Software kit for cosmological N-body simulations.

The tools are provide as top level executable scripts. All tools are MPI parallel. We can deal with very large simulations.

• power.py is a Power Spectrum calculator.
• fof.py is a friend of friend feature (halo) identifier.
• subhalo.py is a FOF6D sub-feature (subhalo) identifier .
• power-parallel.py schedules and runs a set of power.py jobs, possibly using several independent nodes

Run them with '-h' to see the inline help.

Each takes a few 'datasource' strings that specify the format of the inputs.

There are example scripts (which also act as integrated tests) in examples directory. The supporting data for these scripts can be retrieved from

https://s3-us-west-1.amazonaws.com/nbodykit/nbodykit-data.tar.gz

Check get_data.sh for details.

The software is built on top of existing tools. Please refer to their documentations:

• pfft : massively parallel fast fourier transform, pencil domains
• pfft-python : python binding of pfft
• pypm : particle mesh framework in Python
• kdcount : pair-counting and friend-of-friend clustering with KD-Tree
• MP-sort : massively parallel sorting
• sharedmem : in-node parallelism with fork and copy-on-write.

The software is used in tree. First clone with

git clone http://github.com/bccp/nbodykit
cd nbodykit

Then build with

./build.sh

Note that sometimes 'git submodule update' is needed to sync the subpackages, do it often.

It may take a while to build fftw and pfft.

sudo port install autoconf automake libtool

export LDSHARED="mpicc -bundle -undefined dynamic_lookup -DOMPI_IMPORTS"; ./build.sh

On recent versions of MacPorts, we also need to tell mpicc to use gcc rather than the default clang compiler, which doesn't compile fftw correctly due to lack of openmp support.

export OMPI_CC=gcc

module unload darshan

module unload PrgEnv-intel
module unload PrgEnv-gray
module load PrgEnv-gnu

module load python
module load cython
module load numpy
module load mpi4py

Packages are ready to use after importing the nbodykit namespace.

import nbodykit

print(nbodykit)

Note that actual packages are still under their own namespaces, for example

import kdcount

import pypm

This is to maintain the relative independence of the packages.