GPU stiff integrator paper published

Our paper describing a study of GPU-based integration algorithms for chemical kinetics, and the associated accelerInt software package has been published in Combustion and Flame:

• Journal version: https://doi.org/10.1016/j.combustflame.2017.02.005
• arXiv eprint: arXiv:1607.03884 [physics.comp-ph]

The differential equations governing the time change of chemical kinetics are typically stiff, meaning they present challenges to solve that require specialized algorithms. accelerInt is an open-source library of ordinary differential equation solvers, tested and verified for these applications. This paper focuses on three particular algorithms, comparing and contrasting them both with each other and a representative algorithm used commonly in computer simulations of combustion. We showed that, under some conditions, one method (implicit, fifth-order Runge-Kutta) can perform 12-38 times faster on a graphics processing unit compared with a standard method on a central processing unit.