A large number of scientific applications use particle interactions (e.g. Molecular Dynamics, Monte Carlo and multiscale simulations in life sciences or materials). However, as computers have become more specialized, many codes have not been accelerated e.g. for GPUs and it is increasingly hard to maintain parallelization efforts.
The goal of the project is to develop a library of cutting-edge performance nonbonded interactions as well as a parallelization framework, based on GROMACS, that can be used by other particle applications. This includes abstracting suitable parts from GROMACS into a library and improve the performance and scalability of this library on cutting edge exascale hardware (including accelerators).
The project is part of the EU-funded PRACE project and will be executed in collaboration with the KTH-based core GROMACS team and partners from CSCS, the Swiss national supercomputing center.
The starting date is open for discussion, though ideally we would like the successful candidate to start as soon as possible.
Candidates need to have a higher academic education (preferably PhD) in computer science, computational science, or in a science area using molecular dynamics simulations.
Proven experiences in developing and working with large scientific codes on large HPC systems. Experiences with molecular dynamics codes are a plus. Excellent knowledge of parallel computing (MPI, OpenMP, GPU, etc.) as well as C++11/14 is required. Experiences in developing APIs (preferably in C, C++, or python) good knowledge of python is an asset. Good knowledge of Linux is required.
Great emphasis will be placed on personal competence and suitability.
The position is based in Stockholm. For more information please see:
Deadline for applications is extended to 31 May 2019.