PySpace 0.1.0

Documentation

The documentation for this project can be found at http://pyspace.readthedocs.org/.

Features

• A python interface for high performance C++ implementation of N-body simulation algorithms.

• PySpace has a numpy friendly API which makes it easier to use.

• Parallel support using OpenMP.

• GPU support using CUDA

• Dumps vtk output which allows users to take advantage of tools like ParaView, MayaVi, etc. for visualization.

Algorithms

• Brute Force \(O(n^2)\)

• Barnes-Hut \(O(nlogn)\)

Installation

$ pip install pyspace

$ git clone https://github.com/adityapb/pyspace.git

$ python setup.py install

$ export USE_OPENMP=0
$ python setup.py install

$ export USE_CUDA=0
$ python setup.py install

$ export CUDAHOME=/usr/local/cuda

Running the tests

For running the tests you will need to install nose, install using:

$ pip install nose

To run the tests, from project’s root directory run:

$ make test

Running the benchmarks

For running benchmarks you will need to install pandas, install using:

$ pip install pandas

To run the benchmarks, cd to benchmarks directory and run:

$ python run_benchmarks.py

Contributing

Use PEP 8 coding standard for python and follow this for C++.