artemsolod/lecture-source-jl

Source files for "Lectures in Quantitative Economics" -- Julia version

"Quantitative Economics with Julia":

About this Repository

This is the source repository for Quantitative Economics with Julia. These instructions required for authoring/editing the textbook and notebooks, and are not necessary for typical usage.

See LICENSE.md for licensing and copyright information.

Release

For information on releasing a new lecture version, see the docs.

Usage

WSL and VSCode if on Windows

• To get "Ubuntu on Windows" and other linux kernels see instructions.
  • For the installation, run it in Powershell as an administrator
  • Hint on copy-paste: One way to paste into a a windows (of any sort) is the <ctrl-c> text somewhere else and then, while selected in the terminal at the cursor, to <right click> the mouse (which pastes).
• Install VSCode with remote WSL support on windows
• See VS Code Remote Editing
• VS Code Remote WSL

To open the WSL in VS Code

• Click on the "><" icon on the bottom left hand corner, and open the remote folder in your WSL image (e.g. ~/lecture-source-jl)
• Choose "TERMINAL" to open a WSL terminal, and run any of the above jupinx or make commands.

To get git credentials integrated, in a windows terminal (i.e. not in WSL) run

 git config --global credential.helper wincred

Then in a WSL terminal (within VS Code or otherwise),

git config --global user.email "you@example.com"
git config --global user.name "Your Name"
git config --global credential.helper "/mnt/c/Program\ Files/Git/mingw64/libexec/git-core/git-credential-wincred.exe"

(See more details in Sharing Credentials )

Prerequisities

1. Start within your home directory in linux or in a WSL terminal

2. Go to your home directory and make sure key dependencies are installed

cd ~
sudo sudo apt update
apt-get upgrade
sudo apt install make gcc unzip
sudo apt-get update
sudo apt-get install libxt6 libxrender1 libgl1-mesa-glx libqt5widgets5 

2. Install Conda

   • In the Ubuntu terminal, first install python/etc. tools

   wget https://repo.anaconda.com/archive/Anaconda3-2020.02-Linux-x86_64.sh
   bash Anaconda3-2020.02-Linux-x86_64.sh

   • Create a directory .conda by running mkdir ~/.conda if the warning "Unable to register the environment" shows up

3. The installation will take time. You should:

   • accept default paths
   • accept licensing terms
   • IMPORTANT Manually choose yes to have it do the conda init
   • Delete the installation file

     sudo pip install --upgrade --force-reinstall pyzmq
     rm Anaconda3-2020.02-Linux-x86_64.sh

4. Install Julia

wget -qO- https://julialang-s3.julialang.org/bin/linux/x64/1.4/julia-1.4.1-linux-x86_64.tar.gz | tar -xzv

4. Assuming you installed anaconda in your home directory then,

• Within your home directory modify the .bashrc (with VS Code, you can edit directly after connecting to WSL)
• Add something like the following:

export PATH=~/anaconda3/bin:~/julia-1.4.1/bin:$PATH

Hit <Esc> to exit insert mode, and then type :x to save and exit.

Then, from your terminal, run source .bashrc to load the changes in the current WSL terminal.

5. Install Jupinx and deps

conda upgrade conda
conda install dask distributed
pip install jupinx sphinxcontrib.bibtex jupinx guzzle_sphinx_theme

6. Clone the repo to your preferred location (note that WSL+vscode+ssh cloning has bugs, so use https).

On Windows, you want to clone to the WSL setup, not onto your windows drive.

git clone https://github.com/QuantEcon/lecture-source-jl

Precompile the packages used by the lectures before building. To do this:

1. (Optional) If you have a major preexisting julia setup, consider deleting ~/.julia folder to start fresh.

2. In a Julia REPL (i.e. julia in terminal if your .bashrc was edited above), run

] add InstantiateFromURL IJulia; precompile

3. Start a new REPL

(If you didn't run the package compilation step, then cd lecture-source-jl/source/rst) In the REPL, run

] activate .; instantiate; precompile

This will take a long time to run. You can safely ignore build errors for Electron

Building

There are a few different targets, notably:

• make website, which will generate Jupyter, execute notebooks, and then turn that into HTML

• make coverage, which will do steps (1) and (2) above (with otherwise hidden unit tests), and then generate a report about which notebooks fail.

• make preview, which will do steps (1), (2), and (3) above and then fire up a local HTTP server.

• jupinx -w --files source/rst/getting_started_julia/julia_by_example.rst, or any other .rst for a single file

Workflow for Editing with VS Code

Setup A few useful extensions to install (in WSL if on windows):

• https://marketplace.visualstudio.com/items?itemName=ritwickdey.LiveServer
• See https://github.com/ubcecon/tutorials/blob/master/vscode.md for general extensions as well as Julia and Python
• https://marketplace.visualstudio.com/items?itemName=lextudio.restructuredtext which requires the python extension
  • Note that for the RST extension you will need to have python extension installed. When it asks for the python interpreter, choose DocUtils rather than sphinx

One Workflow:

• When you open the project, in the julia terminal go ;cd source/rst and then ] activate . to ensure you are using the package/manifest.
• Edit for the content. With the RST extension, you can get a preview (Ctrl+K Ctrl+R, or clicking on the preview buton)
  • It will have plenty of errors since it doesn't apply jupinx, but will help avoid obvious RST issues.
• When ready to edit the code, use Ctrl+Shift+P and type Change Language Mode to change it to Julia
  • With this, you can use Shift+Enter,Ctrl+Enteror Alt+Enter depending on your setup, to run one line of the code at a time in the Julia REPL. If you activated the Package in above, it will use the correct versions of those packages.
• When you want to see the results, run in the bash script to build the files

jupinx -w --files source/rst/getting_started_julia/fundamental_types.rst

• And finally, you can watch the file with the LiveServer extension. Go to the _build/webiste/jupyter_html/getting_started_julia/fundamental_types.htm etc. in the file explorer, and right click to "Open with LiveServer"

Options and Special Cases

Specifying parallel execution (i.e., make coverage parallel=8) will use 8 cores instead of 1. This leads to a notable speedup in build times. (There are some zmq errors that sporadically pop up at very high core counts, i.e. above 8.)

You can build only a few notebooks by jupinx -w --files source/rst/<file>.rst.