Poster presented at the GPU Technology Conference, March 2013, San Jose, CA.
It presents the PyGBe code (pronounced 'pig-bee'), which solves the linearized Poisson-Boltzmann equation using a boundary element method, BEM. The underlying dense systems are solved using a Krylov-subspace method, accelerated with a treecode to achieve O(N log N) complexity.
The code presents a Python environment for the user, while computing all kernels in CUDA (interfaced with PyCuda), for maximum ease of use, combined with high performance on GPUs.
Acknowledgement
This research is made possible by a grant from the Office of Naval Research, Applied Computational Analysis Program (N00014-11-1-0356). LAB also acknowledges support from NSF CAREER award OCI-1149784.