Parallel Particle Swarm Optimizer for CUDA/C Mex Models

A GPU-parallelized particle swarm optimizer that runs from MATLAB. Particles are launched simultaneously from a CUDA/C mex file that is called from the provided MATLAB file. This Matlab/CUDA framework allows users to easily implement, optimize and visualize their CUDA/C models.
This framework's two components in more detail:

1) A GPU-parallelized particle swarm optimizer in Matlab that is based off a research-verified particle swarm variant (*). The optimizer also allows the user to test all different types of CUDA launch bounds straight from Matlab for easy profiling and meta optimization.
2) The CUDA/C mex file that interfaces the with Matlab. It contains a wrapper kernel with the cuRand random number generation and data reduction routines already in place. It also contains a main function with all the mex-specific-code and gpu error checking routines in place. This allows easy integration with the users' specific model and fast debugging.
Enjoy the visualization and comfort of Matlab while getting the speed of CUDA/C!
See the full description at http://parallelpso.blogspot.com/
(*) Eberhart, R., Shi, Y.: Comparing inertia weights and constriction factors in particle swarm optimization. In: Proceedings of the 2000 IEEE Congress on Evolutionary Computation, Piscataway, NJ, IEEE Press (2000) 84–88