pycola3

Author: Svetlin Tassev

Ported to Python 3 by Phil Bull

Documentation (old version): PDF Documentation

Introduction

pycola3 is a multithreaded Python/Cython N-body code, implementing the Comoving Lagrangian Acceleration (COLA) method in the temporal and spatial domains. pycola3 also implements a novel method to compute second-order cosmological initial conditions for given initial conditions at first-order for arbitrary initial particle configurations (including glass initial conditions, as well as initial conditions having refined subregions).

pycola3 is based on the following two papers: temporalCOLA and spatialCOLA. We kindly ask you to acknowledge them and their authors in any program or publication in which you use the COLA method in the temporal and/or spatial domains. (We do not require you to, however, as we want pyCOLA to be GPLv3 compatible.)

Footnotes

pycola3 requires NumPy, SciPy, pyFFTW, h5py, as well as their respective dependencies.

All lengthscales are in units of comoving Mpc/h, unless otherwise specified.

Worked-out example

The following example is contained in examples/example.py. It reads in a MUSIC-generated initial conditions at first order; calculates the second-order initial displacements on the full box; then runs COLA and finally outputs a figure containing a density slice. The script needs matplotlib to be installed.

To run the script, first generate MUSIC initial conditions with the included configuration file:

MUSIC ics.conf

Then change the variable music_file below to point to the MUSIC snapshot, and then execute the example script by issuing:

python ./example.py

This example script was used with minor modifications in making the figures for the paper. It fits comfortably on 24GB RAM. If that is not available, decreasing gridscale to 1, reduces RAM consumption to 11GB at the cost of reducing force resolution and producing artifacts.

Most of the time, the script spends on calculating the second-order displacement field for the whole box, i.e. not on the COLA calculation. On the 4-core laptop that the calculations for the paper were performed on, the COLA calculation itself takes about 40 seconds.

The script produces the following figure:

References

[MUSIC]: Multi-scale initial conditions for cosmological simulations, O. Hahn, T. Abel, Monthly Notices of the Royal Astronomical Society, 415, 2101 (2011), arXiv:1103.6031. The code can be found on this website.

[spatialCOLA]: Extending the N-body Comoving Lagrangian Acceleration Method to the Spatial Domain, S. Tassev, D. J. Eisenstein, B. D. Wandelt, M. Zaldarriaga, (2015), arXiv:1502.07751

[temporalCOLA]: Solving Large Scale Structure in Ten Easy Steps with COLA, S. Tassev, M. Zaldarriaga, D. J. Eisenstein, Journal of Cosmology and Astroparticle Physics, 06, 036 (2013), arXiv:1301.0322