zr<<6U(tGMSA#&h<#WY9r-_}Q_3eSqFUUa=Pp`6NTs-bt@nEj7hPB2yr$o@gytsLK;
zUwlrY<erIwx$&ytOnQ8>c{qmyZy;MoU1z#51NaAtQ-*Ib=hhf(irseZkt@hCX^>L@
zf`}=3|L#GxbiE`U4Ct`@UYAc`W4k`;^3lrUu!*1{q2Ww_22XJTm75DU)4ti`zrzT7
z33&2RjhCk~UYw%3Y&6;F+GUg5Ql&r2rbjn4dn*K;z3!B4)~>6mwV(TtXG4Y1UiUx<
zu5R7!&Eaw<Q*GMhJDb(%Q0g1KfLm{Luk&mXz6bu%i=3G_L`aM~(HA(j0uOPUoU5qx
zeeXq{4g5QQHU@>8+*_rpYF%$Vdz;p7!cghVk`5d}*M4<BI%u%~I#3|v7(D1GR@r5*
z#z*#uCq6^=2a(YQBBS~~^gfA=U4!5x3IbDO_Z^-XS#lnuEOld$rGCIWA7~7+G`;og
zEy_~2@B$qgQ(4}NmNE*CqNQHq`hn$#t@S?kU9%{EyP1fA1`ihIr6ge-Pp*g1K2>*G
zHQX*RoG<^g&iDUo5hG{gY~eFMa;l1pCDmsJ31H}=p+meTyv+Ipd*>r`K434xhk7qR
z1U9y;Tv>-zSA`wgw`w-$fS>5zU}<yjh;MC}Ck705K@|p-(v?cJ+rm<*U3-a!j+v&x
zqviU4a4ay!NaVEp`8a~*V_(8tW%eLKZY#6S<(j*XQO%T(L}Y8G%+*XyIQ`1H>;2%O
z(K`w1Jnn8dKn`A~P@P?++vcr{ARyA{Q#`XJA{(<!gWyyLY>oP+sMEWd2{;|`y*n)i
zrFJv3c|Oe#CZ2k9uLDcophgWhiqPdQKYUkzrWvn1sO`Dct=R_^r(DC$d~tXZ@NmL~
z1ZPZ%U>15PIR-qF2muQVy3u!AeQE#LcNKP@fw<@t;$%z${h`_z79sIXKD%Z&HX(<L
z5wImVW*SN&sjX?fsAmdYCb?DKv`F%@b-kHYQ1^D2kleJU67E9L%&9NL$yuKsV%H>p
zO(9b5R51jY5Z!X+X4(ZAn#y0-`UY)@-cQ(~!GnP(3=P@ro1(M^igWywV*tvL5wvN&
zQ(v8!!ugcMV)z8`o{a5&jSV5o4-X$hg#L+IL7{b~@3DD&P>Y+-)#5&viSh5v=RUKv
zH=FH!(YL|Jy16@1K+K0MArN0GAQnh}a=5CeAzfrO`kwzFo^*kD@;2?;K{1Al6v#Ac
zeKDK6vTlX-wYBs>Y8t1;fY;Wk4w~%Hyxf8D{AqHHLlR2*rE=vmG~Wgef@Mn;ZzS04
zbHR>7afe0>Y%#J|n%^ec{66TKi!Gy|?1gkyDXi9DLv{EGmL3OzFB1@{s8OkZXHuqu
zXEbG9$En&LKS?ml;We&wQiqnyC55?m*F~&$s2(kWq%m(a<Vzt5I4w^-cztkR4mdt1
zkZ~Q(T`cjDBR9m3y2w|zGeb(;2bmb7kH6ok8`b5hvyMi!ebFs;qoMqB-ROOQ&~`(0
zVgO5B2z6mlH#O^|ZToU}1V@K|u;D%doT>#np$md9Hh&%idK{z}{1?{0!0SvBwY1jd
z6(?25K<eB7m9QiwQhu=y;^wC3Agq(ZN=CtP3H_CBOrj{lxl?7myngZ@z8~J}3T19&
zb98cLVQmU!Ze(v_Y6>wkH6Sn`Z(?c+JUj|7Ol59obZ8(kIWaIGARr)<6^$f+ty#&A
z+c*%t=PPs@8%Q%;wVZ+knG7-*AcLTDN^($3bej`dGDK;5`tPU6qBi#u%gMzmQYvos
z`0A0S(QQPdKVHz2=iBQSuij^Al#wjvadf?lm|#RQ9wm$tDw62BjJ|9H*?znJbM%V7
zTq#O7A9mX-uIT2z>CtHO8V$XFWoq-^QMcxx>&_Cg^ndl9(KJ#dOA_Dbl@vrs1@C-s
zEJ+uiPi~8ms*J>`jINj@X(rM2D^)ayAGRDXBa{j-X~ZHH3zB3kiZex+6w&no{N9-o
zMlQ237ck70ZR(=xOO54jAB>_Y^-DC%yvu``*S>*tV`|&AebJeww)m=l*1FW8`L4ld
ztB-9{_Ju~LvGGH0YP6;$yn1%+SeRCt5*jDVS+OKL@m(xnqDaEs0#cIE3>VPFE3ZO0
zwe73UV!^GfHZaf@%O->xhYp7AkH<>S{SSEu!yigbaLhx4IX*sr`pd=ZwI3)>ro9cb
z`s9bqXG7*OjFqJWQodY&Z&|we`0nGkCjfJTGa+IUOFjZErDp<-fpx_=tP`7ncJnCE
z_$kg3Mub@4ERg|cVA!KB%<kS5)F!Vgti@GzMhl2A^(~f4-Q}jTSnv@cx5ObZ<sOVf
zla=XgsNQeY<{n^l%-j5+At3mIZT8KU%gx=?AK>kN71ZVvV=4Q80n|jS5Ncrs?{piG
zQ|ue7>lG*f5C|zQ1pyO)dzwxl;H%i?1O&ySAYj=k3S>eQ=L@!p2ewHZA2e-;mAY;H
zUIFtaZ!LZ>4>r8GgPn*LY*Ii_hhHJR!~>tij*-jefG@4qq1$}d>Ci*l*8yIx=Pi?$
zaL#;gEESh<P9TMU0}ze@8VTG2>BFwTMJ6G2$%}E3P{PCv7jiW@@Cg^?LtMa=)8rHv
z9G;BLaiPSJQ3bH<kb=}e3Q*fMfn&ucKQw*aq2288yP`Q9`!0NoiOD&Q-arRS2f$F%
zw}rM?-CG@e=BOJy*8pQaIhLxmz-WTN_~Y--p0#}10qE9$Lr#Iku@1_TZeA|4Oi?v`
zi6&%P$8G<bQnxF>1VAP-yA(_uaKR@mvNS#iOhE@NB#k^zd*uV0_7?Z^D8r2H#+uw(
zy~yZ8aOtRPz#Mdq+mKLO`8{~DzI~g!W?!|s()rY7t+!nHtTOhtR{f>PFIImKRxHfn
zZp+++z7OkvrtR~~Hmr^*Pu9LLojy@RMXU3!Y3;eXM!`-=<BN3-1FR;~3Z_59TsKdx
z;JXJCcgzh!_hjysDRgZ^Vc5gtTU^VtH(>&T?P5xYq(+k2x@4;}J^xXrKb3n*C9v^l
zd@e+a3+9ZG&s_mdOOS8xu`q5N|0ZlxXAhzq-Q6>P97bHlQgpkTNp!nX(+&-WCq6)Q
zzg|P+^GUI6DVl#6!^fC?Z6HG%Lm)HuGrIkW)O;F90efnp*M__dMrmx(Td&PPL+^D5
zr$sy(K{=cSg_qc|CqwD7GVBSTs^N{N9)A*oLURMEzShBM*P>t(o1<<}3ONU-kwf1o
z^rb(4MEtf@4D$D?U``Wb8DE=|eR_E{(EquDT@uBTi-nn_giAI>U;NBA*5^f^W<;gg
zJpOFB9e}}4@rPO$e+)hw?g`u{7^cZ93ppB$<s#zufDZ?Q&?bm*+knog?y(Hw1*rBb
zr98i>)`fdM6PC%f0XOxoX@ONno*yItFgq}R)()uNIOAWJHHYj6bAj=mOY(3kr^&^K
zdd?V0>2#>i)TL5OohP*heeAM)<5|1(`gxQKqeyD%vM;*fYmnJGZ;O3UA9#cVYu_J+
zi!<Bb`~#=C)tV`bfBlWwX=amg3T}%dY|O%{fH56apn$soj9u@qG+Y(Rst!wD7I?3J
z#dR6f+!}W;foQ6OYlF{Y-gc%i#~ja#P=izph6TX|NZdCl(_*zTyEVIHF*%kcYj;Tw
zdDs+gZ=MNynG&fkr$jCZOJ|h$s}<j4@t2CwbRlh6xWaHosUZgD#t9Racx|J+wt-j2
z#<2m45?7qUh#$Acez%Z8UpjRH+sWO3y}Hfe-c_dlZqd)dz&$?zW=z*R%nVI_(jGJ>
zh^RGzpU;YAt1nF7HhEi42dWRB-cE-l07W=^x?$zjc!=GN@7IBbDTo|b7c(VKK~0Zm
zT2AS&*TWBPz9eUaCy!rzsTpfb*H*CcA;3<^hOhU)u09b;6_nF$@Hpn!q8hJ%zD_s-
z-UK-|Yks&taaUn@f3gNJv-xn!*5bt;Q3Sd)5lDZrA24qm;4ABM?zzZ-gpl!{$e|Eq
z#V2Gu(1kpLG@B{-#`Ojr3juB^5t%@rH(Qt|q{}xB>qi)WY5sdYhlN(gKA*ifoD&i>
zBNXtp3NrTN5xKx&MDT2O%HmmnxYX?0e0VpLGp@oz85X?KUGRWS+w9z46>g7~r5TxK
zY;OE(3oAXuylVvs2nhf?!>2>u3WzNx$LX_@Z3x9lGV;q$968d>_1=V=4XcYu$K9D~
z_-kg@=}>v2CrmzZfPear9<napXX!M~6p@lgSAr^%iU78y-1+_b#s3Wh)$bLP!8%9*
zIg=5MD1X&jTW{mI6@K@x;Nt}7QarqK5iF3M%gzA1#bA;a+s@>{v6YN59?SKm$^QF(
zN48}vmMtZ5(e6V?i8?&|&XtFBNGRciQNjytyhuU^5~MH^1Q}th7G#Bu=z?&BE(#y5
z&?Excp4K8_nsy>$o(6*yg-+mLB6B73;APM!^nVt-6ne0>2CGIvE-GQMq9@nmwLI8Z
zhsBf^)}wb?SnvxG%Y$FkQ8@664n6oKuZ07@l+a^oQdW2%A(MLeq**BM(EA`f%gFcx
zdNA{#x1I$7a!46K@4?id16)vrQUvrq^&<MHlJL+`1f+;Qp^4BEK^;(v(awtq!A!!t
z7k|dFMo1WsRS4rjp-<>7^)%)MNUsAFp_PCDCbG6bpIA$vPgaQtMJCWCC`4fqD8g#s
zNJdzzv`8?IGqDH7TOS-OAZ*~pxOZqk7x0n5(?-@gK-kC}9aIYol49KgDuJCNe1Z9(
zWzs!Xr?Y`SIpMs*Y+^-&kl-FJCH@4ZP=Dl4(5cqEF8YAaB)oxQokNd?U{r6C?-5(U
zd1iu70mxW`J~B)I3X0j1heaSAK{>&KErba+Fs%{6v4d(D%~_u0@DNjvAVu)(fKY4j
zC2<rXz=(v=!0tpaY6wstLDeC$R~oDt10LYyArX&2(F!OD{u3c{@BCb3mm>S6xPK}{
z_C~y#=Yz$lm`FMkKmF8u^~D5_zwkvVb;-?=n;UMvb2H**UNz-w!Oeu5FUqQF<aT#t
z%1ys|%DI_UY&KMjYVu~}uNC#On(~30j}^iVspe+;q|f$cUEWmU9u&C5@CP^Ja_ts2
zdY{dw3iG1EJgUb2U1?%`TxIhnG=IAUw-w{<xB1yY3pXJBrJ6UXTV)^LZ3S{xIYYlf
zxTv6%jP7=4ZinY}os<?G)TeKK>4+p|Uy}C~{u0!%0Z+O5kIf34)!}Uj#&xTePvfHw
zVVeBvS>Wc?q(NOV!)GF_fjk+^QeR)|ihLHrH8#7C>vhF_QTMvX9>dLFntu&^S5F~%
zGyRtT&7-i|Bi*vOt%uGwoCZgua%&FTWQ->Xy#aM!pY*kE!F`kEd&A|(L;2c*xgnjr
ztDqGX!%6+MlRC@+)`u<G!mz=qt`3^|uy49!xF*ec^Yq!z@1yuwSI2g<ogK-`<)k;2
z5KrAzw)*u={`*@U!(D?e_J38h8TqxEd{trieIG-AI?X4;(Sy9$U;m;~#Izoj+qCwu
z$MG{tY9&62?A>T_%`DH)d)d_=(_Ca9`*(RSdtFQx`D8IiX~%@U>{C83ma{=V=X$yv
z`7Ixg`ac&B;xm(Yl$jyv>mC@)FjL?JQj%RvCI$F>=IWi7yzbSOD}O)tvY(6DFrSs2
z=sJ6!eaK$p1x4UBFFja@&)P<*P>*UACCYu&=uRrs%$kxk<-(WCn+14gzm6u~vWxTc
za)s=ov|V<Y{q55S{(JRxv6%jtW%u{@a*{7+qv>1@2H6=<t%2>R=l6g7v#hp*3o@eo
zoh--W>w|PCTv7VA(SKN%WJKNHN~e-UHQz?2J=`+HRyv#&B+fx~x_2<MVBdI$(+my(
zE1y7BkJ@_m?6`DaKbI2v%?p>_8!zGtJ+7a*#!s8)fU3NF#)8?$SuwcGVM}<~n|C6+
z${!Yw{o#@Or9JnH@p<kSSJK6mU~wf=^hbPh?N6d~sAd->ynm?5MXig{@!?9L@PIZg
zT-0G*oK>X?0dyDQc+q+q7#<OUGkFCgcxm|n*8zq(A&eAc!pUy~!znBI_}K-9Jsk|A
zgX9x0a)cy=urWp^7{vmX^1eF^#8gb((n6gs;`|Lq2-Hevf|bs<lR=ZZ1D*B?ns)7U
z1eGA1+vq4NNq?!GACB<Sm~b%F?4*}kO4-dpwY;-UHP1~W;c*&C?wM(1g(IzSCi|j2
zjZBa}XyJ^4e<0<<{)R8iEXnZX9qs#QHY;Y>%@+NMEM|Av_x@m3%(HQSkw0Vu=wddy
z$tL~9&8h>)d@iTM+q2V!e3H&;p{<h8oKtXn!WQj}G=F3oL+$--QIts5wbf~29*Gz6
zj#@!W&$pt1q~Uu&a>xa~1&DqW^~raO`dvDWPzihmm%LQk9;GcqFhohQl?>4bwk_;5
z&RJF<-9?DrfijXz-gJr9gT=sZ<i0#wnf0S^<(tT7#VcQtG8DccUG;X|uLR-i(N%9w
zSqe~ukADP#q%RZSy(TN8HGB~(gH4SNxai-Ek3~U;5}Be5*l{hk8--O(GE&177EVao
zJuRWk*EMpLO{U-OvYVp#2AuPgWekl-<)VN~r4aZAUrQuwE?In}|CXLY!ajzeGw^@y
ztI9-MCv|C^rynKq@3*6TJS;Ue8Z7T`v?7%ai+}scxabe(*@w4p-~NCA_~CMz5B_fS
zvAo-1<4^$>wv_~kn3V*EcV7Kxkia58aZ1+kQDhA5l(a06!ItsA!2hPyYPU>ZBh(R5
z$3I5F6wV!|<;;c+3AvyHAs;7nitU63@tz(p-JHR<2eKoFHvajKnlk&r^WW~qU<0Px
z7k}k{N_+XA!ZY(ft@FI5b-&5?{zavmL5Y_5cQaU1`?x*bzzs~I6Blo@{9nt_EFWff
zv;Oq!XfVga?cH)T%uka?Hd1Wzh^HV?jQ-iAK)Gie|I8bZ$5&F>`2S&xr`OUtt!mgr
zNEalm@u*?>r+<xf<GNoaaK@n6F|pMNn_n-q_K2J}GMLm($caifrM}06NZqJh+UdB=
z;(vK)C-fwt!48BrO4Lgo2n`N5t_wz@(r~SWs1bwez!3F%@UhE9UcE*?RLH-)UN>$p
zUmNq3e7z&Bu`RCpA3$liM}xpgx4=mOd{YECFgP=pK%D^=135A{lTjfjf2_A-bY*R~
zCLFV3d&hRgw#|yQW81dviYm6viYm5k+y3f#-#&eMoUi*&_xZ8MSog#=am{<KwaJMU
zRp^CG?2LfocDBy+%nVH20C@`=BNr!SI~#dBE?IhIpqUHcubd5zoLs~aXy|NVXDe#x
z4CDrA08Id*Kw|(43xJu6e~Sx_93Wz6@8M`+X6_82Qd8ETrlqC(Psu+f03(n8(ERmu
zvM{p+Q2c#x1zOwL+W>8y|3ds<JE{PI0B3U`z|_JT2oO<F{2?VT0icqQR|7}@ZGn!4
z)&NBpBWnv|fUJcv(AEh^4KTHH1X%z305G<*HL>`oG$)3?SVB$!e?x$iJ<!<VuN%<a
z7-;{GiVk28bhNQ>a{BuWuy6vHIU3qJ|LuXZ9l*lY*xJS9p9KEOP3``bkiDbbUkjVR
zy1!tGc23St#*P;D&VavB6-CAWou{+8q4Phnoh<&U0d}T;Eluo<UH&QQU%J0Af3?nr
z7Pd|RXP~?DKe3E}e*hB;Cwpr{kH4}1g4sJ-{3{I?CktD%|IC06;0QD`bTqLBIywD?
z`HTHecmA_Z!2eRcp}oDe$G^Po{$=$)bFgrB0$Q6gz%jG@jcM%sH@2CDEga)Nvm#|{
zY6oCu`nTM~#r{8Zu0Y3sHHhk;nW6rhgrSL@t+fZh1ZWD!e<*L~{5K?k>i>>qhW|d2
z{~IX&Zz1@<h2H-+a{ntw|H~!*f4$HD6<XZI+FIVw=I;Rfd&dC&E*V2xz~4OskOlm+
zXsit#|9?J)HWt<%|0l2iYOMkMce?)vFezulzikn+HTz2q69dz~g%(cY7Vbb3MGI$R
zbAYL#_1~WTe@m`rYXWq%wy*{M<@R4~1<*4yG5r^vs=0-+mF+(yu>D&Fv^Dv!75t^^
zU->f%OR3092-E&=7VN(a75@%`v#N*v-`xI3HF9<)|MTz<jIgksJHU&cnVl0r&%(+4
z_x}A&h>PXB&;KUmzkrzk^DJlR>}cT*(EeLD6Z5~SfBzr9|2*mb7Z@>HV>^?7Mn}cj
z(AMPdwEa)PKWbwaN5{V;|7!vKt@=OC|C)Os&>d(Dx3Xww%oA*xo|WNTi<LVxE~@>z
zml=F0#J;3KwXndy+OEehJ$#$1!r;9ogrRYg`?Jks;ne<PUz+CNPcNo5Wlt*bG9TZE
zl)A3~e`SNBgM;=UkWsG$A>&LV+Usgwb_+s_ok{&*XIx2HulOB`V6ubN5$5WN+Mm?5
zz#pCL(H_y)IITVoqZ7Fi1hw|p=@~`R@zW#az{v2>V9yO?-##8~P8=*bk0F+C$|-TO
zvxkA>eS`5Q<f|)(Q{_T-i$*m{lA4<*-rJ%Je-YV3>^2&6#sw>(*_ROL9+TjrQJjq1
zWmqVDEK6Y`KONDi$dszIUR~#+)Y&ro>Uljsce+*9GFPkx{9$9%qC`4rA=~CBiaM2`
z<iLEER%n8Wq&gEPa)#0m^(ek6j_-B20_0-+h*5&I{v+kdJAy?qdL7O_gfM*FAXTdA
zfBkho&lKs0_$<KbW0()!BJUk|nFIxYSccM^ML$BnDwlna71-s)1xwv%#1t-{u&kUN
zT7iTXNN<YmLaA|+x!faXeZ>hwk=(FpB;t;i$|b+^zDv<->Cx`$P8VYfM3@sWm^%91
zC{w%x<IynmL$dn<q*Ml$KUO}^_j`)gf7u7(yiv_yV<Vk*)UnQX1zvW$aB5ahB>QeM
z4?z~-@#;I~E`_UrT7K#%La2MB72;g%U@fZj2<U@8qZyO+iFCClY(0RdLn{hf+qC{U
zKT^yFEY4_o$4~ZsVjWuK4NTn+EV;sMJ+I9^ih|sJ57yh&HESV0FSiPVbgzqWf8it8
zrMJl|1r+PYOp%U#Lx%nOlq*#BT57y`p^?=e<7xtpR0vQs)d~JCr6y^-GWT3pcU7y+
zN7-zpF*~3vxkfaPj5lphm4pdth3252WPOqBPthti8)Tb+&{F*n_c|wa<VRQcAbWJ?
zhXO@bArM7CGuOQ)vMAD>&_$ufe{&TH8e}A^1*nNKwtjHVu96@Wm`R@esxL%q&BbG)
z0)1Z9i|GF4cUx8`eOq@6%<ntVLLRqqYi9g?mb9?^=Sqc!$V^{5^L_bZw)|C)c%UQl
zx;ZDZAv$+>tp}oqVYj<wxyzwi^YU%fB8YRDz3F<=I-Crc<@C1wJY04se<sXi>zaO>
zuyf01Cs&(Nzk@bJKUiyJ3Qs340ve9yo_c?(Ynh-{GQGzJbcYj{fnzIbWPwe78tTM)
zlNgrxIz4bQXjWr_h!kzUmq;KM)5dR8AeBFqc{-}rCeUDePEuD|ZWU(x*s)DiBvrAf
zF>|o<6kM;je+a(Dy}C)9e`PfhOn_43^*XU)Ft4qj`qay2f`g4jSmM)7<KQkwu};^r
z(Xv@5rTeZJasZ}i_v%cB3V*S~-{a>ZbGojZ;Xh5H9JNC2#<u!Qz4TsK=+LOQYc>8_
zJFOz;22<Z2Av|)~wgo(;6kP!v>r;Hv^D7uDF6JJU7lwpoZD8#WfB4yM!gtAaWl8i2
zIsz{s`3=%{1wsyF6r84@tZB1#AC<&Bpi`i+6bPG}k#n@E@c^wFDA1+Wne@Ufe@Hg^
zzvge1&+V0cIW7;JshqqH0~({l218yd$K4&bPR`ll2+8A+zNRfbEZi9BAZL({jEy$_
zfR}EUOz-OP=?68&e=*E2YlQagMyp76%NvEgb$K<sduv{2%=F^#67jij!c62j?G|jm
z_>@rC1lhXnCkFT0@Zt87*6aw(oMpc`Vps)0f!gv({a#op`m=+N=ue=S;i`2&Xz@^H
zEilJWj~F7G?sl?|N?`t`szg??;q>Q;IfJ7N{#~Q4XcQM^f6!1ogy&K`3kDUr=L_#6
z$se%z{Ooj;Hw66*C!0LFr3_!X=o+L--{-x1$pp2n9?+B5d4X|DJv;-u3?JoK%&sM!
zELK7^@U2OF+3w9o-Sf(<4i(jwW{O=H!b`Q&by^3N2}AwChNNdL12vpsY3!rm2KEZc
z_4QhXO=*KJf0}k<EK)y-G#!4`f;9BjxI~A@c9Fa-d!&C`08Ie)%pS6cLK5GQAhV?U
z`;E`Apff%188&ayv>E@_+OJM=4CF=;Z(|l}-t#Fm&Mk1+X&D*5Z~Bg~YZaSMX&S+B
z<*ZCBPR^0HH7hXM@@!*V+Ey>6k#h>mu5Qk%^AzHTe=-{+a93CmDen<+6}nMa^JK&)
z7)YI*E=kyVCw2pYA5E4)cBx9bVUvCokW)#-R}L6OwV_*25?fMN`YXJsfAo7za{A1z
zg>acZXUa@}#OdXSV~`$!j~-uFzyDl-X#dRRK3rC(^}Sbb((R4R4O-QNmkhTkA_Vs4
z^Lx~Ue@j9}k*~d7csZ>WE`VA23>O~q`Y%c*#NnTX`3kOXG^Jzd;=-F+r}kW5tEHOW
zScJi|5LajK92Ic5NMVwVqY*)Q9n*UHUnp*eGr&%m?60TC!!I@ZJsWO+ZD7HU9~xXk
zs1%3%q7L<ON=VsX#aaQi*joqVg?GOI`vB{*f3ja#EN&1P!akTbAcnF+Mag^w&Hm#Z
z<Jr#0ZSyU1d|1FIirW#HKW3Q_o*GHQY+U22)T(<EYD~F2VkjyVruIk*hWN6yevm8{
zXZ-|CF!h?HO<cK^rB|lBdl81LKc?pnC!)SC0<Xz-$nl{s-}J%Ya+CktarMni{9AzU
zfA55b&#(#Rh#uFIFoh<SEK}-Ch>8n3PRl@X3mall%biM*1w_wz@S^aB{aM1|Zi<B)
zTG6l?F^$%yKY@G_7*<yFxcYEL1UUdTK03LLlF<NfgDuuY&WLk2v1dbrE4V)WDud~3
zWbxr&-!^ZYT2W<{hM6epT9)l_g39o1f9i*zUrP)=X|eSvF(LLZ4&&bVKZ8Qb%K$6*
zyX7-hJY+*KeaLo*J0B|8%I`Zzgo7MDhCLvPudOf6d!35OQfdJGqJ>0o?oE9Aa=x+a
zKe%<6rK)eQ)l3`W_=mZk&#n7y2fXeY^6RLs8?tJ_H+`+y^5|m`UDRNT-PWR$e=j|e
zW^O@f1x@AGS+-V`UIu`bGhagx=i(lL(?mBoiP^)DVbh|T`XItII5aoy9mV2jtf5ED
zBf}B#V##MSr-Q^Cyt?^53KqBlk7#K~e2_HZe4@h9US?{W;i-fq?Zn_3a2=hWjj`WM
zNVc27m(y6Qer#~yq$*l8HQU?|e+IGrkx6mqlZjBBCRh&-7U@-}5f5uQQ8Ybo#d&U&
z=?Mw##!(M#yWJW)&5D>r=AI<WLY_Al!L`5I=96N}os$IH@G7z&m57swqURb;Pd>SQ
zG!Jj1+wej88O7+*yZM!L5>96S8nY47#4l^_x69t~Y}E(4%}z{}rY3<Ue>w;8tOEaQ
z6eMEiMN-a8Ym=_j;JqAEqkzhJCY5RLYa8t3(KZ@-Tm-@X4D_@gcbrb6DT=K+SDP|X
z{TP&`fQ3HMWzp=~X^F4J$}WO>kYVQIRX%f9S?;ivI3+h~+93~;sB)v5-v#rqS{Jq2
zoXM3jJ&;f3wn1sg)*5XJe=9*JIm5)Q#JC?Tj<bV31YH_I3X1_YJf+8L4rVwLJHn4%
zPDwqq;4QM9()@}+bV2qS#%}YlWvp5}1Z%9ywb({CX`;`DPeYovfE)BRc{gxG!5Z~S
zW{{3F_xwHr1<a>sANcdw1o{Xr#Iu(Aa9D{H=e2Az=|FdoFb!Ctf87zQ!GRulMIF;_
zUFQ;`;#~<+lO}N^L>~;v%mom5rzf=wjP?x`bE4G$5jCG#CGkL=sS>Q$@^&5H9AsbQ
zLR;f64%kT~LfAuqL(hjX2!I`O!!D@PkFf`oY*S^PF9ysVPyZYwe~Biz!qZ&TtiClk
z$Hb#giMWgo`u>n+e?1V9QI}-qk5`mqiZobXx0|YjPxm7y5nue+Zkw<!sR`;}cT<64
zChZ<6=bM!@3>Xkwrl*YuUD6=cMh^$Jvgho3MYq8rO<nOyorKYmCaLCGFI$v5fAg1q
z)U5#&rUdNls7T+$I=ROOX+$U&wq;`M4V}_3VFiToZ%)+{f6Zw<)}KOV?V2WSDzC(Q
zTmJdY%3JLpl74J@3|#Y{uU(kkkH_o5VJ6$>5>oYfXd#LEk05&^hNIjQ-=M&5<X>vE
zDD~FSgot(+vkSSXUND4})rWPd6>m6C(8BG^H6Z!nqwk^e^}|CgdR$mtWTfa<of#;d
z%R(xYbKSd~f9CuL9g}2m55;e>yhV-Aq?zQy5OcDAR0?&;lN-XGN^6ze8HdNxiEAxM
zWxJpHOyJe)>D}*I{r!Hirn8m>&SCmhB8g@=Us+RbMYZpK$_#@qSboDhU{bJWBA|>w
z1Vol@TU^cq<QT<{;yw%2Z1)Pz$B#XkriruUbde$)e{t@FJ3)Q5JcM*gFSRcemm3uY
ze5h1sd1r6?-|#eDu^rB7Ra7ddO75Dz3C}oJQxJ4B3~!4SSVZgthF+s8bHCZ6yQyru
zwlv$bUy`AiL9ZZ`axq-~5bHFF$=`kBhVUTZx?0J;yx%Kh?%FJ!adq>a)R+ca%9>Kw
zymoulf8H3T?ecus7Z6{wI+wjOgFt6$T4qO9zAh=N)rKLr;n%n&&YL9>Q}xN~g^+lw
z?f;dpS06wr2{az|!{gMDE_~+DCkqHvKq4Mh8?4YEaPwDbK@fmN&Qs>~y$R!+B!Prr
zofvKQ!mnKC@pShRbll1?*I2rFx|@(Kqq9V;e^M+FL@COPab#DhL8?{)fs|0n#)cqa
zYFhE*de@#=o087+PCu&}a+mfdb%<Hw3)isHdMEDR|LF5pHpr?GHd_euvb{_SEkU1^
zd7Xz=H$ny3A1zFvZguQsqf^DL^M(P(a%ueqjkj~YC(Ub<VKFWk=Jb1ZBgVFEwSMWA
ze+pA7l0Tc>l2?TYyLc03-}7X*t#hy9(7L<q`|}Zz_;NWofHnV;3P&By2%K(+QdWcU
z^S8_|cVr)>GOLJ*t<`D>U}O9_9j%D|Y&_*)ppO0^Kti%1tJ`%>wOJU;$9)lmgX5-|
z3nudmDlYW?8gqD?Q90K?1T$=%OT$Z=e-<=q!<+)0(yaJFO<`rm9^>2csG`h=;?8w^
z%gK;rT*No9Z2XfHPCb_V@4{zAXT-Libmj<;KOcjb5JwsxQ%13w3ES5iyUYgUyGpaz
zSg9FDis67_=@Y+61&h&N?Pv2?PbFimUVc<#kxTK$`(z6&;BLizr>EqDa35Fse+Jw}
z4|X|ckdg@Vt-C8w?|;&Q!W=(GcrAl&vmS*1{)e6P9gzpTLwf2k8IC4$9shSVaW`#-
zlOFnlwDmIQ0U_rRgRU)i50=TvxTUaSuwH3KvBeW&rxd6@)p_@;g54`+GK%1t=_jKC
zRBg1rXT^gcGr)X_)I?EKFK`$hf4UOE1|yiy2A|+?unf$N5+d)e3M8$mPyIJNNzw0H
zpC--7pAa1(buyJf*+MC6HPOKsFrd_!A1L_gNpG^#le5Sz-77pF_Dtm^wFQ%{-)QV*
zKMckuHPcrQ#>S+?l$BG|gqsM-sjZY*FbAdr$TIIXrVU_jaE%bbFibQ+e|%(o-Rhom
zFf4BG7qJuT1k(tXp|FrldczNvgn7j7N^t8A-JypXsVKc6_588E<#0(#eVI_kbY9aC
zgV~M3nyAHazWqTemNm6uDo^$49b)Y()l#Vg^2VUa?!<d*`v|%)kD=BnJHYze%#N8s
zSUsT+yBkAR#r^QwhZy(Jf5H1!zan3eKNcZr*xHLBI&8f=BX70O_Z>ddT0uv6U!LtT
z)6_=7)Jf2DHI227@}>#cawyP<L`@WA=e&xj$OT&6Ho>ulYX<o$t*S>SuwGawOQKze
zKBh=TsXKN4-C8?W;L2c%+*&U|tPgaYx3%t3%{b@;l~<nB3Sy1xe`|Fd<zP`LL3592
z*XSBhDN+f#7PaUAw!^|W9b<{RkX52XRms#M_z_w$@k8!_@sltneVV%L64IFta|ljE
zs0p$|9BgCI3VT{N*HfDm3X;3UIytkhMBFvM1Uj9k*K(`fnJ6;{X=4TZ=RyB%o{n4F
zPf`7T0OU}&gr*7-e^?(vx&OCUb3EOWFI0Kn_;V@9%v<35c;;C6Oulc!qNeE0@VKq-
z!5pfnyNjOl57>1i-JY*MTiNu=XycIssqBfZy^XabuiuvG(!I@df5i;c^32&teF!Du
z_<}~<Xl)bbAJeWOV~7C5Mi8~CN)@zFZVssg9C#+Gm)h&xf6acsck*w?ni2R-+apY+
zg<5k}#<GQ9LyN{gik~1II;kluTH0w^Y!1olZY0uq2ZpMsmZHdN9z3<ehR4O!4;}BL
z7G2dw#N;4@5X9l}LKZv{u|C&~TSHJdsyiZB@3)JDK-es$VM?v1!ibi9+Pg)!IZg^j
z8bV_qnSRv=f7RXi-pCO32%pfQ)?1RKoPg-^#s_ouy@i#k9{2`RlwD#U+uX<_2ZjpV
zD%;(c-p#h*gY_v*uB1Uv$PiqM)V??Kes3l+>`<WKdAX8>+-dJy<;c$n>4O!CD(LEJ
zfFJW-k<52w7KV}dIBzyv6vLcox<*TEGwgtt-#XQif2EITE1n8Kc_bb)!Y;-hc|PsQ
zeB_GMma;Y>a#6B<0WD}zkGs%|ZSt*ngvcNWlpS#C@*6Ouy{yx6fF9ChWUIoXAkCa)
zFRc*fZLvBWLz(OB|H>ksTb^~f{*ogXe>xqa_P~|6`y7%<gB4yacCHCL$=>go{9rJi
zDzenzfBK=X*Fw2Upw!4Jg)t#2RhX<%xhFH!Z-=($vh^dDMv|LGi2DguTWV>J;{GQ{
zo%f1}!h{?~<+Xi`%wgaGCjIcY1v;i7FXcKLcL`Ri{Bf1i_Pg|;Y6`T34pTRrl@|mt
z0vk+oqBaR%8dQ<7Fl9cn#IxRKeRZ^*ZX2a}e_+P)?~Tvo0%ZN)%jsJhjJ!3XNdQ@k
zz4jm*-v`jByf$<`ocRb0WW<ML&Jtds6-Ef`?@Cvrha0%d&qO>n)EGNf>geU|)<q%y
z(vi-&$P1(|GNSY%kSDhOk*Q-NzhoGY%5=~=G_Sceww+n~F5RQ}Qu7rd6iCl-J?ky>
zf2O3zYo2e*_iGzkbN4k=W0(>~e$LK^O|xERy0l#6h=^pFoPz+DGI!pDgNYJC{O?Xh
zW`nB)9+$=>gk64R1>&GaVIj2me7>#)t(iGD;qgQ)6=0`G5RZ0E!zq<5^N|(JK$+`?
zK9zI`^Wl0Es}RyDDN$y9;RQmcmDBR*e-%P7mt@sxC4Ir{xijY7Arq9Rn?B=$b=^t5
z)=}T`hmLd^y(92e3_BcalYg64@u#^iYpV^os9ImO?{AWic>Y$wt@a)N6$d$8**}b5
z9++M}KAX3K8<O0>LjHgb@@mrc+gXI?ht&~*d?WQi0(UqdaOV}bESm7$ANsQ?e^4mv
z-J;ALa`%~6eGmcHFkJt<-INxa-L9EQNHBe>-MhY7wT7V9-!e6K`IoeB&XG}`xN?p4
zx*v#=?yqq;wY>`HY(@KEJ>Kn|8}WPSl+nu(4VC2=#Jop7q++A=085=opeMyi6-s8L
z+BXyMjjnA4*;|nYC0%MlKKKNie>Gu0z0#H76_JyDE+lgi8#d6DOth1x7X2Po83o>k
zjVjRg&RRT8Qi&t@tylxb%@-)WCDG<4;%IRYZ{1zIv@C@uS@9vXEKd;OaBZC%bRAE#
z)O6}tQqozi^2Z&(1t{(U9U6iiuqDA8`%i~fQ%us7`83Sn_5!xs$E~#~e{NFpm&y~m
zs4(u#=2{3zNr39<Ol|e)Tt96Hrks+Cp@E{1jCd<F0`uX!!Siq`UYW7cs2d?)+ghQ8
zFvCzQ;vjmOrLz}=`N?QE1%9L#@06u6sF2zkxJ)=<lk3)urAsEv;%~}+Nu2xp*)Ewg
zr*IIlZ2=nCiFBU0+5A|Hf3tJUR&Is*&g`W$<=)MYlGd(0B8|BZAN)O(dREOhl}}YM
zydhJPSY}qb8MnL#UNR>-lfH<%-`~yBNt2YJ$7^dFXRDrRuGle<i=9;4)ceoddDTR6
zM*Gtc&kT(CsrAL7nTprzOc-6Z$q8ulL8oSnmA;k9eO%!o-f8qIf8}Mtu0|S3zW1%W
zFtKSR2+O^*M8GXi3<|lcE40lJbu7~t>?Pi6FhT24OE3oX8_pZ5J#A=*ew@gACl*t~
z?PbsM#}Nv%9#3&z!tWh%9!mB)HRvM1r$nB#n=Ks-dLSrNh>#RDQ8E+dgNk`$7_R=9
zaK)y*(CLBBLxdLdf9jTegGBZ=3l_GZQQ;X_J;e2(Ov0x0mBarQqx>~>-pTN@#VCtW
zC-ezt4P0}VG}H5<)8u&G6yr_{_zs;O@=ZCj>lo8fn>+RH8ilo_FmyQ%0rfyidJAfN
z+W6Y{vD<!ms~igpZhGQMBH$F@vn-Jtds#&*dF5x6=8}`1e?Y|ksRG|a&w1RK7?!^r
ze>`D(N4{Q_h4%gM%Lx@y_>|n75cG$4roVY$G+m>g0&$zKP<82ATNZ{5pPss{{bY9+
z#cmjaPig&#?vqlf;UXc0gP&VsV-ad+$D(O2!yahl5pl+Q^(USeAqjtk(Gz=evz6|~
zJ|ZlS)Dzzhf7~nyjblE1m;RQb7JJNYg+)@(V|rb;%G_H)i<QL?bI>MIWymc`K$NdZ
zBGb1SxM%#JHdt>&OZQUH$hNMC)yOlmwNR9WY;CyqxsZ`7jug&BvzHxIcLEir%v=L2
zh&rnBM;5GM5QVP%vptueWCC;DZOd{IDFG>J?BRQ9f9ynO%6Ik}hrKydPeq*R^chSI
z!^cx&JcU6k?PxIk>}b_0HALA4*GOm-p@2T+tP0<^yTwG(?4PbIJh-vrmsj@7l@9Yq
zGRUY*?dhmbH&$6G?J6xz*G1u0X1RVPiX(5s7MU0nW8wU(!#`N3pgO|xztLAfW5uI%
zVPOu1f5Jfy<r75F^0f)&J`HOND7;#`S)pY71mPi;Z!Yf(_*sl8g=^V^$VhwvUSyfo
zWea7C=WM$zunZ({6Q_ZUf<9ag;4-(#_<DprF;}`q027mxV425A>c>X+C>i>d3M3xf
z7x)YaYKdwd`OX^q>`j3_ryAK<nf{0`ZxGfnf4&P;8WEB@orCVIiHaNBo@r5t(Q#+j
zfA$H2cnQ#LUL$N+*z=KZs6L6M)1{t<l%E|~GZwr^eJ68ed#hc5Y^!FuN24K<wrd&4
zGG+}VwUZ*GF$_6+LxtqdrnQm5o8+RpplPK8wr9|TE_soA{|u(u&vNu68`Gv<@1%G)
ze|xvVL1NagOg#KCtWidftdH0~lBud2zB0t{Xzw-Bx&CH$(`@ncq_LRzyzG24{1W2x
zq(-1dseUrQ#kQs0IFVu?QJ4-Rp=ueFk(JAl6M1!M66OqMBVN1MWz?x?6xfu*NGn*7
zwj$R+E70Cq%$7FOnK#f*)sG&igaNVzf5{Tkyi80y>W#~UJ`oXf(yC{O?I4^6B-cYK
zvRcFKRwgx4R=hp#Q)(hm6RVnn(mD65oSe$a)8HUn57NfLs+H;IgHoV^4a>jWUA0%K
zSv#(nz3-V7L+{#ykIY=fm@3J8Q1e}_9UL0RmuPD@|K)`@Vk8PbG8*m(!0{)8e+Naa
z#39a0qNPn7IpZu#R<<Yd?NL#MD}LapWTW7R7T@<33e&SG@)ts9tnV?10@hb0)*i2Z
zXdCuQh1%~5Uf7&PwA%^~u37s~g6(uHc%ndGm#0w{o=vTQZ*dGIGT-aM&3x40DELg6
z#KfT&QOFp%HMeW3G?sS0gnCGLe`kOuZ#^#(#`Yjda`^R!fAlen%1-M0;ih+RhdACM
z-t<oloT2{iF8FQd8oS@sH-y@N?<t)qW!sv5go|e26>~#VQGY)^wOr~uJo6iRQ?x#M
z(xo=`kM0x11#;?Z9zV;O<0XwC&ZCkt#(-B$0@R^`sf|WPX`9}VXa0daf7fNkG`+x4
z?<N?^K*1j&Ah(l<IR9CKGjjUi&T%$kD#?dBDO|DziKHgyVO6B`!5KE127rZQNyrV9
z2Nn^E0c2cn8zRF$Q<N73f^HWaa9x|Jn`{8iu*n;5rah&p>HLJ~I3=qil5MPafBi`L
zfb7tXPl`CwYO4>ag2+41f4X?Q+HhRs4aOXCR7I+F^MFtW1AAR3Mf>VES8i-Gq<olK
z7*QKiAO3yZC8E$fv=q_|QqISxBaeW}A-(;c|NYuxi=xQ-Wz%yoZ|u4B6xM>kPZdRl
z)4@&Z2VuDs>SbR~5v(lNawF6!1Bp*EZa9kQs=;lcCp`c{K)t{AQpqc`#ecpeYf?fq
zcgmJO;kqbU{Od7~QZn^ije)Li=s46dm<?Nz1#^Jm;G7a))`k_=A#EpA+4}lqe3*hB
zHxq+%k2@M~-_LZDKxKvi&xA@-Z&BS=L(RvXYifqVa}27|{sH{iKu)#XCt}?J3E{0X
zCE6cBY8`O43hAJs1J?(Qkbf%2wPDNH)i!RQa}o{rc1L;DoeAQBk~a$O;3&OTR}L@i
zk2qZW>_~_bc(vR$h*W4lVUg?DK+IQq8tEXr+5@z8MYFccN?vpbOh&4`Az;^N3)?(l
zXD*`^)gm}b0G^}IoOWZ0TwQLbNT>{eOiW29%5~n&!1>VA7HW?pxqq+BYoS65g0Ohr
zgJ)C?$)eo{DW(GjDZ>lj@UyTHzNcxRxS>(Bd1Zed8GoHT3AvOR=P4q009n=^W%S2%
z9g7)e<hDxFQZM0zh_rAJGPDN3BBq#VGzAF<Ev?Y!Xa=!lQE3*XB;J4AqblWb(kv0X
zJy?{wi%0vci2cA)u78Ork-q^{GQMzYX=p)IYvvBUMO*)e;1#2T9YS!EuRCPk<kD|j
zvRdRVZv9d}u3;>5v6z#^Isu#W5$9Eq^j=Z3m)(&(PSu$OuL}ZWY>Uid%bI*VgQPfE
z^}5D6%R_w}uA2%5#G0J|ZzOAPxXgSOfGs_q(>M5bSI?K{Wq-N{0VkSefYpTPZjr7X
zBP_$^3%W-Y8miKTGxb76JHr5exRY8%0*5;@WW>F9E?g&B_E2{-9f%M1tX+ptTbtm<
z#ytPtrn;(rCi;-T%{GHA4n))!x_hpjpDxttfa>pa25zqhKQPRDa8;__D@`y+qar^<
z1>JgdG{RH@hktQ7T$Jv+@Y$AHHw(W%_eva~ELEP1bQZ8}ue6e?=vk`5(}NR)@h5Y?
zu-dVye8?~ft32}8^}0S6DahWUrH+TKM)NMzt|WozPS?*J9gd1ZxoP||0p`~k#jqwI
z1An5b!<&3+r1M7Lph$;>b%s-7FYVc=?@MdAgi`qm9)Ee6&2OK!eu*>C+iRuY@bi0S
zd{q<1Ms?Af>%Df>s(cMxnXSrQ9{jFJfzK&W#b%itvwFXTbM@dm$4?QU_EyimpOdkn
z=#4V4b~Ua<kyb*XzS~Frn#xi5=3O%xEF|$uMv1iU9!>Bpt-YYUU=c^zGet*J@Mwzk
zE?aCqW`D&DOJm`o@mL#GCZK>K{x)&JsF&u|oDMjEvJ|`{LZ#bW)WnJl^(kT<M<JBQ
zii?M;@E>1E*tQq1LX0#CUTI~3upo^2@R=Q*F`|zcL)`mxKS%J5$s{ZYsxj58Rm{!;
zUQu&QCU&pmov1Oc4ARjmqEc`wLxD<Hyu(^we1GjalV9vuz(zJ}dAeBJWg*~PAh3tc
zh@@Zj?at*~^Y_WC5(G}oX(;+e;WWx#P^;_pQ7GxlqNv#o@>NAHIJ6@Q4S5abV}N46
zyO{n4iLW8P$Gmhgt6}d3RABVmixeGky_P9rnCGUhS^}74<Tu&3eQJ?r2xbswWy6zI
zFn`#$uoez5hr#yv@0A-XkEAW!PCF%kaxdf<kwzK8E@3B*u3St>X>64`;rU(dA#|Zz
zW;_MJy3`=<QYVA1Rnw!R^zo<Houv0F^)5}9tJAlXH{rZsHk=#Om)K{ie)ma!Nd^l<
z9~0fenT^Pm&c-qt4@?K22;{t<$u@1uIe)msJ~*1tE`!-up|7R*Mk7La7s9TJm(%@G
zz;VUQ2YaPDLhcxBDKXClYmCE9>g<dAaji!04-h8YMEIJ%iLAG`FSKqs?}G_N3g=bl
z`j|L&(455=JoQP|9<_AXXZb0Ew`Qo)0>v;NWiNmyWvoH@2ep=;++zEsb$zkhdw;E7
z1Qw|U3Ef(hF+D*!;qoRT+R2dkIu7$s%;0=v`{UEbtH1U576bBo!3rBZ>RvF9zR=(e
zKl{tFOw^BERImZFTCJs$l5RG-_etdy7M-ZB%Oo8j@*{L1TJ9}kxm}f_qa@~%-C-!j
z-~R>TT-gWlV!-C(F89pq?~O}j?0@jhGU`L3ZnIR__@Uf#lAZU3E1-Wye%Bo0MqY~$
zaqvCwlvxbYWI)o$ann51HEr6b_)A`M3|gaqbRH%^1Y2}U+ujg>SoWCMkPno#al9^!
zGikxF4-S0i8cN#&?YmCi9N)+lDQ^uGoId&PJwk~ze-#lOh))`OE%(&rB!AN4C~i?N
z`80%pl_arNc5C8AD>)$cc`L$*OgT4sJuY2)Mo(>I@a$AG>f5Av!Zu(GueoE4rNTRx
z1Y7F2Of&r9P8%}AH+mShNhb1q1v*=F9jPwNe!=EeYRnTtFppdg^T5#BsNEk*(6~Kc
zKf3T4{^gG7vc819%5~n2V}Ih1#DpJeHd!;S@xz=@y-|;krT=%Rtnsip7i(p4mHUf6
zr(kB_sin1mX{0Q;s6nh{nMR!Ja&JoQay}?!+E{6-=?w)b|0zy2L8ruG_E!mfCmh-S
zas5Tc8DI;KrL9T9KX4;4uw!zQ%wW|jrDO9a55Kjrw${dVATsQO`F|nCWHLrf4jd+l
zsIx6IBNlg`I=2_D-{dzXL6g;lKnnkX&5)2Aj#w)}U0<fFpH1n%kY3G5tC-7u1@H$H
zehVqrEFa}|vgct=eN&J@%QNUr0ow^Ytv#Ba#_c3zM)IMdY7GJ|x20sJtF@&`7WRyl
z(0+%)U}3unS8Jj|Ab%-1;Czqn)GKZugM@fNDF#ej*e>>W=O00cvp)e!>Zj3&`WvW1
zM;wI)tMWNrP>)w3gO#Dc@Ro>WNxoM(s*0~%V3}c;@<u{W2bIKJ{WNO2JItO(CCP-}
zgR1NOBe=QQpsh(Tk60MBp_Yh6ApikIaf0f}EkOkVy(9p(?0;YX0sB2^eIj0qWpte{
zW!DGf%w>ThmkXqO8K~)0l$eHIJ2sSqY`g9EwGiLE?l77;Ux`}#a~}6==9i@g2h8tt
zP!Os?-Jho6gG}YmrOWPkslip<pMs0fmURePm7tE)Il;kbsJlg$4RT{0*59o6N^|6U
z5Douu<K;5-hkw)AAM5LewYI*9_a=IPjW@o21n*c-I~T*Xv_z~{9|@hQ#BvYsMX~aF
zuSt#d<Cf3@`#Y{;H?ri<StfjVy&3pIuV+6+HoN~^?Wa-_^Aytk@IgA@kGj)*?aG(~
z)whWre{pcGn3C@SqhdBs;Zz+t4P7Ov5d75;ZfOlE+<%K-X(-~yHpur_ixx>GP5sL;
z7)`n@O)Z2DB^B&8?$!BLRem4xb8)@cVukpY?*QQ!*Ui?6n|^z~l$Hi1jv%xHEA2J1
z(0o?$TBpw?IgwTh0+P|KYZ)d%-)2~_YS^&~0a+M0@K+=S0iHwE?$HqKJtgkajK?a;
zkI8lEOn)FLbjwxLmp^n0tf^4=apV2^p^~zA_QH2#TyE6|e!rFan*^31*;DVRZQ;ma
zT-Q0(pTw#fG6?;NIfoxMTj6vef_)O{XdJ<OLDVOhqHBcZcw=`4Xgw)83PF;fx<B3B
z*J7IcY&51F>9k$E-(wfsn}8<)RJ7`P!FYn{B!6+ui;YBa2gsO(gn8C~21<`tD9`7b
zwpqTJQDQW|PpZkCa|c+|4ANoKuFn}{08PGz_BgqqNN8H@x^D>jZgCFo2ug+U7D11o
zL%;U5yP>4m)}&jXP1$>^TLRy%_jGOYUCWNFLeqrqVpA+DF1_rRXGumhXxHT?Ewn`n
zX@B776;d#M@Djp`;f#2u-c3whBjl{O5CHTm$hq1A2AmDlq|Tiw?{N?)l<grnO&IR1
zZM~+*t#@e=Q3wYvC2>|<b3V!ONWHMl99IK=gz<?s{-_TdEi$#gHg*>+@UrmUuE4q)
ziSj!4$AU7O5`dk-4a<y#)2M=Xd0|xE$$w^7j^;30dcJsyJ>nbnQlO=jrflNoHt)+?
zzq2ZPA$l5`!YIj(0nSgf{$`)fEx~3Rf!_}z#j+Zh$uFY44MR~sUtg(reivjtQ`}^T
z!~9}n#|fPA>LOS_TudOtRiEEfnnT7+QEVR}@qqBb;59PuTfs=L9Xga`BB*%Kdw=U>
z7#{dt^~f#|^K#pYE*z}pYgI(nS8Y5gmY7z=P)>a;{g}yOHxXF>qbON!!)E>KU9~tG
z2Ts+~8<uvZgIWkf|10qDUH`B`<w?^du9enKVN?WJ75}-QRUv(@Y(1!$Q4v6EUut}<
zSJ-peHBp5xYB}bm4d=kVRnAAg+xL$s|18_J>i>b{YM`_T^lyz>!a)xZ0p%Bug9
zk729u=VHD7G;aS<n4%{)sDEnsXQh<B9EHPOw%;OF4D2S+kk-T!h<h7JZ&9H`D4!&e
zJH_)yCk8>Y{mFW5{v}DMNAA+^Irn^zOh1low{3_ua8`18tS1hG#cutv?0?N7Ub(^_
zcV!xNwhc#f@x4wsN41Q5as#Jn3Fdj(K|bf6Lv?XZs&p%2w+otw4lpq4%I%OX!;lX=
zW<MKa0`GXEG_9>ylGi~Xo`1aT;XCBTK755BQlV6B41_js#_D>Pn~fbdy<4KqhcFbI
zMQAoO^WYSv{9@QKa6=?4Ie#@eDt7-~S9swNY@5k<Wrd9}W*Xa9%#7<nT{Q-+h+)c1
z9yLz%XQ;+ekK*As;et{Lb~B@trW99Wr;N!aGBlEz7*RuhLtN*MM-DXHknjV&k1`#R
zP#E+ak3+lgwkYg#!FBE2hzEHlzYz^iNI7WXQ{pjn^*Q_k+y*29V}H2rGSK_{>N;|R
zv73p;R2|v>)O!JrPt{&bK!xc0RPQCMxY}gt{-TkOqhj8+z!>j6Uac+B_$_;eQ!@Up
z1P#kic~itP&NcD^#hER!w9ib_Nhp*a*B;P6=51Zge5N|<GXmQu0nZ*S$%0l|uHu+^
z^lt_yyeXG4V_7{_(|@wCXB|ajv@-fgpQPr2zp_M1tUX<6WMyQ}<@kJ%460tMuQzO;
zd_3ts<nm6vn7|N0!tdQ3Hj9A6ZsU=u<(U!o;vzw)-;kR5EF?#z7C}CXMN_`$z4(-J
zw0|o>;Xbt}_}Y9`-;#sk3rR6~s1+w{ns^3)7s;tTrG23x34hxd(5ru1)UJ5;`Y@nG
zq3M3E1vjK`O7FEYb53DKqrb5E;nN~3i}h*rckda}3^wx%PKR`vX~qd>>;NlV?9?+M
zQev_S#7fP*R_^$4SX`|{_fA!O@S0JUsOZM`!^n?BZDYf$3H1qg64l2QRT4?A>987o
zepk2;<-Y;LM1QQIZlkXVQbFFDk!kEW+I#R?1B_ca$xwcDY=hY)nAUEo956qc`+0O}
zM*co9Ed|o+7szu>z6el+p!0T<fxbZfEH=p%;BUy=5s_E84so?vEECJp@fML>#b_V%
zrC+5NkuR?@4$5L<fwb+vU^e{e0MSeq`opCd_jc+4X@A6{DL|pXgM2jL>DPiYVG|)g
z?M;Mux~-!^`C;Bl)0dxM!m{ZAmLoT*<;9n}E9&DL7NTI6;aMIz*Bbek8VE*ZG}Sdg
zMWAINaW`^q#8GcuAFQlGBZ!WHxj=|E@M=-|*5mQv(ji7#3W-s6$Z;kxtp>~cciapx
z0_6cN$$$O+9pmb87R;jfelcPlr`Xy%OwlRkdbqxhy%)MzgWiHft?wX|LZKPRkFeZk
zRI=TAP0o~<cP#$vt{@_D6i;fK)s?B!%aVt6b-Ga2h#7?UXT=pecIJdmSrphDp+ZAH
ztaV_`?%b8E=UJbaJ5~wt;WPx{k<&WJo67rJQGfYLV-<(e1^s@Pqw}u0(Hv=v3m9$O
zIi5H(Kzqi-`3Q^Vx>MBDm3J)O;|ipr>@BSzH}}zVa>jCIFH@444|gT!os5=+!5<GK
zdl7QnBPO#^eDoC3Jj$B^&^@vb3>(`S6hP)_-7jG)1y0w<7d%cFeqexRm%}XuuVp2#
zu74lHYhw8o1-8PY`61h(!kQN4uPQ*HZb;bA)OfMrt?tw!p`~A7_)p!ZOo1OIFS~%L
ze1hHWfIr|_dw^==d7qnUWIkn%4FneAAVst8Q=^5|8ec-848lUF-#+Rq(Y(1e=H@+1
zxh@k*IWyI!XY99YpT9R3PJ`%n($~(i6@S4a-rXrfaTbT}l3*KYm%=O4JX5bw@>c0$
ztik&>{e8(Y^;6$emKeQ6$w;RU=3Ge0sC|;mWDD^Ijt4n(P##}2QKS=`O)+yYE5a-w
z$mz_pBUVZXD0H}jYhZB}PCR2j4{!G0wM7VZOTRhaaH+Ni`qrDb$rO|da}Y(S7=PQ`
z9q@k-Syl<{gZ}f87;+uwE~MHZGvXE9)h;)DAWHQ_6Jvvci0f>Zon&|mm(O@&!?v`S
zXYQFt=vw*Cl#F>KZn6|#2=WqN5Y#}Wd3muO#q@`Jy+9kHrWJPuLG07!U_9mEy~UX$
zYpE0}6Z6vRsp`Y2r{ECZE|te+Vt;>yFR|fPKa@WF=T!I10N5p+jJy&N+n1I)Kj@j-
zbc22?zCSLeq5W$mOL!qNyt4e>x=^)-(e%59O9PAHmRQ_y(XML@F;a2d)grgngXX2z
zRh)SJa1h2#{)0pqk^9?(TFt^66HlR5u9g}vKDfcU>C-X)A*l9g>#Tt8gMYiz8UnWk
z^&k6}bar`40WYUDnRU(`7BhGgnI2I;WU9V><|S(DH=NBSOWtWxEig!T?(ajEzecZA
zYLSPEjZ+d!N(FlJ4jrkg=l1J_Z)%Wm_HVsCyl%Xfym^#aCN(bMs<Yt-itlSBe58!6
zz+-1<hw|LCOumyd-Wm3Noqt{V6<&@1`o^Xet;r3xHv6n-e&Q0qZ4;<aO7=d!@h;{{
zviE$&nYV2b6sL0&tIZKd5Q9=t&VE&*%%_4IzMRH+pS7mjayel$RzI7GfsjvvUg)An
z4SOn;nq$j8Rw1u$e2F_W)^nCp%76Hx+g&Mwzbp5+C~+oLPAqgA+<!$U&0@uC!EYxP
zX8B;p^R~fAuoc9u1>r;-zBP>gb-S1at3oBx<y$$FC*iboA{~@PL*k)3$0U#`Q(V59
z)kMc@R;l4Wo_D5T?KqeEEi2)w7xJ|2(bD?+u-aTDv1`GuqGdCi?#YSiH`s_iu^yV)
zj1Q#YP={Wq*uf^ZT7TUjU&IZ=4wLgtGQMrk_^&iMjL0E}N@8$gQ(5o5KBXKH2)>WU
zz=WD8Clor0Nshav`v)};r9uYb+|o2<7Gu_ithAQlue$Wz%JFM2XKwvp<74VZqB%V@
z0K|NB9y*<V?9lK_hRR|T685v%DhPeL)fY*C=^Gx+%MM6eMSsA|0!p*<C;U$6ZxcQX
z49u4BFdVxY9?1&p+ioRvguJGV-HIR0o!3{{eyRXn)wK}RZ63MmwBmeHJ?l9S&tLHF
z_pb_5e;x_~_O@{kRMI15%lVozJMpaZVR6+c+GtQnFtu~d%(1Vg<Vqm?3%gF#r0gXR
z7Mj|R3(2NYLVpV>yWCW@4|AE9#pX4I<oc9m;(H4ET0xKhMBi3JT)wSMPx{{tC&7(|
z5ST`MK1){RGgBtNlYA=JxqAB#(dMRgmy@lt1v`22KsjGF`!kn8=f=@Znp(m|3*h&;
zev*4?=cLF#`?!oD9)+O^C3&ZWwnG>8QR~L5radu70)OW(2CfQw3?<izd&|yNWWtPI
z36{;D#rNIT)j=~N6Yrt2!OuIL$gC5Dk!Qo*WpeJI%qa=dNGEhE#L4ss6=VWzrGF)J
zI2HRWUWDqn2X@(vCg0=|NMVDEmt>k{^f`8OD-`0nxQY2#uapJRbuX=1U!G831J?dr
z;bjRV1%E;Go-qK&wg}(sSDH#y;jR}F-wyROoqrKbG>G=zR2_Pza-V?nn?N1L%fdh$
zokB1N5>BLAkA@w-HIjeV%}=Vh9<S4;M8p~#I`rK5=F!Zt2$*s-p5O^H@CDdnG!Z)C
zY;CV*w=}3^_>3rY!5k=9vHf|M?HBA^@uf2SJ%2K_bkkhw=-ySkB#OPreW6ktncSU2
zjuE=|Fu_Cbr1y-N!0nTRq3{dD)LxE~yh(V*VLP59pK5sVIO*^=7#;0S|2Jg!2jr$w
zs8~!d+cw0p`v**DV|C%<L(kF`-Bm1>QR22x0u|p~mt|3`<jjxI04TwA)3Hpi6`61(
z|9^yKTcL#QAFTH$+WAk4#_6#`ZsN0?Qtq5Q0qrAm8zWBh_IJr<gtfOV%M&vJX-s%^
zFi;ja=D=LOX?_|W|1-*vy;zS0SBgz9;_No1NqjW5zR>EypWUQS=kN+oZMo7fA^dyH
zyrB8+@sNEsaWYb>8hOfIhfw(dJel&HFMlDx-fkV`k7d^r{t4AuajxhRn5MRm7x*(J
zh2z8kXmK4&?B`|C4hbC9ocbnBU(#mruZ=1?!cua1_`R`b1Gcu#O^|99d!4tZ{JK1a
zFvRB56*V<=2?Bmt3<Dx1`ft{>G2phxiArQlDvi?Lr1UGEPn918`zILniXkHtmw%T-
z^%}AZtus){=fSLR%tpPpV^`(M>OC2YaXp*^nwiH&Dc^@z;sL8mz7<^DY0Sw--7xf-
zIAjaV@FRb6NLbBu8|;;8)_XRqei_@-J+vc$4wQ-=lsJy(e!4cXq-CG4V8KDq6(ypr
zB~9m%EDSw~)QobDVytp(D>V59GJl+XG+)t$Te*@i9(M4PGNTYXy>=GknxH_w;Auog
zd&1|alO@+&9F5R+JBTu-O*Kv-c1$ZRQ4mOKSKBk<`}E@jRgBhY_lF;00WP=w(g@SH
z;cb^R(wuBHP#24SZdPxUARFLh)%kgGx7lPkHr#)j*HVV=7(Mypl9lNyV1Lr(Kv5IQ
zQ6V0p|DfkY)*PNdvlOwr{zg4KD|+lAP1lbDYCp{wl442jmmbuKTz7px@{lMGjpD4b
z3bfkd%p7a;%EraNOEfr9X_eiSct(%!k?5-7K^wmiVI;i29Vbm-XF+5haDC@Y&<Zl#
z8r1L$jbG%}F|wosw6OeWI)6SMVkbqI5L3FR(64)RPgjz-dkL#?TN>6R+~)LPrF`O=
zd4r^oT&&}^brbaI%9?#_2L`TM=na-3S$wT`xdV(c{K7<>wfZHt`jnrO{Z;iYFrm``
zyy&E)TR*u78PhX<m0HIp%QBuwlm{laq{o64WdDvg8S7OPC`6GgXMf`W$48=2(5R|x
zAYLpZ^#aHEwQSjK<@t+YAm<mRkiFZ+T@sMJszhOWY>C1bZ9v&;O>rmjL+%gOL|!Gw
zcGM@|H@47?IYNuICrgxS9Kd4#Eq<X$z{N7{&nAuIi4;D|g_u=;5Un8(;tAAIY+@t(
zi!r~jPS%?hbsvk(>wiD*GyzF&@gCu=kRzz3p0cpt{m}Hw6UESC7sZ=$4S&katzi3K
z#39Y7ha*)8>dy}u>hMFVq_l4dXP<wclAnscn>VNmR$saO=?8GLFHuz^gdQuaYmguQ
z$lII99VeDIJf7K3KvcSVEwPj>KAbku*aBXCcEl_C7PO!Tb$?ykX|7m_^+*S6u%*1i
z$o_z>vBrTDd(bMoyjVkdGU9i9@%ki;Q6#us`n{3Xlfe*cBWNhxIqPC3cAGC8Wpq8l
zJlWJrtPWCppaM*F4#{Y~rQkHUJ?|XLg>_htiz2)V`$2={@9jGVut)LNNwCK%F=%$8
zs@2WEDzTyw)PGQ0)8b|nZclJTz!5@MXnq+ddq$<?JMQ3Za-TMKMeRBB^uJ4C+j#Ya
zW`iAO1@jy%r7@1(tP#$BD@~FJ_1$S;_<|3(!=f=Alj*Y3k-u2DQw>(PA+U`Y7gs3{
zc;3e!vv@nO<lpM}y{ab=2~)SGTIFpJrW<EZaxyGxihpPC6vDc|lzSHp@{2q^;M+WT
zR4a~?c`?_Osm6Ko8Y7ZaO(%*>u432^azRT_!Ji#7X|w&e7%DlkG`!=7(V4mA?7No|
zUvf+ppS2Q$qC7%DeOOx+9L_f4QI$j}rCJ(u5rhhjP^+WM^jW_`;LH(GN3f~JFX;)R
z)z`B@rGE$%fWA>U73Bw0`F)ylTwIO^Tgr@JsG=CFIxU}Mxd0(0;`W?+O|#R4Y|1L+
zg|M;`eXY~Q@Yo2Lb^zvYWjn&}_K4?Cq(ejw@_)`~kJ&`8=>$5P(F}18)(ndRyb*np
zXbsHQoy%g)SsxRVkSdgx`?D$6*rpXquBlILD1Y7LZUiWY3@r29={h<8RQMePd_zj}
zPWLtyx$Xz&*Lr4dh!%dudPd0UTu})qmiD2o9SUR_D5)S|WEa4-CFG$>_+zYyl`k3j
z-iKU$B;z`-M*u8TzstTRT;RXNPB_4<W6=<eRknKXhP;%v!lvDiQAm<&=<Vtld5nYI
zuz!kv<Yl&RRh$(ITl~3yO7ZT*&90{pJsS)bwX389HyN3mc^cCZ2JE8I?qCB&PA%%O
zVUDRFIIlyyMqF7Dg-Py63;l=$(DA(zy%w?tBEci7+-PcXV>JjlXU9sueL<-~o`qoX
zj68qkib=_+Gv0<4$Tko-gD~)e`ZOzB$A4cy&1Xf#xMe$1p+s_H*?ELtdmy>1@{?3F
z*}k#SLnU>&)VB4$dv4VMpTZ~G^A<r7zdN%r2)NF${t5PfI-vo1O}VO+uOVXopo&oQ
z3`K>^Io^*+w-@=&KFalM!uY(Ql`<IM_XfMCTgw)hJ>g98z&)=3aLMb4@)E?2sej><
zvi_229q~d9UTHDYqiVH=F^f(0&l6yeE<eRv16v&`Vbnavh$SyFQ3@u@h6bnq7<j#J
zFu4s7pf=jh=D$<pw{9b^igke5By~qmmQeu}!fRNzn(wj!8IZNRo^z&gQy!gMxWSg0
zb6)qXUBTn1vN5~F>`xo~{{rj?6My#*M&e9vF-jXZ<LvJ)Mz8W78(pq>>!UgB)tB`S
z^}E+(=m$KtKVKr7d{2wcVy)ne*j_p60o=)}DK_Brs*1O$c3d{f3ZDYFcsk>4gjpoD
zWb~-$<T=x1wN?d3<`<w%Mx%T4)DYX6**vY5)GsQWagXD-y^wKicN!cK?|&(;GYp(r
zF_aSCbFPB$)q6@`k=Y_%M#n@W(GSY6^<pyk`-Z&Nk=}pk?s`;zFmRtg8P@OPsMb9q
zcElsWA3gI(oNX(%BpFxfd@>7)WQ@=wv_`^+IZ;RMdUghSVst;3m1#}KEd6pChGwQH
zFpEx!u+gNKlgbu-2TkHFjej4!MiAg@zjdrkb<h(NenDGtRn@YUZk{aCyfjdp80VAG
zphYk1UVWs4MM?K)EU^bO8r@H0WMOz2l1sT^UAt$71zdA>I9Udz0D$ok;v5?GD_zI7
z0=8oUx@M#Xz}JII^s;^~Lg(~?*P8@ER@1`W1)X!DFI6I?QVPbCd4H*c<d1R&YKa2o
z`dP_c6_wr}$!bh9gQ1dUMi49EJE(og`v|yp%SA|;B4f|OB9(j<$NewZ)YDX=<3~)8
zzB-YV_)za{sjJB><dUq*20xDh)*9=?+7QfuUY-96)oS^7ery10?)yThJrB>zDpRPC
zfu8j(MeWCx?KIhrVSjkA9{OITtg(*Vv5`tJvt$AE+<~Ern0_X*<F|Vbhyecp&;%f0
z#6F)}v)<uAIb7UG+p#@%$mZT=Ji%EP=}g1hT`tL@2N@}9;A}y|)@9mAA6zikqmtls
zQvoFI_t%RJi+zOeSk=y;HPMOyo?bj=+LX!ww6tJX4(6}@ZhsNDALdHyUFvuw2V8g<
z0ZWREuL+dZpkY;#v(BnwTK}_EErUckcqV>$`UH0fJ~xn8M8zyX&EBg*I4N2L4KfrB
zgUaH%P$uc3yJQAchiAS|3%@;)Ehpa2-9PwCJyLd*EQ^d-?&plH{j64!U}?;?S40yW
zwCn?^E;;}&oPYaei7Y`~Dx=}RH8z?)Wjw<5H)5lH+=x2)aQ$MGPy3U>h&wP|MB}n6
zM(8_X3$;v*%ctl|2uS~}Ilp5xkWj_h1Z1hFuYlm%TKir$_Id9v92~|p+QF9$nu)xV
z4xE&8WzYoX!g@G|Mf|D74ZgQu&+&}eoKmfUR<v~R1b;Bi8BaQHxjbe?KX(yGtGLm5
zO|MVCLaEmxbR^m<to*tdVHJ#?3rY@c_0RI_4Z+R>r8yzfKBE#%t5q^$&3^#huGp?&
z0kLQpp9)^q(-}#8+1Mx457<r1doV)s@Wf*tJn!;gbKefB&jf!q_?G7+SnP-sz|fIP
z<^`l56n_;Iqv++XShEtpvRRqDoD-Z?QKJ00lxn9XwR>WhEsv1)n>XzaNV@&SMqwkb
zLHgk~aXjgvi*&sx?q_$;L@z~j-c2=YDlAvuGv`^9q0WApLO>2X!=`?DAQ+&u_!<Oo
z38NQQ)r|T>)dp|8KKRDrkS2*)T=EZ^1eUnz6o0V1renGv7pZ1>`@4)=0k_Ho5KRtw
zBahnXep8fK7Pp~oxVaTuz*Y@mVL|1FmS@;0E{?knm#ZaU_jx6Wf=y?q>py^$2T}Qh
zlqxv45)O_B=`H?E$Sw82+ew(a%C6Yxks|5sz|*&eHuCZmB`~m6@UoO;Iarw|c=Eh#
zVt+f8$`zP0$cpw16#UAR^}d`d*j-^9aFk4IkV9VMw8_l#!A0`qGtZcLvByulbn+4;
zsJ9=N*|6k1=^JJ*B8l5PU5vkoM|1er-cmZA&|KrmT+;jGeic7OP<5~Z3rN|N*%eel
z5Uk+)tBH>5OTJ{*@}PuZhrW`YYCzBUYkzs_bZ;2z<!NNS4_}xmH)^<N^gE1)J2)fW
zNc%~U$cQE(m(fnE2*{*FDnG0L5Q^+mkeOXzuj31@Z+2ntsxcFeXwh!q@ifb3E%kEa
zp??<~h_*cMLwYxv?$Aip7;omT1g=+J29p4>&(QQS_TSS<cKcm5_Ru^9W+1}YcYi4x
zr4?1*YM=n;_zL!cNC~L0K&FU5$bk;7x8`p~5Ve=(k#&^=tIkyw_iXKw09Cv-HJdY5
z$H$4a5R*>(%`Mh!x%y%oG=N*A*@ZzO+xH-m?gDsL8gzD(*)ma^xtBNRZ+8}}TFUP8
zSB84rb^a(<p1?n}vWpsxY@qdPmVW`W6oJz?2eXX-Rr#o*P#OeCM#Oxvc_sN@&VKUM
zRxO?F7NE^0Xxc>;C5AZ*N(Z3YFUNMAtVq>WyYi9f?^Jx!rTzFEFoVUH!F3xTs><$b
zCk1t7aNPHLQvI)}$Dukt)MhS)S#y+aQ>)1^4XD}AFv*SnE%{}?_Z=}dRe!FNoSsb1
zm}b^*!BZ;WK^T#8!04_^Wfs%%3f+yc8{A+fJ9124;B0Lfa!RAaL{DzfJs%9AyB4&1
z0%>JN!R)G~=R2{SBE2r3Z!Y3oea+cN^(Yu&wwu2NC9+ldJ9jYFgN3A!JpuI@ju&>C
zXzw0Jkn;fiN}D%xPXoj&q<@8)USfPjz{5ehKX9JERLg)s?7K-<w3@)e0Q)$mFlbH9
zcA*j?YVixJy4`1;dFOUJp)zm1k+Wo&#Ic>_6No$6-XlP(Q)&#l6JT+Jk5UhLzxR``
zza`CrEXqyUnrs1no2_Su`L6G<KF-<EpO1aKMzs8cR&2*)z7jTfGJm1mR#C5?sNVTB
zhMZnARRc{g7c(F4RYRJB#!L#l*?Ft7O2i0zh>kJ4MlM0hTw2&N*V|~=axDbzr{9_s
z7>H!&T@MH+I}jP#b?L~J+yIqJ>Z3Y=rS6}j$J+f-y!&z(fL-A#U8KbCAA$O19kQ^z
zABpb77Xt<xPln}~hJO-ks6T4J&MEK9Kh){RT7pkf;No4V7^pWi^PI?Z8w_5|`v=be
z1YG_1k|QSgUk9LobCtGeGGiP>r@W&&OS`eOpb^92SHvy(eJc)7_Y~wDsGs-@0Ynph
zl3$WZpS2kmO$&^5!s`OZl&Imkml2S-K6j0(D@4%Cp%N0U69a=@HJ4fB0X79OF*7(a
zmq48X6$COeGc}V@At!&Wc4br?+>$L8g1ZGv<L(yR-66QsH10I+4#C~s-Q8UR!QBb&
z76QQ@-*@NEops-vKlA2C_v%x-YFE|Sb!s&QiIOUVu!)^9NW#t*%)r9P%mYxcurYRa
zQn9m9u;Z3vPyv}aOM{K90q+<G0tJPrBghDBVP`951P1W{G(dkQ05K2{z{(0>;pXN>
zpa6*4*}FShn3;nC)M_dkG<0<I|FZn$0x)*}C-NTYWMO6tpnU(~0<yNVw*lFL-zomb
zfvO-70BjBdm|9qa0HTUYTG9%V0BT7EHGm|@7UcNegp#wdwFM9$X8{D+I)P{argn}1
z>wg>oKs#F#i@$$bb7Fkw5_SR@0i5hXK#TV<kQ)$W{})6Lum?HXSU5So{{~n%0n8kY
zY{Bn$0JZ~I*aEGcP5vt2-EM04cR}`!cJCfG@7Oz;lAROS3Fv5H4+gwvRT7i<N1b4E
zBk*6joh;sAfSu{Pr->cV`LCA##@=b(QLvGPtrGwYasz+=mCG0eFtKp5w>ENr&;3qj
z?`ZLN8O}}?wr2mTfF9roGBa{Cu?9Igz0<sN|8+b6>J#u^x;L`7w|4(K+U{?!|E$3R
z>;$qlWkg_Mea{I5zvniyuti|{YhI*nP3-_I%>S^PINSdda{)R2eL>WJ%?!<Z5k@9<
zw$|<d6Oex?0+WIr_&p_n`hTV}<Nuz>{|gfTuL%6VBJcm7xc{!ve|g0Jzwi0qsU@7P
ztrd)H-Ur|xdj{~nXpC$D@4E&d2l#8*SQ|M4{#rN|!2c`M$i~9j{l5$TcW({QKdSx@
z&VPshkwVzk>|G%S77j*^f0!+tBrM!OCQ24ypgDiQ)X4h%^8U7~*_wbHtu1Up?;`(w
zzwbfJ%>RacHMam-+5V*l$3GyDt;xUl@UGk69WV(iNvOz5(fv2u_HSpU_o)DXb+-or
z{v(ryyq(E^9{!>d5wUXvcrvhXaswDxxtQM{<NI^uW@Y#Of294Jh~-~rc_Xl+g&RPJ
znUQ~)nFa9v{m;+8PI~`FBW??{Gx=+dRKZ5JChw#7pN79+ptGanyW)RugZIw=dH(wm
zfIw~_Ai~O`9gsK3GBqO&T#K7CJRzpj-^&6$9Bg0G@U<}CuiCE1H#KaVyTb6fCYZ7D
z2hVGp`@)I+%bpDF{%9|bHC0bC=pv8Mn~Z;^FCTq_vV)6mKY&TU1Uc<gBhvG7UTzCU
zo0D06e`i8jMZfqNmgq+Zn<L!iZyG-`mwZ1g@_Ty}A7Dy-40b14BLqfm!O1CQ!qM-0
zs==|5;h~=E4}E(CblEZR6ud^bK1nAeiC}j_$GZmL>xVxsT+WED$ex&KFV1BCwxoZl
zcA|A2vyZHP)Ey<afj@LbQk)uA%o~|KE{E^TMfu49p@hND>0Dm^618N}z>u3Ow7q`b
zU#QH4Nt<*sYUy1rnts0;mXndaB$%k#($A#bZj5p@j5OKJTX-c_sa4j9G7weoMVytJ
zk9_g5;`6eWz!SrH(>+n2YwBhD8j62x#i`<sp5kHi$F^`>kEDvd<VU_v!m*)S2T@>n
z#O3@sCcQs%6j7;GHDwYwsawluF0Qa#iM;W6t~ckB^7)~H)s2~lIqR)IC|wEZHFC}L
zyM?8>I$zBAw&zfEvkR{*dzEuKxQY|%A|acPR_G~7`ELIhMk6}vs=9aw9&LXCpd05Y
zUORm!SDq$N{LF}ZBHAioT+x7w+c32hIQumOPOhBLgjPP~Q>=NoGO~~1Xgi`*+dLHM
zoFycF;V4!_kuQ!;3m!K67J`=NB++BcqS%W74}3mBEnW}M#~4Lg;WUU(LucIq(im$e
z3z;7p@mqFvX&C86%5wW#7n6Sq15f|0G*vP&UIr2lAE$nYFhtm~w$3cX0o_8=kwn*B
z9~I=cuO$&bY(~Te`>K24yzlIn7$pe(+^>&{Fx=%`k-71q=Bb~@%M}NI2wYBKG^s8>
zYiUNrQ-CV+``qiZe*7q+kY5t~6455(Q+RESl1mkJlJ}Wz-=?auR<wV|?Q;`0zC3wc
zPyReN9Ao6~T;0wfZk{qGBKR6&i)#x2XsbS9WT``q+@iZm7R$Ygz<FP#Mh=I83(9*n
z6{!FDs&Sw_b4~cTKmHg}%tj~=9(Ei_&?Lv*yzy;zSQUg=&x)e<=Tkzo9GCM=NUk;-
z`qbUt$gOH;yT8VlkMMuT(=2mCYE6IcY|hhjr2CdB+k1Okg#(UBsGe6U0)1IGD`?2|
zosWY$b^20V3vOdx&8wW)hlQ}74>3&yH~oG<$WD<^;Lx1I=SYgd?!R1C^yUkU+lpH8
z-zazwS8;j2IW=JIK|l&YaeYHe0$Z`yU>ks$B0VM{!N!8V>GXfXTaw0hMjkd*Y7&N)
zQ7WzC*mdY+Bj?dsE9{8oVwq54q$Haf-#J5m!o2i^59}ga8-C9n!G*r9)efv$IY7B*
z`#^rlN6VuQP0m#w4Cd~^$wF8^gAZ=o>*c;3;AJv`m`L=Tp$>z+;N|EhDy$oeWl#Oq
zE#@38EbON@*|>jYq)*>0?Kg<skk&7nX4Mot@G`McFfGS4QlF>%U7D6Ui9EoM8bR@M
zg1#CB9u*=G`*f;fwkJBKhGLD^#w_{y2Nilkl%aOzhZl-YOxbv_Sphwu^2^NC_v}LO
zB=d6A831LChu+O}%+)MXgK#3}t#h!Cok!mTb_`uLX`O#5{v>JtY(F=812;|jkAG`s
z=_sx`m4a!?Zz9$R=h0jwT{#a&&2VaYW4`76Evvz*2TbbHg?OmMSr?Tj>M%{it7JTN
z`n3S#YWZX*;*Y7$G+SiWw!yC<P)3DDYkO~48Bh`&$z7Cg{O2=E9IIn{rY4|@i%Z|a
z%ZUOcj{AR)zxwDal;N_e@xM2cD1Fi`e|=n-cd_axI87>@+>-t7z_i`tGbU^Y)jxo4
zLHU$#NYXOuR_$oAScx5O$dS!b$dx|iv;@(G+nBELhXdAR5em)ig1D!vTl*sWF*rUE
zufkj=z7fy69o$U$K;{k;fj#q`Quo&Xim4}XCeDAQ674yTRn>?!2)akyPecu6#VOXs
zN}^|2hKEE`2gLkoT1l(&Vzhc4mAA#SCR~GA6BjL0OrN<({-W<EieZW|)E=Dx#wD+s
zQJe}^MCo>aKfQLxeQuC5LH4&U@KSL2Cr<66E6qAYIazajOs>%DRbpdaQ9BtObEKi$
zsX~7p6@o6v2rIP7GN*6b!EkaNEObZ@$Ah_aawg(rmo=RxYm+T~j(eIjf!2yREz4(F
zp*^-WJkmL%yW8FlN!UiMAK4-)|7`YWgJ!dNt7rpAybn1%-1DJ2v|-Rtm`r;js3))j
zRE|+?ojLay)fxQQfAUh|W&A{gLdObCt~7sC0xF%RYl|Z4dkb%WniXxoAQ<L<yqMY&
zl%FdzNaqLR&<E2bKL(<)g&f|6%%JIuxWK;hUL?#FOEvb?DJdLjKB7;2Tt5g}eiI}8
z+Ou4kikTvd4Y^%0TqY1feow~vDF%BdEY-?_qgBR0FuGF!M<(?c|H^x)2<$^^q^o}i
z%vLXn0$5$BnjPiOFM#^l!vE+MK_RG_?I|~f*>7!p6~?shA}x{*D8K^XdWn0*q$*>g
zi}Y;Et(ERI#Jz?I=o3SggMQDB5?`}&jzJ!{eI!VWrg$+K=9#9#y*K~768?)D)8p&#
zNa#EjByEJv%V9A^2T>CxQsV+&%;0}M3t_Nho}ayVlOPw?IoZ#XXhn8%YsB2l$)VTM
z?U<$9uZsp4s&G>ck{#Yz1yH;C#(IOUl#u7Ncp-j!gcoWIzqHBpOAfyr_Y6BhQxtax
z`*CtvD{57j;=-Z8NPJn_@fyO5&EvvuX_{!IK80H;Q;q&C{nEX;(e50EG>3mk6`MKQ
zXD2vYMd`CJy<s1cOZBVeW$Ptld_!^DO5W&J!+L%G1`FXMRK<}&!*hvw9XJLUTa0;p
zFp}3I@M|C{9ZEa(BXnh*qT&Uya9~)P&sScu@@Ck@8Vcisd@w$BZvIUxB<gtm8N%5T
z12M%#Mf)*>t{8IllzFXq_2GZHoJ-g4Rw*0(3ERRkUjTu=t8U6NlkH$9K1489O$b)t
z-WS~HIb)Jt!G;o<T-vP)txfxT_e6E{<%!lxd(F0HYso@@&L!7a3E7wrv3J{2wkhcv
zUDGFIwkm`;8Wll$cMcj~httgbK@$|b34%Tt&2ei{>^$#)%WvAR`wV|(;~Sk3y>D1s
zmE080<-_wz_nxRP>fl<cbBm$Y3*}t0a)P_-;8=O0r-mjg#fJ9m6*4pgYA;E9x785j
zKMlbjL%;tz50LzJ9YV6Gw26f5{4?nA=2{uF$1IVuVM7<&b{`~s!ni;V9m5MGHE3h8
z`PiIq({K|V-%&9j-`{_QK1kjt<OO5J-n6bPRMg>PQtKqkr%<~;aIXPfqi$XEE&g6y
zU#?5;VlIm@&}-P>r%nlxL)|IWN6T*o<R>DvlSV_K$&F!tNr!;8kS_(nZE=rNi@_m%
zog%}IwCrPt%91U{QmC(Uj=)8_f|*o?0vGv7oc2br-Q(`i<t~2(5SUoCdYkRU2{*3!
zIyBYb2!!|y{dVhg{DO6i=C>$Bm)KVfi?`qIARa>?@&RhiLLy*SfaF{>mDI$h<-{&7
z8b%X-`Wmfjcv@#u=a<wJ<j!!BBcV`S*bY32D0Wvc9M$KtFd@I6p2dIA^t!EKXZ#{2
zE>8upXQnMtnMQw%FAtSu(N+@HbJRI3Oh;F*Ob}PDV-y9f__RKs?e-SjrX_2x+JX}w
z>>|0jNr>XiFbc6c(Ed!>E#}U0jvd`|Yee&seTfp90pO(Qi(>c<C=$1k+x3pFK%XJt
zPihbM5%iHvPf75O%O4_A!;-FK#w_n9Rrl!v!?vsRFt~rJ+yfdoPmrUw@A{z&Tj|f+
zpFjP+)S3`r->-#?1XPGVbqvfk-3|6i4(idRhO1i^`K(L{xcHVpLJ1@5_y#wz4oMDX
zVnzO#hoK;0s~Pm9Bb78s$<__=Bez9koa_h+@gFgy5z>c`s$%IULV7$lt6;#p-0Fe9
z3BGu=UIc%#C1J#BFk@+jWHk?9HhuqER+VEu{~Hoiy5=e1$7<Ix?j5HVU-k)kl0vgd
zq$=_yZx?F*?H;D<4XSf!<OhXDnoIUM%n`)A`JE@+N1XHcD$0BqSwfTpcw2~Xi;k3K
z6&=!q!aRACEwYl?#e5UVcRzK!TT7R2cZO{BF_nM5_nZ>(k71-)>+ifqwzS+qTz0e`
z=fFOAW2)?jYtG)wwR=-+AW7B<I75q68mWZrRl?MV&z*ONw0@xdtw77}KzTCB<@i~B
zI>G&mn*S}Q-0t%}-{L){;C1<^YtuYCDM{Qo@>ZyvJ>ra_F=A}7FIBC6yL($CC}(G!
zayfrx#vv<rF$4zPwc-z%5etKX)KC&voo3PCvD_S=({PGFR>6FaWlNBf0XQS9e-Y$E
zPIuagjLThm@U<}frtU5Cs$<fdeh$%cp^Z*d2XdlQJkltWlM+exI9H&<D3_jP%6hl6
zYT)Z!Z-A96l|uF*H>C_{Q9lUPMF!kTBL06=`b96iRLya_TEWF+&AhlgUv5wN%Avb{
zYk+)P-&30<RnXBDXG$=IuH*&5=0TFRi4o~oV+|e4g~^uz*`Q?+ZG2)DP~yi7Jj<{n
zsQhBDgsyFfPd&(`K3WOi?r}nBm(Hu+`^a%vZN)yfYgHt-;9X{ysV*=g7G{trSB8Hn
zWm036B~bKO07^C?Z;`{Y#gTE{#*1_xBEEQnoFvbO+#naz^>U!^?|yPy5*w>{o=(c6
z7cx*MSa}*$Uv_Tr1%&-c;TRL^m5s~rY%WHg-}rN+yLeSN#9<#L^E)@vk&ir^`~x~d
zh}bh5`lH;7RoT-|`01FjC25cK#vy-7!(?+>({0(VN+RK5?iwfyKFMmT!VrB9<E4Vy
z1~Duw37Ya~t&pEwr-!OgaI>ZAH6P~j#Nntmj<a3OkX;YER<vLTh}kL`KWCT1EAPc4
z6i0L?d#x9ABGG4yWvYK+EUEhu9oy{!_|(4hc*xSMySrzyF5^@kg7%QwLNk9z&p1cc
zLrT1XFj3?dABH{msm8+zDk0^|2!}#^kaP2_eSi+iO$e@yqfpTeM>Rpg@a<RM8wpqt
zv1aJ2Kl^|QWjX>Lc`MDN%g;(Z*zC;nPrBegw?~N}a?(0{){IAn+nV~hhkJz&H6MAR
zYn1sV`>6UGJE7<m3)-2It4M!`piYFQvE(yGQQk~kk47aHnFSo2J)&a8JsBblbBNt6
zT#p9=Q2%&YOdySk_HqqMh-{clMTL5XjPqkEi>@ysgnMGnUXl+N_i}5Mo^ej0c-onG
zI1!|HTwf<1h~v<yT2{YahY574Mt|zOWdNX{)`0oHDj*5B!iU?NXOMp_Xuf@CZ9jR_
zcJIJ))R)5C)(9}?3-$6D9JR5Vb7%m9uoU44_MYS6u_-CqIbp{ZHo9t7#mLrM&#yeA
z?TktZI+Hydv0GLPsX@-N8k)pc1gd7Frcokme(Bg0DmOD$D`Q4a*56er?_u<#bRMn)
zeBoV6e|ca$4Ay(;eyM-GaEbz4^V_J*k|G{Xy}+(I3nMCgp7u`9?es|6tI(*UNl7RT
zHwY^1%CBr+=dRiM{p9@li}}aTz_r^<!bFO`5_xy+Vcnfw2#5uTtQYUofaHp~;lPQC
zKVbSs*vCr~BU;hvEYJ|A_wmFLW@dj+U&(K2bRz2CZ*@_Qbm@QPESOM5ToLpG9KH++
zvBS_2&RXaZlxG@~>zx~NqIR;;e^@3XW3zZioE@W)@<K<<h*xku)0k``BN9pdR!s(w
zIqIO7SX@ST3@u@aO<`TTaO;{ZQl2+~qMcWnItE%omrM?3*U8~mFLp;_?yTK)Yo}gO
zjbZGy#)3D-dcS{J40W6BI7KW@Fn!r@VVwE*KP$RBxVoIqs{1FYz%QrR_??NxMb%6$
z)X)S*jIovIemw-tuh7^ZV!KN@DA@msU=UHUbtUP>Z$k3^&CmJ0J5ReAfuJp?PfR+(
ziciFL0l>JGtB2|!s%_uc;#z*Dr)eX_UmS;JA%8%Z#7=*>{Paf1;g^GGS@V(!F*t6<
zx%OqxNq;yY%!`#>#uZsyVzkSVy)0@D4_b6I=SW~*T+8QtM-E=nz<+WAo9D*oZ8T8`
z#)Kp(3~s(_6^+xg9YBo0s}%r8G3$5blDJQXvHgtOBgz0@Gwd1!bkg~-jqruupT*xw
z9-bj+-V%Ri=|>nh4fiRTs-87j&-4wwYaCygG8Hqq8&xCnfF-?5Y6AP5V6AnTRIHgY
zWjr$)e6Hrtew$=$S?YjXmVEoRrY3kzLfKm;yk{I`ERuO#eGwX`Q6-lTAAN~$pU11^
zsz2s8#;e$CO)X^830?R&o2=sM-ekII4}Q;t1R{S#`j@{dOnvGfP*56sxvB7^{&Ms5
zHRlyi!Xkd*NuscB2CcWq{(~D=EAL0_z2oYNB><V4!k7+!1N6k(aCz@A<jasc5AJj^
zC2itx?AY-uB0~|QC8Wnl0QLNV;1Su|YR<hkC>d1EFZ~p;KnlwvqWN<1Apzo>Jc4^1
zHHm*tiPM;IN^DL!=lTQsCm^GMeP3W;T6j-!Q&44Ag6?N?uA*9~QLzKW>5tX`&fB8J
z`hAb7(C@cNimUm%^cT0aCO(vR9t3-H<0!ggY4lNTBR{@k?Y~afcHcA&OR$7E2>Vks
z5lIh5Rlpc%85yyMAujhUAl)3VY%w+%Z;pRMqzn*jFFqxETA!`B3_*a!9`}5wgVMm^
zeFwYWr2(5LWY;*fD}aJ#2_YRV<{_(&os^0i83o$z3~9I-3zSYDs6fFj7IENmA%>6W
zB)~M;!0*K#a831#-c6tq9O2yToYtEc<FO^%wrAjOc6<d=e<n(F3Y{0??0D|t*YW7X
z-;AVxX24`<Jff`H^b{dTF82F|1-)ANX@~*EAW#E(UayjCP26(nTX2@3wFs)}20@{G
z`G6X!IzBrDQL-LA+e37B@vp&r;cFlX`-iO^ftH^&&fy0x@Zn&>dcs{C6!!@m!!VqL
z!AT`gbMef2m8B6sJN%;E+8NygFAcJ{Do=-hlw>^0$^KB*tUviphHHx1P=5AUF3eW`
zujqy|_K2TW3wC)7Kfo_(j{6)$1kVtJUmR%;A9Nf-fBUVj;oZP==lo%3&gqvSDl#my
zsIw1HJbT!n|9Otp=JD4?^H54##1n+39WWzTpI2u>+&7E_T^>O&Qoo3V!%^*vk0Qc<
zEyEfE#ItNBA}TEm;ts6}`kU~d*Ht+anX+K|<*!6Srf;nf5(>Kk;Bn`$<-Xbe)%l($
z&DW~u=}&D$VHCt+iqheIfInSvQiAi%UX#%$CIxc`9sI`j#?G^kAzlfpA1X~=n;>$Q
zJ5?b&;*BxRej-<@d%98A0In0TkAJ~`Xi{F9`qqYgz@RFsC-o^{9x4M85E%XVGnVDr
z{26qBxk#S8q*>k5B`rbL;n#))BZ+DJqOFttaZE(*;V19WPS;8kO)N*L4;GD((?;N0
z-p56QDQ7}gH@~MIap7&>yY}!t{fCyh8-}<6#8;R4+N;n-abAq-MF33vx>t#RsaZ1!
zH4Qe1Lsk&t5977ak8wQj1bMh|`{}-XSiTC!FP+Er!D!l^sr`h?dKhWV$$%wvox>Zu
zk)vc^k~ffMBFvOqv4uz#&F@{jB2E1o2t*{mRoln?ZWs!-N&f-_BoX!K>D52eNs*Zz
zzo5ed<t}K%o;}HCf1RITwcJC05{R5JjZ>XhH4%^LJy<bQP|kLUw<eLLkMS8(Lw|g8
z@0|+%;d4*&lAU%#a)a}MJO3-P{9p#8DC2lU-78dN?u(f3$8Vna3Gh(hPy4d#CWB~(
zV!3kz17jOg&Hi<M5s2IoWx_r6mj^C9o`pYam3sQmmRS_jBDuZSu5_4xC%v7Z1eg5Q
zAOWv3CjpJz$Rh!js@%_PRZn%=hM9>3h-uL|21$S*t+Sp_B+0kTsN(Q4M+FJyK}9*l
z+%E9gc<Z4*w)drGLC@GJaS5H<#*0{`ltu#|xsX$qpZ|15Md-@`UNv6%%eo&W_Zu4J
zZ4gY+pRX5CKU?=V?QzV1D~8H8SOe=xqmDB!wRFm#yM8rXTVj$!m2|9-i&EGG?WRea
zO6rj_kDf*fimw|Ul2sW_(XU=y8lHZDw+vqlGS}3S1D?0jNwjg|aug4C7PleF=<2(2
zr_O)MQu>(ba?S@F9Y>0B$~z(-&(w;^l8%&PO#d3&9HXnW04$?_9a~>jo9zPb6Jx^r
zuXM^E63No>LSHx`&ZO9pf$_eVE3Ub4cr{H`1m+*8X!x)s?j|PMlvVFJ?={6B)kX=$
zyO=J^l?2rwJoP4vFec(<8E)6)wL1KCh#<h_$53$BB5UhFSY9xt4Zg4n8oCLA`5G>j
zWsZWTW!J6eCIZNRg_quBayk+>EXmv2>Q${i-Yd8~uV?*?z*$$<YdA*PhrH{mkj(T+
zwkL6o|82)ENEgZ`1>bv)JMX5#^4ZwW!WY5R-PRo@_v0ScwxsC~TTJ=*o{pcRQ-reB
ze!hruduCR<y`-rDG5MLsEI$n<_d<`Rx@y&J@)32e9Fqfo9FkIL_6b#}C&r5E%V{(U
zO3O+<N?e__TNfF;0S^=9?$Jb?dcwxUeTet<-S33UL~!Zf)R2E-?_8ku5BulAm}w>1
z-4);1=H&fO_=5Ni8lOX>+FvBWzq5E{%97x-`@<&zCU*NA54}D=!@$7eSQ6nE`3rIX
zv~jAwz;E+^QoHIycMRTO&F6<9W}=1PyhtcAVy=yDrI5(ac9r`M>p~R@TAK)no%kKs
zi=gr==}<vPaRR!BQ|Xb%IIb&)>I1^iNtTyj+Z5aR=!i3HxMbz5KM2gTBB`T{A#Gd8
z8A|>e(HQ|5*9om>M7VTNUa8nGslgC<o7)b^UiU74++CiXU88&7vc;8x4{JOR1C@T_
zu}hL8XAE6eh921&bpkxDzQ$|6<>S5B2;1zj1TY(`OFWGXN>q8wx6n*{3w%xZELcu(
z!&cUqiRjObWY$a=XwXh`9o&nDE;YN-j1KQ^%y8ADi}4cABa%yHPF%E(i`<ZwoQ`ab
zfnZ91(q*3Zc|l8%C54nPV{%DDKZ6%%X^CqCN)kv1_{JNEM{x3TV!J7Un(Zeb6brI!
z@qt84knhyAs&zRfFJ6t7QH@gn6sUG|<6YIiYYF)XqP~a^nI|{rC3LvWa5^rMG66cN
ztqXnRNihhbBIJ0*Z|}ew{KS3s#0#x@JK1}GecB3KM4g%J4dbD*<`#uAoWScOf>1Y#
z{elzgbr?v%Sel}@J=ud0Va)s3?s6~^{2kd*Nv>3Ez!DYC6URE!E`Xk7Gzk7JZmgT3
zLx>x8uRWn)@@W0XDUcXr*nGA%b=uuXj==XrWpYvb`ZR}XQL>U#rkD=We&$O!oDh0{
z_2MyJ`{!PvnvYygQvA!}pZDGze|kdyxY}Kn3!<I!L{2mY)5OQA?{ZS3<C&$SNl4*T
zr-E6sgylLXKz>@V$<vy;mJv%SDEKYfb^Rxn3*9hb<|4w;d7*TD7LT;yux`%v5s+4z
zdjq76s*}oQq5C)@v{xo%$a83w%`Q!U*ks(Ic>NXPZ_wnEdtAu}wQ;gP3UN}Z-zJ2`
z4N-&B&8N%A;k$F|KPd*ZXc}narcbMWXFO^=mnzI>OYVW4kE7IPB5>A(lh2%oyL_w2
zhNOJjoyx&2?<HSd3u5QvEeMS(P$+S6?MzUEa0HlMQS_lHKC^0Pta5k%Y>C}}S#moe
zO+1oebtl$O4kS`^mS{vQz9^wQ&C8;WfO3}BHds}|NeGTl@at1owf`ce+MiC}>VtLL
z+E(?mK7jEt=jFF|@H}&VwzgYr!=+huRnSC)G_zjKeQ{k9X%}}B?L*_%ny0wFOoAtb
z40jCaw#zQ5`O?*7AuEO4va9BQZ~5)~Z*UKJ^a@4|I|P>fQW*5RjzkY!y><|{Im~(;
zzYV&k$~w7D4jfrI%Tb6IbnH7J+$Yj3htbQ}15p5?R$MS$Hw16ZJP|p<nu5;xGi>P+
zf*dxEZ_515!gsW8-fE}1U;WQMC__a2v8__>b--O(oRgd&gx}OOp|53sHD*M5P@O1B
zN(Swr{v@NY<g^qI^?DQ<6OXhy`iLZ<qV2t5@W{*V;#sKME=w$_5Nj@j1$y0*iaUG$
zHOX%`;an_;w&ES!*c!thx%;a)X8lu<$?UH-a*MXPG=F3s?x0ASv%3#!W7&%W_GSE>
z5fbZPXzhcd5t{gf3ygGszYr=*p1>e__ByLD8{u1~64eebDrI)bv+RjKgny^vX_NI7
z4Xrq7Owd{ImD<{3Byyj(65JaA58=|QxQA_SuB+6>Rf#137{iP0zk!NRa)obJolAVl
zF1&tI{*^{W>bsr0QjY0uroAOzuBXwkHRB%5Fg2br8)llz`3PBmXtE*Y_jreT!12q^
z-221teaxhDvnnxc|D{5K0am#;{t6$AB?nI4t&nGedXD<4ilGa;*$AFYix4owm48)`
zx0cHvZ?XA@(GitehVigXLZY*{4Cix{j)_`<U)Q|#gIAZ@9Bbiy0h%bb3VAN_Ukpzb
z3DYiGov`4l!YoC9XcD4zbtqV$!|bd3Jj+{pHn&GfYt1O9Yr-=aAiDF|xF<tXQ9`2F
z4a#aI4IkR_v_EO1Od_xyHOI0N;}fRI<h$C%dnwLs$A641@~+N})WJy?0NlH*B;8>v
zuDSAfXVbh&?3o-uO#;mNBOyxh)ios|7tabHb*t?2ayrs~#d1{eHo4N2SncyF{cD41
z;hNw7p$^h-gjAOdHeSt&7v4M%*h|zsXndCbO0?^A#UT)TXo>)zJr7J3osQ|O*>*Rl
z(GMFlz-X9}%EyAcCDTd2M=e?i&zu?UXzotK@H+LlVQ0Q^Ek+V7-4(xc*sXG|sVSKy
zBKz<It=N};Z7s1c1_QhyjQ8ha#DHP7N3MgOoY2ilZj4<^K%O#gk%HSaqBO*{v4Z&m
zO18XT4x6JVRBb|4h;iA=H77~Ys?r+Wb~DZ!zVF`pm#4Dyk^L%`0|s1rs@R5i`&$((
zr`y>Xm_!?U_^L;KBcRX-TIEhEs$qUQmzXx1^UkV&t6M|wWfN=5Qh@)lFVQQ+`I)%U
zQpurdmV~JhXU32=kBnE<(|Ny~6jtagrRdpliZYWAr2<v)^onBh&16Ga{OABmP(2mD
zYTQ_Gg+rw2{93C1bDcr4Jv>3?Lyli$E!7*aSU?pX?woijW4wv^0ROWz?^KmByghHU
zGd$>jg}kNc&}kCZJ@gNsmwuNLv9UPwO%Kcsy$M5Z{S>xo=u&vZiBB(Bbs&B8RP0k@
zO)v(`Gby)*aVrZ91D;mBMlX`ONx69)8usX<X~E<*=4B!O5JmgQ3Ln_L>NCU*4XZ8>
zVX1N$+@r<*P!(d}?>K{NDDOoZ-Oo*?U#<&(QpB&CK5t#?s*N^@DSqk#r9j=_>_X(c
zYTUQ&r>&0J9<X$fm+dgHR%qdW#k^XTIDu`LH3T*^Y6ZFea!xye%bzfQZkID$>WDjg
z{qTlW5_J>hTn(;t9JpfZ&T&<oW@<3*;fi^Pu-`0m;kNL^Xfu#+51&LDI$+S!RBT3n
znfv&(fW*b)q^@ukQCMyTwP~qbOF(~Pgi@||-dX~Cl;?F#zprT77ILguEV8>G04>J)
zs$C&n@~v-J#d4kvv}5?YdWr<|m@7~ZiF8$nAP?16Bm--TSN|-T=lr`|nHRHfi}8pf
z*ONL=lP!utj#{oLEwa+&bvkjf+5%C3RI3m*ZedkR5%S*e)*WW=xgi1dNyIC1i79M;
zU$;T1Yf+vVOPudcbt=IQ#Du-RF-lgTt~^&YHt3$*1%6sACqtDtNKA7{e}%(c*xReI
zmTc+iU}eRD{AKN~nR*#S!pKpT%y3TXrX6zAX4WvL&aDdij}>~FLD;)nmB+|`^2%VA
z9`TP<MSZD&Qbe)CR!f6hIve-ib>ajb(w{i8>T;q#Wn~CZXL7>B!MQyj0xI{LgQ0|H
zh?)(tkAUe!bb9VXi0wacBULh&ggXT~wyX%^Q^1c6Esu0gf%<yr+<5u3LAa+0#Dh@d
zsq>VFZ%t!MFfT7VI(Vd0?6nJj5S7O?lZr~y1n|2+w8Q9ZGMCk2_kB%ef||RFK3@oN
zo>fpYB`V66V3){5z_>tcvR{L^3-74rFD&m;rVGrvq8ou4d{%i91;Uw}x9{cw8qzwY
z+e6nrhE-KZZ5;4u5k}{1bsiOS@C)mYAze648(2v-WeCDAY!uXG8C^erY8#Cd#Vl&w
z*t<O)f5z@_yqJqHbBV%y+;f>P`!lw6N;tOEy{8jqI_P&D@WGzFtxAnQdy_#(wz!cw
z)u$Q1skYoDL;ie^kDi+uhyj?VxJKUMu^m~p?2NV6AhIz_;k$P}<!r9LUb0{leW5$k
zuNGkpTzEwsj!;{Ix+L{~416MUMA8YLnZ9ohYd%5NnwbS>Etg#LAHl^e-eG~ZspXUX
zd`Sn}tmYrueNECIJc`su1djZ5?j!I#R;M84R0uoQM5y)57W(KfZcc++4LuM|XD&hR
zZ29zy^vnSf?1wwx$q5;Oay9vAsmK6v4k5|<@Q#Y-`mxxZTX2Vernh>f;*|uTCqDR?
zBLinSNDWs?aOVCbuhe(SAMgcF0Eup`ab7uYi>UY}uN}Vv1L%tKOS}Lx+iF70OP_PQ
ze6{ggqy|cfY74mVx8nnr{BJ*Bmi8=pwvF>{8lrs+erHS%ov`Torrn3h7!v62k@B<v
z^(}AXB>sjSG;<$+;^FhWOxNA{$V6%kqXA62OOg3#8z&GOguH`4Uz(hvy}`tC3)ye|
zbwqDfIjpafGIRWwuw?W4#Uuy_zwyAi`7;1=7WCIgNBbp22EhF|VO*p-Q!5sPu%xi#
z<~@w$@P0kF=|DIf!$(CNQTG>r8=#=QrYP<gRuj+6ezrq@?Q53oq(2?`gT;h=!`;Zd
zT-nbUk|mR;(tNbamL~jOh%dpbA*HCW-%b-t+VrBFuiv_V$4YE}Eq@-TVd__)GE0UV
z+eStu+(AWyU;u4=;wJ4(mmT@-2*)viH92&Ozn9*<Hds8`5hdBdKREwXs}~WVTiZ-U
znX99{h%6_6d}b>4){NTnq<y$4$>T0O+^@n#oPu**^qYke!=N&LYrWB({@K-1CCEPJ
zXvhe%K8e^T)9sS+A*bfcbcIPG!<TOrGSl{-11+y+hFF(E6NAvgXP$ikWN<v9S0t_s
z*w&jGkI9&{Sir_GXwnFn+weK^X8=W%lXMwR3`ZA#MVgO}hitsp5W>3MgUT+-Vxn{o
zw2Lzf@F%%YaokrwqXol>+`+?$5x>0b8TF0q2svm*Juf^yl9I!U6N3S7(Wc=Ko2jYT
zso9)iGrrS6BoF*IuKqAr3flDcCg<23U=iOfI0wzwcMHwfC`7t?cWu)*c&N2X#G|q+
zex{v&e47m=3d)70YQC!_2c9-k9tMZeKJv*$JMeyq(mnT_(mb3;ycGuya2hmOfWIyS
z0l_L14Cx*UJmV+~cFTflYYEAEA%Z$hiG*#8j^j)CQ5eUAY$Gjx_<H?K-;1QVEo#I@
zzT8r^uu{3d-&0nl0g^2WiyO_2woGXA+c=zm345aait(6)8#P5*2>qV`j4TLYROV)U
z+ZBsh_NfK-PQyIhgg^*g=sOy+M&Xg&7E*~9YW0-u@_U1U<CQKS{l-gTle+p^36v6~
zLjn}8=yYhayE=#_m8>QyrRAeF7G9D|KH3bVtrWO<w$Hx2PbbWC*=KsHnB|b9TMoQ`
z@?@mNk~cQ3=WuydP`d?D7+PJ9g|`|9KL|Sth!&6uCB~!ZpaWIc9g{a3{n)ianInbq
zh-HWbqCJR6QA=u()$Sft(N^((27g=3z#KFP%X4;l`pt??F}5m;Nc$zTL4g0EjeWy$
zI$=M42x4th*)Neb_S1+MZ=dYscJpX|EwP+>$oT^sp~yn5?UBF6SC6|YJ{lg6C}FNV
z{KI3QvA>=xjGeQO*hABsw}wjK1rFCYCoRMD7!62od$BO6e(GpjCX8KzSGw-u%I9b7
zGq!TvaVI}4Qq2ubE`b0S9PXU;VYz<4u!6z$i$9;44|H5obin0B_h}sveHfO1fi~$G
zCU}XxyASlu4?6Ij)2gpGQWfN-a|%7{?AS&ZRzRc+cUy9T4&u{MDw2XQ2Uju!GqtaH
zxrgJG@r0;92KCnj<jpl4pdY7~2!*ON^OrEH?$ZYw<UkRqDhG;8-Wl-@&GSfs&2nUI
zp?)OtUOUM&eYDR*>OD!NWpX@!zxcq=)CaZHJTYl<+Wza(a^#OI;V2uC+WGizCN!rH
zW`HYmv>_@Rz(yzC@k)&y&}cb6o`ras^CS!8$9v3UojaH^PRgBK^+yB0Q7dc)m!QT`
zCzVRhl>D2I8H)NZ`v;84oi+@+&32RcEnw%*vkS_OE%*$AhdJm(1&!N(ulqGE5p|zb
zHtYs0Py@N4kb>^o84IvHRte@g61^4qT&Zi8dx;YARnde~+u3-+s+twaBRviVBXuAf
z!;p|{OIyt`rB>qR9{Am7P1=Zx+~nC-YA`nH@8YVOE@Nc7yL8xZOZw65kX)d{ow%P_
zM#TrE_Dli?lsI7`cDea~VkvX=zK|l~Ol^>k6!aLW><WCDd~6G2lI}(fxoyYR`NqlE
zaQi%J6drOe@H1VieidI0;k5pCK2?+qtB|~yYklZdZuKPr^_cWw#s5AAyK<-BzJK2U
z`!H60fuV{a(IUd=Xery;iJ;^P^;dWy22J%$(LS46P30@p=7z_A=Aep9gA2?tZti0x
z>12pA8QD{n!b2>;#q#Y~G_-~r0_Guqr`2WKbmvWlQ=7AZpZJ*f<;Rgzam(cTws1p%
z%8+;cUKIz;m3iP(`ILaGl&IySCs~8(RDEY*pO|XQC>Y_x>j)~7>zp2*B}R&sP&@35
zY)~X}f!FWmv;u{HiK98BOBQt@S%nV2Hx{$clIBK=J<o&Ni{S6XPf2CvmRYo~8Ci{y
zKc;Sv$388aowd+*O7e{^R~ee11B5kyDA##KZwROleIx($W@6mV7EZ#oDI4*$B=Vv%
z_YF3K@ChMnPY5I-q`@!vRkPs<H%=;q_r$hY)?GB*&>_1iSLNv2=gT}0FD)g=@>D3s
zR)s9_!48_|;(@@wKu0Qc8v^iu03x|n+P5YL0#+dfH#0UhGM9lU0v-i7Gd49cm!v2H
zjsi6@mozB?Ck{9YFHB`_XLM*FGcq<blM#(5f9+aXa~n4je&?^iW3(<2gBz%FrAo3a
zS!+wyO1@ow7*QjM2@jcv<Ls~BZU8gH9K5o5j4cu~J&nfKCmJM~FeD>PSOOPJIASbe
z5-7~U$OzMLEji&Cj)HKcNlCcnFe)OrfKd~nWsgZiWTxQW5(zXW9g!Bqs*EU(BO^*1
zf1(A^87w1U1Qw<QF&yrdB$i7U6|sW%m;w<Qtj!e6WdZ;whZi=Y0vJpq3=51PlxMi8
z25=hDYCx5t3=PV0)XQW<s4VL-!$I8)kQ+&aV<5+9A{j#-z^(upvl)@b!6VRAafTF0
zlmVTYvY>#1l{Etm06y?)qg2m0q8+Zye;6?t-~(xffeDNvMhKVzYZ>W)s^JgpY2eQ?
zcnWmqGGYv|OoIV9*g&ctb8s!WhIL@HXgOjP7zGrw3M2yzs~w=X#2Sr;C)NT#_|RHs
zz%j%+Zvo;mu6tl;$Akh$!U(u|MjS^gYT|?e=K`Mu`M`!aflGmc3|4R&%?l!(f6c%T
z@X$F8ZgOywp<+HpU=0RW3|$9K<{5|vFJJ_Ln9Oj8eu66pD{_n${0CPcn{ht_z0D2s
zS#-B#J)U{T!)ni;_vi`v3RDQ&YeGML{sOYVo}kBMIU0TI{rc;ArcCy&<J%dvW&O;u
z9T=7AYcTnCMj`rOzU_RO?Q1aje^y52c;Dz9bWz6jt=>TwrFGvyhHYn*qHp;w5}E4@
zNY|~5(y2bi$96^o=GnpAQh}#;kte{>YzLi%bA8@a<#ak(kmt_{J;jWUGfsV^>39Rt
z1}To$Fo#~vrh}VeLB0}tb#hARyW;19cyV71=Y8Py?;1P-Ws<Qzlo4b4e>T1gYx|Bs
z1&jBA57M=-;><`-XCnPu@2k?TuVB;}??ZKf!X|}*c?yc6Vo*m6>jxUt7Mb-0M3_QZ
zTp<>PPc_I=M8c~06e*TlSPADtaw0d_9TVPHg4!GRXu>;cu+J-<Ybpu;3N*5iS|ZYU
zVyC9H;xo^^6$i*I<#UAye|AR=EBF`e2-OSe4oVbwCcI!91kY6Wc#SWdhx-~nS6Jdv
zPQqJ`el2)j-g`cz0}WroI*%Dg2|OjP5+klrj;U^Vjjxh!;@p~k)>ZOL8Hf=_cwwwX
z6kA@%$tC*epZ9;>x4$Qm#US29jCmjDG3R2GYplii3}X^0$dj%4e=lOUhk5SvFI(jU
z=9Nk0S2=`TX5`<uK9o6zGRha?bUfxaw#Z4&mw8^+2h6=H@8h-4)3P4K+L6{3hrMXL
zMtW%zudjAheMsv<Ii_`^!XB?phu$T|)Fx>i@pZy?o|zvN94S_wC^`J5k%|litZi_E
zdZ#{&wFle#SiM`3e*$Aiq!^rCR<B(m#akeHUMW>U^zdvFCY9Mt_LoDYl4G*f3%3%H
zRj#hws!ectbPKv5Y<|{1t3tro6<-XSRE}396`ID1vaG(Bz+@jQaZKseoOo^Hb@{xS
zgZoMHEQytGzRh~~=Tkv^vvpF;2eaYRVmb@W*JVD2C+|O=e|$bad-meu;+$phV3a@1
zi3}Xa$J3wW>mi(e4$=5<C;_b%2&PB#K>_WXcEI@}e|l96A0FY>3TOa}@a&M`!}CQx
z8V-&o52FGoK)6{H;|~C{M?b|`&_;rg$9#s(-7mCwnh!_Q34NFq`J$N7@$w*`IURxw
zJl;V=ST>Q{f3$mfa&~(45wyErryLyhqH=Id`ahJ*O4YPmtJ?kWM?t5Pf<6^96hr@8
z%%*ki%s$#(U7vk;{Ry=*-O6zjm17u^e}i&ny>j#6&$=pZA5~7zzudey232l8uk#cf
zeh;1YpASH-Hfp5|X!h5_P-Yv{84pMKtfrD-`>6ExfAxP}e0lTi;^z2bok~yw4<&TA
zTu6Z}?n))88~$-Cm9jO?scEA{s?7^}Nw4d=h<$W<|K;q}m(R~`PT#%561T=du=gJd
z^pJq_oE0_+H^yq};BDn0&iNXRRLwtGU858FlAh9Y(CjU}qSy3>zM~&#PH*XLKKOk;
z%IA;te|DA+icxXD2v;+d<S&0O78Mz;!-D}GOtJLb(SrI#Y4LrTk7)68Fv`dDo<7h=
z`uOM5V=<va`a2!bF`dw9Na-r6VH{5H=yy7!a~fbQ=;Co!6m;>&lrHIVa#zgegXyfG
zKj<IyC;hWs2+Uq(?(+EP{Qc>(%i;KTIlsv#e{T-2i-%>mm+)h|Lv9sTJVu-nNggqt
zddc6eyu`#BFIlznlH#?v3s&@X?|!<R(aCTUD&V~M0Y-$Hkyxf}guSZa>rZDVpWa3z
zTu;ZBQ}^cZqPQEbv4h5F#ZP2$&xXR17RP!$!P}J`GPlMKte$|ia&lA!>=}IpLHmYY
zf7Al^BmERYcUx=@rsHuQBD9Rk`(a|?T8REu1!*vR8L9cKnD{09RahW~tq)psOD$yA
z$LU^G{`man^6a?bq79*nF5<Ry5toUJjIFsy)d~vJJA|rKS=~d&4IllRzHa&HJ(L*q
z)o1WlJ_ei3^U2*NzuhiJqoUJ&lRGR8e~J4>e~$;h)9=e-z8J!4-<Q)xad$haF8q}8
zCd~48;nMrQ7!L!~Rf={Mu__|kF=6-GCXcR8uV26FhzWGyy10N=fE#SuSTioL{Yv6O
zyILzCnyrAUIvblrhGfmicnL9q5rNTA#zJ1(Fv9MTZVOMfobuLOtSnbIS;ubte{wRI
z-VG-Ypuq6{z5q85iaGhpeZY*M>n|%n=}$NeBr%!zwJTac$ry|=)*{y6YxJoPHFS1x
zyN^vDs6HgTI-$J#$Sh+fMi0@V3PZ@wZP`UEZzdRVEX^u7w$&1gIO7ZEpl1BvYUU4H
z@FI;?DAHzAHpFxd$@y(3hg6Orf5Mba02G%U?MjrUaqF}+nvR9O1BzT}GhzfihZt>G
zi&#xRGz`_~epdEUN#oIJJ^Y4gv;7F;V%1Z@ol5$uZ|#hS<XYY?{4Ek8tB&(|;XnCr
z9Xx9V!hZpY89`5nv-tvZkzkkjCXY87Zaxn07LRlPm8WNYH+?^W=)1#we_GQ%Z35Id
zD>m`iCLUoDJW3>ZbQ_P}A5Z;s*MQ?cda1!tGKC`(II<1L><_1TLTduzEq|qecnZk0
z0onZlwa+9?NUabOB7p?ayTYU+{4&S=*V8SOZhw@`Q(zOLY+}@DSlKkpMGKJ2_6M}~
ztk6W(3V3PL2s4TCvZnEre{ExB`(xeI(l_ARRzmQauUP_D?gE$Z4|h|0+l0$E!o}9K
zf-81`EB1%Gt@CO^ZiTAU>rAZNE|_H-^Hu&(%%S<3F0q;G(bvN}*kZ%SVkG1roYnjQ
zyHyyO4P*G)xv>qsT`oq@pUfj}5|u#Mqzr_xnyDrZVIw8+gGPEpf6Cf$h0PP}dM-Y>
zniW5Ys#o?fm6cOVt>e@#=CT?Bv)r^!1aXPqdh65`q)fk{Q<my<$Qp}@YO!p?Vj^9v
z3$%2l9I0z^%9&n_lpQlGvr4{UW|jbuwnI#1Vqw>d5jqvqfksru7qU}Kn3;9VtSl)u
zS(2xUs5I#-aNccEf5*~ovc4Fv%+2fO<{dim%EG*EVV;&q!fmoVx0|AdSH@;`gRU&K
zwy5dK)3oDq(_*Z03EPwpGFgjv_=i>gVRiq=L`X+qRhHIWv;1<5cVuz3S~un}F>Odl
zYX2y*Gzl+-;}IF^-!CWA!Qss!pDhkZXtc;LukNSQ#RSeSe|>g<ACNSu9H}f&&*6yc
z*+OiTgzUiocrl?Mk8q)k(l^8LaM4)3;``Qw7#`rNGD}5@1OHpdgdz|8uObtQI`F@b
zOep%m|57rcm;?Wt$p*A))rxBkFni#CYMBUhNHGZovGbILqk&{<AdUtK#jm8`tWZqb
zxa`4-s~Z<bfAgr+JV@XvEHoESL7=sGiUEx!Qlq1-1nb0v3zkgXkkC>hbq^X!L`3=1
z6P%5QKl?`qL`RR^4<D8@ABHZXdV*aX?+r`t!BDOn6PhWqv><spV0JbUg(k_;wY*8P
zgy2mQeK{`%_%X!45RqjK8lvJ&ROO;6TMJqB%zI2Uf8=_v$r=&Yue14(Q*{zbb9EBP
zYI`$XKH*#k6HTP*%A=7~x%%?9Lg_u%K}N8wMHk3kslTF2udZ4H8{ss$+tAbJu)1&^
ztAP5lYjrT1Y-^zrl}hB)6!EIWNsIh;6lyDLWp@);#ayWn`c_u9sxG#wF^XAC6dubq
z&)BA#SKAeZVFCPwa1bEkC^yE*8l3o3;xJur!qJ)y#{*#ES;xl-FLF3UOdoooqbKEg
zr)<ZQozXiRzdMW@=R<J;#{>F)UaWfew@`JC2I0fRsvZ3w%v37UlfgO@mwq+^UQshO
zIYT)zHZ(ayIYvT8FgQX%K{+xxI5$K^H#ajdGc-ORJTo;pLpd@wG&w^#MnXm~I6^@|
zIWjpoH$+7@H#0CZG(KGlFHB`_XLM*FH8M1p0VM({e;w1iQdB_@#_{gKI|>LWvg`uO
zuDXDTqR1+XH&8@T5Cv~|K{BiK0ek>c%QETE+{l+Olz<{JFwnp%%V$96|0Tcrw(FcZ
zv(r6hX8)2|GPjwEk&b5KD#;<iF=|&xO|S!wQ@c#s3A^AFwM(Skum?_4yGYs#`(TXP
z6sZ~Zf5RDS4blN<fpKb+q*iEybJQ-74njMye4HnBKqs(`sFM!CVVI{jLF$4daGM&7
zrfw*}9cp|$^*|9;sGTL1pqHQQD%g1HgMKK(01UcWtIo6nRUSSGqc993pi>)y8gQ^l
ziK9zAmrei=Cl*NhtNI^9Zq48I4Hiw>pAJW-f2KCp-MYT$5uL7%U42tSXW$xKhgn#F
zX}9u(Ub_jm*w7&92F&qOySfYaVGS1DMz-|YGAzM8cmNOK5j=(`unJG%89av<EU$LI
zbZe|?y>)7&U@(Kh48}1S$6yG9Aq++@7`<TNf`JRhEEuz3xT3{V<C~UZhJqOiCMTGj
ze_&oh*MOl3MkRCun3rHmqTn|FOAGY4z3Nsfy1oCYRC4?LTB+CV>u;q#x6L-Ce)r{u
zQrUg|uhM|~r<zj5{m%!bs;9!1c`E(Ll-o1a*-WjN8Q#c@u4TsFWX9iS>hJWL?$$KW
z+Mu;TYlGGX%?mY;u7<j*TcK;9xu^TF4(Ms}V_tM_+x`PTUTg`MUpWFD3^gz?F*FJ#
IB}Gq03JM${xBvhE
diff --git a/paper/paper.tex b/paper/paper.tex
index 4184aaa8..9d8d3b04 100644
--- a/paper/paper.tex
+++ b/paper/paper.tex
@@ -110,8 +110,6 @@ reviewed, or in any other way considered for presentation at the conference.
The margins in 2022 are the same as those in 2007, which allow for $\sim$$15\%$
more words in the paper compared to earlier years.
-\cite{abadie2002instrumental}
-
Authors are required to use the NeurIPS \LaTeX{} style files obtainable at the
NeurIPS website as indicated below. Please make sure you use the current files
and not previous versions. Tweaking the style files may be grounds for
@@ -461,7 +459,7 @@ Note that the Reference section does not count towards the page limit.
\medskip
-\bibliography{paper/references}
+% \bibliography{references}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section*{Checklist}
diff --git a/paper/references.bib b/references.bib
similarity index 100%
rename from paper/references.bib
rename to references.bib
--
GitLab