Parallelization of particle-based reaction-diffusion simulations using MPI
Author Information
Author(s): Guo Sikao, Korolija Nenad, Milfeld Kent, Jhaveri Adip, Sang Mankun, Ying Yue Moon, Johnson Margaret E
Primary Institution: Johns Hopkins University
Hypothesis
Can the particle-based NERDSS software be effectively parallelized using MPI to improve computational efficiency?
Conclusion
The parallel implementation of NERDSS achieves close to linear scaling for up to 96 processors, significantly enhancing simulation efficiency.
Supporting Evidence
- The parallel NERDSS code is open source and provides detailed documentation for developers.
- Scalability was evaluated for bimolecular reactions in 3D and 2D, showing accurate solutions.
- Parallel efficiency depends on system size, reaction network, and limiting timescales.
Takeaway
This study shows how to make computer simulations of particle interactions faster by using multiple processors at the same time.
Methodology
The study implemented a parallel version of the NERDSS software using MPI and spatial domain decomposition to evaluate its scalability for various reaction-diffusion scenarios.
Limitations
The current implementation restricts assembly sizes to below the spatial decomposition size, which may limit the simulation of larger complexes.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website