Simulation of the gravitational wave memory signals for LISA and ground-based detectors (LIGO-Virgo, LIGO Voyager, Einstein Telescope, Cosmic Explorer). Model selection and parameter estimation.
Based on the paper:
Boris Goncharov, Laura Donnay, Jan Harms (2024, PRL)
The first step is to build a singularity/apptainer container image. The container definition file is here: /.def
. To build a container, run ``. Next, open container image: singularity shell --bind "/some_new_directory_on_your_machine/:$HOME" gwmem_2022_20230321_3.sif
. A few more steps are still needed. First, move surrogate waveforms to `~/gwsurrogate_downloads` in the container. Second, this needs to be performed when inside the container:
pip install ChainConsumer --user
git clone [email protected]:bvgoncharov/GWFish.git
cd GWFish
git checkout development_bg
python -m pip install --user .
Note, python setup.py install
did not work for me for GWFish, the package was visible but not its modules.
Now, you can git clone this repository to /some_new_directory_on_your_machine/
, the code should work with singularity exec
. For example:
singularity exec --bind "/some_new_directory_on_your_machine/:$HOME" gwmem_2022_20230321_3.sif python /home/bgonchar/gwmem_2022/gwfish_analysis/memory_strain_amplitudes.py --outdir "/home/bgonchar/out_gwmem_2022/" --injfile "/home/bgonchar/pops_gwmem_2022/pop_max_o3_bbh_only_1yr_20230606.hdf5" --waveform "NRHybSur3dq8" --waveform_class "gu.LALTD_SPH_Memory" --det "ET" --config "/home/bgonchar/gwmem_2022/gwfish_analysis/detectors/gwfish_detectors_et_10_1024Hz.yaml" --fisher_pars "ra,dec,psi,theta_jn,luminosity_distance,mass_1,mass_2,a_1,a_2,geocent_time,phase,J_E,J_J" --td_fmin 9. --f_ref 20. --mem_sim "J_E,J_J" --j_e 1.0 --j_j 1.0 --label "20230606_m" --num 100 --inj 0
Make sure to adjust paths, and check out other examples in gwfish_analysis/slurm/
.
An alternative approach: setting up conda environment
Please follow these steps. Create a new conda virtual environment like this:
conda create -n gwmem_sxs -y -c conda-forge python=3.9
conda activate gwmem_sxs
conda install -c conda-forge gwsurrogate
conda install -c conda-forge sxs
conda install -c conda-forge matplotlib
conda install -c conda-forge bilby
pip install git+ssh://[email protected]/ColmTalbot/gwmemory.git
pip install git+ssh://[email protected]/jblackma/NRSur7dq2.git
Additionally, I work with my fork of GWFish, locally-installed with git clone
and python setup.py develop
.
Optional, useful: conda install -c conda-forge ipdb
Optional, for gwmemory
notebook: conda install -c conda-forge basemap
Optional, for plotting Fisher matrix errors: pip install git+ssh://[email protected]/bvgoncharov/normal_corner.git
In development, gwmemory of Colm was required for jupyter notebook, gwmemory of Moritz required to test waveforms:
memestr
gwmemory/gwmemory_mh
After installation, deactivate the environment and activate it again:
conda deactivate
conda activate gwmem_sxs
This is necessary for LAL packages, otherwise they might not be visible.
This also helps in case of the following error: ModuleNotFoundError: No module named 'lal'
.
Link the conda environment to Jupyter notebook:
conda install -c anaconda ipykernel
python -m ipykernel install --user --name=gwmem_sxs
Limitations:
- Bilby will not work with NRSur7dq4 in Jupyter notebook, only in the command line.
Sometimes the ipython kernel will not see modules, and needs to be reinstalled:
jupyter kernelspec uninstall gwmem_sxs
.
To delete the environment (after it is deactivated): conda env remove -n gwmem_sxs