julia-physicists-cheatsheet

A handy reference for the lecture Advanced Statistical Physics (and others)

What is where?

• If you just want the cheatsheet, go to julia-cheatsheet.pdf and you are fine.
• If you want to create it by your own, install pandoc, the eisvogel template and run pandoc --defaults settings.yaml -o my.pdf
• If you want the edit something: The content can be found in /content, from which the markdown files, listed in settings.yaml, are piped to pandoc. pandoc can parse this files to amazing many formats, e.g. PDF (via LaTeX) or HTML. Feel free to create an issue or discussion if you think something can be improved.
• If you want to understand, how this works: pandoc takes the settings.yaml and parses the markdown. During this it uses the default-code-class.lua-filter to set julia as the default code-language to codeblocks (so we do not have to manually everytime). For generating the PDF, pandoc uses XeLaTeX and the /.pandoc/templates/eisvogel.latex template + the custom title-background.pdf. This should happen automatically by GitHubs actions (by /.github/workflows/ files), everytime there is a commit. But I do not understand how 😉 (there are still some LaTeX packages to be added to the docker container).

Introduction

This is a (not so) small and elective collection of important code-snippets for the Advanced Statistical Physics lecture. For convenience of julia-beginners there is a small basic tutorial, which is directed to people who have experience in programming but are new to julia. Another good start is julia express. Other good cheatsheets are the one from juliadocs and the Matlab-Python-Julia-comparison-cheatsheet from [QuantEcon]{https://cheatsheets.quantecon.org/}. More comprehensive examples can be found on rip-tutorials.

Specifics of julia

julia is a high-performance, dynamic programming language and very convenient for numerical calculations. Especially in comparison to Python the speed is sometimes magnitudes better and close to C.

Differences to other programming languages

• julia is quite strict with the definition of so-called scopes. This is the region in which variables are defined. E.g. you cannot call variable from outside within a for-loop. More on this later in the section pitfalls.
• no semicolons needed at the end of line (but they can be used in the interactive REPL-promt to supress output)
• if, for, while, ... have to be terminated by end
• strings have to be placed of double quotation marks: "foo"
• indexing of arrays etc. is 1-based not 0-based, end is given by end, e.g. f[1], f[2], ..., f[end-1], f[end]. Be aware of this when looping (out-of-bonds-error)!
• a range is given by start:step:stop
• Julia saves arrays "column major", whereas languages like C and Python use "row major", so use loops along columns for best speed. Details: link
• Since julia is very fast, in most cases you do not have to write "vectorized" code. Instead just use loops.

Recommended IDEs

Although you can use any plaintext-editor for writing code, I can recommend Jupyter if you are used to it or like the use of the so-called notebooks (IJulia). If not or you like more plaintext based IDEs, try Atom or Visual Studio Code, which have nice features like displaying plots, a built-in profiler (shows which lines of code take much computation time), workspace (view of all variables and their values) and progressbars. Read the respective section for more details.

... continued in julia-cheatsheet.pdf