SaclayMocks

Simulated Absorption for Cosmology with Lyman-Alpha from the Yvette Mocks

Installation

to download, you can simply use git clone:

git clone https://github.com/igmhub/SaclayMocks.git

then, you need to add to your bashrc:

export SACLAYMOCKS_BASE=<path_to_git_repository>
export PATH=$SACLAYMOCKS_BASE/bin/:$PATH
export PYTHONPATH=$SACLAYMOCKS_BASE/py/:$PYTHONPATH

Dependencies

The code is compatible both with python2 and python3. Here is a list of needed packages:

• scipy
• healpy
• fitsio
• h5py
• pyfftw
• numba
• iminuit
• linetools
• pyigm

The pyigm package can not be directly installed by setup.py. Please install the git version of pyigm, by typing, in a separate folder:

git clone https://github.com/pyigm/pyigm.git cd pyigm/ python setup.py install --user

Running

The code runs as follow: First it reads and interpolate the powerspectrum from camb (stored in etc/) to create a 3D power-spectrum. These steps are done with interpolate_pk.py and merge_pk.py

Then, the Gaussian random field boxes are drawn, with make_boxes.py

The quasars are drawn from the GRF boxes, using draw_qso.py

Then, the spectra are computed for each quasar, using make_spectra.py and merge_spectra.py

All theses steps are done for each of the 7 chunks that make up the desi footprint

Once all the chunks are done, they are combined to produce the final output. This is done by running: merge_qso.py for producing the master.fits and master_randoms.fits files dla_saclay.py and merge_dla.py to produce the master_DLA.fits file dla_randoms.py to produce the master_DLA_randoms.fits file make_transmissions.py to produce all the transmissions files

The logs are organized as follow: For the Pk, the script run_pk.sh has the corresponding run_pk.log file in the pk/logs/ directory. It gives the global result of all the python codes run (interpolate_pk.py + merge_pk.py), each of them has its own log file.

For the boxes, the scripts run_boxes-$i.sh have the corresponding run_boxes-$i.log files ($i stands for the id of the chunk), in the mock_*/output/logs/ directory It gives the global result of the python code run make_boxes.py. Its log file is mock_*/output/logs/chunk_$i/make_boxes.py

For the chunks (QSO + spectra), the scripts run_chunk-$i.sh have the corresponding run_chunk-$i.log files, in the mock_*/output/logs/ directory It gives the global result of all the python codes run (draw_qso.py, make_spectra.py and merge_spectra.py), each of them has its own log file in mock_*/output/logs/chunk_$i/

Finaly, the scripts run_mergechunks.sh has the corresponding run_mergechunks.log file, in the mock_*/output/logs/ directory It gives the global result of all the python codes run, each of them has its own log file in mock_*/output/logs/mergechunks/

You don't need to run all these codes by hand. You just have to run submit_mocks.py This code will produce several bash script that will run all the needed python code, according to the options.

To produce one realisation, just run:

submit_mocks.py --mock-dir <out_path>

and it will produce a bash script submit.sh, that will send all the bash script which themselves send the python scripts to the cori nodes, on NERSC.

If you want to produce a small footprint, on your laptop, you can run:

submit_mocks.py --mock-dir <out_path> --cori-nodes False --box-size 256

When not running on cori nodes, it is advised to not use the nominal chunk size (2560), because it uses a lot of memory and threads. Also, if not distributed on several nodes, it can take hours.

You can also limit the number of chunks by specifying the IDs of chunks to runs:

--chunk-id 1 2

All the informations relative to chunks (box size available, chunk ids, ...) are in the chunk_parameters() function.