/lecture-julia.myst

Lecture source for "Quantitative Economics with Julia"

Primary LanguageTeX

lecture-julia.myst

Source for julia.quantecon.org and notebooks in https://github.com/QuantEcon/lecture-julia.notebooks

To contribute, you can use GitHub's online editor for small changes, and do a full local installation for large ones.

See https://github.com/QuantEcon/lecture-julia.myst/blob/main/style.md for some basic coding standards.

Online Editor

On this website hit . to enter into the web editor. From this, you can submit suggested patches and fix typos. This will help you create a pull request for maintainers here to examine.

Local Development

It is straightforward to install the Jupyter Book and Julia software necessary for more significant changes.

Setup

  1. Install Julia, Conda, and VS Code following the documentation for using these notes.

  2. Modify VS Code settings and consider additional extensions. Some others to consider are the MyST-Markdown and Spell Checking extensions.

  3. Ensure that Git is set up correctly. In particular, this ensures that Windows users use the Linux end-of-line characters.

  4. Clone this repository (in VS Code, you can use <Ctrl+Shift+P> then Clone then Clone from GitHub then choose the repo as https://github.com/QuantEcon/lecture-julia.myst).

  5. Open this repository in VS Code. If you cloned in a separate terminal, navigate to the directory and type code .

  6. Start a VS Code terminal with <Ctrl+`> or through any other method. Create a conda environment.

    conda create -n lecture-julia.myst python=3.8
    conda activate lecture-julia.myst
    pip install -r requirements.txt

    This will install all the Jupyter Book packages required to edit and build the lectures.

  7. Set the default interpreter for VS Code's Python extension to be the conda environment

    • Press <Ctrl-Shift-P> then Python: Select Interpreter.
    • Choose the interpreter with lecture-julia.myst which should now be automatically activated in the terminal.
    • If the interpreter does not show up in the drop-down, close and reopen VS Code, then try again. Alternatively, you can run this step at the end of the setup process.
      • Whenever reopening VS Code, re-run conda activate lecture-julia.myst to ensure the environment remains active.
  8. Install the Julia packages required for the lecture notes.

    julia --project=lectures --threads auto -e 'using Pkg; Pkg.instantiate();'

    On Windows, run the following instead:

    julia --project=lectures --threads auto -e "using Pkg; Pkg.instantiate();"

(Optional) REPL Integration With MyST-Markdown and Julia installed, you can ensure that pressing <Ctrl-Enter> on lines of code are sent to a Julia REPL.

  1. Open Key Bindings with <Ctrl-K Ctrl-S>.
  2. Search for the Julia: Send Current Line or Selection to REPL binding.
  3. Right Click on the key binding with juliamarkdown on it, and choose Change When Expression, and change juliamarkdown to just markdown.

Executing Code in Markdown Files

If you installed the REPL Integration above, then in a .md file,

  1. Start a Julia REPL with > Julia: Start REPL.
  2. Activate the project file in the REPL with ] activate lectures.
  3. Then, assuming that you set up the keybindings above, you can send a line of code in the markdown to the REPL with <Ctrl-Enter>.

Code can be executed line by line, or you can select a chunk of code and execute it.

Example Operations

Building the lectures

To do a full build of the lectures:

jupyter-book build lectures

or

jb build lectures

This will take a while. But it will populate your cache, so future iteration is faster.

On Windows, if you get the following error:

ImportError: DLL load failed while importing win32api: The specified procedure could not be found.

then run conda install pywin32 and build the lectures again.

If you have Live Preview installed, then go to _build/html/index.html in the explorer, and right-click to choose Live Preview: Show Preview.

Cleaning Lectures

To clean up (i.e., delete the build)

jupyter-book clean lectures

or

jb clean lectures

and to clean the execution cache you can use

jb clean lectures --all

Debugging Generated Content

After execution, you can find the generated .ipynb and .jl files in _build/jupyter_execute for each lecture.

  • To see errors, you can open these in JupyterLab, the Jupyter support within VS Code, etc.
  • If using the Julia REPL in VS Code, make sure to do ] activate lectures prior to testing to ensure the packages are activated. This is not necessary when opening in Jupyter.
  • Finally, the code is written using interactive scoping, so include(_build/jupyter_execute/dynamic_programming/mccall_model.jl) etc. may not work. However, shift-enter within VS Code to the REPL will work, and you can execute these with SoftGlobalScope.jl if strictly required.