This is a rewrite of the General Estuarine Transport Model (GETM). It is mostly written in Python; only performance-critical sections of the code are implemented in Fortran.
You will need Anaconda. On many systems
that is already installed: try running conda --version
. If that fails, you may need
to load an anaconda module first: try module load anaconda
or module load anaconda3
.
If that still does not give you a working conda
command, you may want to install
Miniconda.
Before using conda for the very first time, you will need to initialize its environment:
conda init bash
If you are using a different shell than bash, replace bash
with the name of your shell
(see conda init -h
for supported ones), or use conda init --all
.
This needs to be done just once, as it modifies your .bashrc
that is sourced every time
you login. After this, restart your shell by logging out and back in.
To install or update pygetm:
conda install pygetm -c bolding-bruggeman -c conda-forge
If you need a customized version of pygetm, for instance, built with specific compiler options, or with specific biogeochemical models that are not part of the standard FABM distribution, you can manually obtain the pygetm source code, build it, and then install it.
To obtain the repository with setups and scripts, set up your conda environment, and build and install pygetm:
git clone --recursive https://github.com/BoldingBruggeman/getm-rewrite.git
cd getm-rewrite
conda env create -f environment.yml
conda activate pygetm
source ./install
If you are using a different shell than bash, you may need to replace source
in the
last line by bash
. If you are installing on an HPC system that already has a Fortran
compiler and MPI libraries that you would like to use, replace environment.yml
with
environment-min.yml
in the above.
You can customize the build step as follows:
FC
to your desired Fortran
compiler executable. For instance, in a bash shell: export FC=ifort
. This can be done
in your terminal before running the install script, or by adding it to the install
script itself.FFLAGS
to your desired flags.
For instance, in a bash shell: export FFLAGS=-fcheck=all
. This can be done in your
terminal before running the install script, or by adding it to the install script
itself.-DFABM_EXTRA_INSTITUTES
or
-DFABM_<INSTITUTE>_BASE
, add cmake_opts=<CMAKE_OPTIONS>
to python/setup.cfg
As on other platforms, you need Anaconda or Miniconda. In addition, you need to ensure that software to obtain and build Fortran code is available. Therefore, install:
Now obtain the repository with setups and scripts, set up your conda environment, and build and install pygetm:
git clone --recursive https://github.com/BoldingBruggeman/getm-rewrite.git
cd getm-rewrite
conda env create -f environment-min.yml
conda activate pygetm
install.bat
To update this repository including its submodules (GOTM, FABM, etc.), make sure you are in the getm-rewrite directory and execute:
git pull
git submodule update --init --recursive
conda env update -f <ENVIRONMENT_YML>
conda activate pygetm
source ./install
In the above, replace <ENVIRONMENT_YML>
with the name of the environment file you used
previously: environment.yml
for stand-alone conda environments, or environment-min.yml
for a setup that uses the local MPI implementation and Fortran compiler.
You should always activate the correct Python environment before you use the model with
conda activate pygetm
. This needs to be done any time you start a new shell.
The best place to start is the python/examples
directory with Jupyter Notebooks that demonstrate the functionality of the model:
cd python/examples
python -m jupyterlab
Some of the original GETM test cases have been ported to pygetm:
To run a simulation:
python <RUNSCRIPT.py> [OPTIONS]
To run in parallel:
mpiexec -n <NCPUS> python <RUNSCRIPT.py> [OPTIONS]
How to contribute to the development:
Note that all communication in relation to development of GETM is done via GitHub using issues.
(*) If you use a service other than GitHub for your daily work - please have a look here
https://yarchive.net/comp/linux/collective_work_copyright.html