Guo Sikao, Korolija Nenad, Milfeld Kent, Jhaveri Adip, Sang Mankun, Ying Yue Moon, Johnson Margaret E
TC Jenkins Department of Biophysics, Johns Hopkins University, Baltimore, Maryland, USA.
University of Belgrade, Belgrade, Serbia.
J Comput Chem. 2025 May 30;46(14):e70132. doi: 10.1002/jcc.70132.
Particle-based reaction-diffusion offers a high-resolution alternative to the continuum reaction-diffusion approach, capturing the volume-excluding nature of molecules undergoing stochastic dynamics. This is essential for simulating self-assembly into higher-order structures like filaments, lattices, or macromolecular complexes. Applications of self-assembly are ubiquitous in chemistry, biology, and materials science, but these higher-resolution methods increase computational cost. Here, we present a parallel implementation of the particle-based NERDSS software using the message passing interface (MPI), achieving close to linear scaling for up to 96 processors. By using a spatial decomposition of the system across processors, our approach extends to very large simulation volumes. The scalability of parallel NERDSS is evaluated for reversible reactions and several examples of higher-order self-assembly in 3D and 2D, with all test cases producing accurate solutions. Parallel efficiency depends on the system size, timescales, and reaction network, showing optimal scaling for smaller assemblies with slower timescales. We provide parallel NERDSS code open-source, supporting development and extension to other particle-based reaction-diffusion software.
基于粒子的反应扩散为连续反应扩散方法提供了一种高分辨率替代方案，能够捕捉经历随机动力学的分子的体积排斥特性。这对于模拟自组装成诸如细丝、晶格或大分子复合物等高阶结构至关重要。自组装在化学、生物学和材料科学中无处不在，但这些高分辨率方法会增加计算成本。在此，我们使用消息传递接口（MPI）展示了基于粒子的NERDSS软件的并行实现，在多达96个处理器上实现了接近线性的扩展。通过在处理器之间对系统进行空间分解，我们的方法可扩展到非常大的模拟体积。针对可逆反应以及3D和2D中的几个高阶自组装示例评估了并行NERDSS的可扩展性，所有测试用例均产生了准确的解。并行效率取决于系统大小、时间尺度和反应网络，对于具有较慢时间尺度的较小组件显示出最佳扩展。我们提供并行NERDSS代码的开源版本，以支持开发并扩展到其他基于粒子的反应扩散软件。
