Praveen Chandrashekar

Centre for Applicable Mathematics, TIFR, Bangalore

[ People | News | Codes | Talks | Teaching | Publications | Calendar | Hiking | Contact ]

Clawpack

Install using script

I have made a shell script which installs clawpack from source and uses conda to get python packages.

wget https://raw.githubusercontent.com/cpraveen/cfdlab/master/bin/clawpack.sh
export CLAW=/path/to/clawpack   # where you want to install clawpack sources
bash ./clawpack.sh v5.10.0

Install from source

See here for installation steps

git clone git@github.com:clawpack/clawpack.git
cd clawpack
git tag -l                    # see available versions
git checkout v5.10.0          # Checkout version you want
git submodule init            # for repositories pyclaw, clawutil, visclaw, etc.
git submodule update          # clones all the submodule repositories
export CLAW=/path/to/clawpack # in your shell config

If you want to compile python/pyclaw support do this

conda install ipython jupyterlab matplotlib meson-python ninja nose numpy scipy seaborn six
conda install petsc4py    # if you want parallel support.
                          # Add flah "-c conda-forge" if not using miniforge.
cd $CLAW
pip install --user --no-build-isolation -e .   # note trailing dot indicating "this directory"

This should install some files in your $HOME/.local/lib/python#.#/site-packages directory.

Recommended: You can also do this inside a conda environment in order to not mess with your base environment, see below for how to do this.

Test a fortran example

The following command does all the steps: compile, run and make plots

cd $CLAW/classic/examples/acoustics_1d_example1
make .plots

Open _plots/_PlotIndex.html file to see the results.

Make options:

make .exe      # compile
make .data     # generate claw.data file
make .output   # compile and run
make .plots    # compile, run and make plots

Install using conda

This does not seem to give the classic fortran version. But it will install pyclaw including parallel version which needs PETSc.

unset CLAW PYTHONPATH   # These might interfere if set to something already.
conda create -n claw
conda activate claw
conda install -c conda-forge clawpack ipython jupyterlab nose six seaborn

Test pyclaw: start ipython and

from clawpack.pyclaw import examples
claw = examples.shock_bubble_interaction.setup()
claw.run()
claw.plot()

To go out of this environment

conda deactivate

To delete this environment

conda remove -n claw --all

You can see a list of environments using

conda env list

Riemann solver book

See here github.com/clawpack/riemann_book and install like this (do this inside your claw conda environment)

git clone https://github.com/clawpack/riemann_book
cd riemann_book
pip install -r requirements.txt
jupyter nbextension enable --py widgetsnbextension
pip install jupyter_contrib_nbextensions
jupyter contrib nbextension install --user
jupyter nbextension enable equation-numbering/main

Examples from the book

The original set of examples designed to run with clawpack-4.3 are here. To run these, download Clawpack-4.3 and set CLAW to the path where it is located. Visualizing the results requires matlab.

Some of these have been converted to run with newer version of clawpack, see here and here. You can get the code like this

cd $CLAW  # or where ever you want apps to be
git checkout master
git clone --recursive https://github.com/clawpack/apps

The book examples are in the directory apps/fvmbook.

Documentation