Prerequisites
This section provides a set of instructions and prerequisites to use AIBECS.
1. Install Julia
First things first, you must install Julia. Click on the Julia link, look for the "download" buttons, and install the correct version for your OS. Once this is done, you should be able to start Julia by typing
julia
in the terminal. If not, find the Julia executable, and simply double click on it! This should open a terminal session, and display something like this:
_
_ _ _(_)_ | Documentation: https://docs.julialang.org
(_) | (_) (_) |
_ _ _| |_ __ _ | Type "?" for help, "]?" for Pkg help.
| | | | | | |/ _` | |
| | |_| | | | (_| | | Version 1.1.1 (2019-05-16)
_/ |\__'_|_|_|\__'_| | Official https://julialang.org/ release
|__/ |
julia>
This is called the Julia REPL (for Read Eval Loop Print) and is used for interactive use of Julia. Great job, Julia is now running on your computer! Congratulations!
If you want to learn more about Julia, you can read the documentation, there is a Discourse forum, and there is a Slack channel if you need help. But for now you should not need any of those: The notebook will just require you to press Shift + Enter a couple of times.
2. Julia packages required for the notebooks
In Julia, you can access the package manager by simply typing ]
in the REPL. Once you type ]
, the REPL changes to
(v1.1) pkg>
This means you're in the package-manager (or pkg
) mode.
Note that you can exit the pkg
mode by pressing the delete
key, and this will revert the Julia prompt to its original form:
julia>
The packages you should install are:
AIBECS
To create a global steady-state biogeochemistry model, we will be using the AIBECS package. You install it, via
add AIBECS
in pkg
mode, which should look like
(v1.1) pkg> add AIBECS
This should only take a few seconds.
Cartopy
In order to plot things, i.e., to look at the output of the beautiful work you will be doing with AIBECS, you will need a plotting package. For good-looking maps, I suggest using Python's Cartopy because, well, it looks pretty. A few steps are necessary to install it:
Install the Conda package.
In pkg
mode, do
(v1.1) pkg> add Conda
Install Cartopy using Conda (within Julia).
In REPL mode (i.e., not in pkg
mode โ press delete
to get there from pkg
mode), do
julia> using Conda; Conda.add("Cartopy")
Install PyPlot via
(v1.1) pkg> add PyPlot
Intall PyCall via
(v1.1) pkg> add PyCall
This should only take a few seconds as well.
You may want to install Cartopy differently, or even use a different plotting package. This is merely a suggestion that has worked well for me.
IJulia
For running the notebooks, I suggest using the IJulia package. Specifically, IJulia allows you to launch JupyterLab from Julia. To install it, in pkg
mode, type add IJulia
(and press return), and you should see something like:
(v1.1) pkg> add IJulia
This should only take a few seconds as well.
IJulia is recommended but not mandatory to run the notebooks. You could install JupyterLab outside of Julia and use that instead.
You may need to build CodecZlib to run the notebooks. If you see an error mentioning you should build it, then build it! It's easy, just go in pkg
mode and type build CodecZlib
. It should look like this:
(v1.1) pkg> build CodecZlib
If you followed all these steps you should be able to use the notebooks!
The final step is to start JupyterLab. First, make sure you are in the normal Julia REPL mode (i.e., press delete
if you are in pkg
mode.) Then, tell Julia that you want to "use" IJulia:
julia> using IJulia
You can just copy paste the code above (including the julia>
bits), and the REPL will know to not paste those automatically. Everytime a package is used for the first time, Julia will precompile it (which can take a few seconds to minutes, depending on the package โ don't worry, just let it finish).
Finally, you can start JupyterLab from Julia by simply typing jupyerlab()
in Julia. It should look like:
julia> jupyterlab()
If Julia asks you if you want Conda to install JupyterLab, just say "yes" (i.e., type y
). After a couple seconds/minutes of downloads and installations, you should be all set up and a browser window should open with JupyterLab!
Just navigate to the notebook of your choice with JupyterLab in your browser and double-click on the notebook!