From c532dfd80ca3bcc2aee031c7a628a753fb592842 Mon Sep 17 00:00:00 2001 From: Tikhon Jelvis Date: Wed, 19 Jan 2022 14:05:04 -0800 Subject: [PATCH] Updated to-pdf script to explicitly use APA citation style This makes sure our style is consistent and matches what we filled out in the copy-editor notes. --- bin/to-pdf | 3 +++ book/chapter0/chapter0.md | 2 +- default.nix | 7 +++++-- templates/latex.template | 2 +- 4 files changed, 10 insertions(+), 4 deletions(-) diff --git a/bin/to-pdf b/bin/to-pdf index 2f2cac6e..281264ea 100755 --- a/bin/to-pdf +++ b/bin/to-pdf @@ -30,6 +30,9 @@ run_pandoc() { --metadata title='Foundations of Reinforcement Learning with Applications in Finance' \ --metadata author='Ashwin Rao, Tikhon Jelvis' \ --metadata reference-section-title='Bibliography' \ + --metadata biblatex \ + --metadata biblio-style='apa' \ + --metadata biblatexoptions='backend=biber' \ --filter pandoc-crossref \ --pdf-engine xelatex \ --template ../templates/latex.template \ diff --git a/book/chapter0/chapter0.md b/book/chapter0/chapter0.md index f2ec2953..335102b4 100644 --- a/book/chapter0/chapter0.md +++ b/book/chapter0/chapter0.md @@ -61,7 +61,7 @@ This jargon overload is due to the confluence of terms from Control Theory (emer It is also important to recognize that Reinforcement Learning is considered to be a branch of Machine Learning. While there is no crisp definition for *Machine Learning* (ML), ML generally refers to the broad set of techniques to infer mathematical models/functions by acquiring ("learning") knowledge of patterns and properties in the presented data. In this regard, Reinforcement Learning does fit this definition. However, unlike the other branches of ML (Supervised Learning and Unsupervised Learning), Reinforcement Learning is a lot more ambitious - it not only learns the patterns and properties of the presented data, it also learns about the appropriate behaviors to be exercised (appropriate decisions to be made) so as to drive towards the optimization objective. It is sometimes said that Supervised Learning and Unsupervised learning are about "minimization" (i.e., they minimize the fitting error of a model to the presented data), while Reinforcement Learning is about "maximization" (i.e., RL identifies the suitable decisions to be made to maximize a well-defined objective). Figure \ref{fig:ml_branches} depicts the in-vogue classification of Machine Learning. -![Branches of Machine Learning \label{fig:ml_branches}](./chapter0/BranchesofML.jpg "Branches of Machine Learning") +![Branches of Machine Learning \label{fig:ml_branches}](./chapter0/BranchesOfML.jpg "Branches of Machine Learning") More importantly, the class of problems RL aims to solve can be described with a simple yet powerful mathematical framework known as *Markov Decision Processes* (abbreviated as MDPs). We have an entire chapter dedicated to deep coverage of MDPs, but we provide a quick high-level introduction to MDPs in the next section. diff --git a/default.nix b/default.nix index ec871399..59e40d4e 100644 --- a/default.nix +++ b/default.nix @@ -7,10 +7,13 @@ let tex-packages = { inherit (pkgs.texlive) scheme-medium + biber + biblatex + biblatex-apa footmisc + noto titling - xpatch - noto; + xpatch; }; python-packages = ps: with ps; diff --git a/templates/latex.template b/templates/latex.template index db79a325..1cefcd75 100644 --- a/templates/latex.template +++ b/templates/latex.template @@ -55,7 +55,7 @@ $endif$ % The `babel` package, among other things, lets you determine what % language you are using in a given stretch of text, so that typesetting % will go well. Here we specify that mostly, we are speaking English: -\usepackage[english]{babel} +\usepackage[american]{babel} \providecommand{\tightlist}{% \setlength{\itemsep}{0pt}\setlength{\parskip}{0pt}}