This project presents a GPU-based tau-leaping version of the Gillespie algorithm. 
Only NVIDIA(tm) video cards are supported now.

How to create a local copy of the repository?
=============================================
git clone git://github.com/CSSL-UWM/tau_leaping.git

You can build the project with the CMake tool:
cd tau_leaping
mkdir build
cd build
cmake ..
make

The CUDA driver and CUDA Toolkit must be installed in the system. The latest version is available at:
http://developer.nvidia.com/cuda-toolkit

to run the simulation, use the following syntax:
tau_leaping -f <model_name> -t <finish_time> -n <number of realizations> -r <list_of_species_to_track_space_separated> 
example:
tau_leaping -f xamples\test_128.xml -t 0.1 -n 10000 -r S1 S5
This command will simulate 10000 realizations with the final time set to 1s and will track reactants with names "S1" and "S5".

The solver is accompanied with a toy system generator (toy_system_gen).
A toy system of arbitrary size can be created with a command:
toy_system_gen -r <number_of_reactants> -c <number_of_channels> [-mnp <maximum_initial_population>] [-mxp <maximum_initial_population>]
example:
toy_system_gen -r 1000 -c 2000 -mxp 10000
This command will produce a system with 1000 reactants and 2000 channels and initial population of species varying from 0 to 10000.

Two special types of systems, that are weakly and strongly coupled ones, can be produced.
Weakly coupled systems (Cyclic Chain Model):
toy_system_gen -w -r <number_of_reactants> [-mnp <maximum_initial_population>] [-mxp <maximum_initial_population>]
Strongly coupled systems (Colloidal Aggregation Model):
toy_system_gen -s -r <number_of_reactants> [-mnp <maximum_initial_population>] [-mxp <maximum_initial_population>]
example:
toy_system_gen -s  -r 250 -mxp 1000 -mxp 2000
This command will produce a strongly-coupled system with 250 reactants and 31250 channels and initial population of species varying from 1000 to 2000.