Massively parallel implementation of 3D-RISM calculation with volumetric 3D-FFT

Area of Science:

• Computational chemistry
• Biophysics
• High-performance computing

Background:

• Traditional parallel 3D-RISM methods are limited by slab-type 3D-FFT, restricting parallelization.
• Analyzing hydration properties of large biomolecular systems requires efficient computational tools.

Purpose of the Study:

• To develop and test a new 3D-RISM program optimized for massively parallel machines.
• To overcome the parallelization limitations of existing 3D-RISM programs using volumetric 3D-FFT.

Main Methods:

• Development of a novel 3D-RISM program incorporating volumetric 3D-FFT.
• Testing the program on the RIKEN K supercomputer with large-scale calculations (2048^3 grid points) across 16,384 nodes.
• Application of the program, coupled with molecular dynamics, to study the oligomerization of chymotrypsin Inhibitor 2 mutant.

Main Results:

• The new 3D-RISM program demonstrated excellent scalability on the RIKEN K supercomputer.
• Volumetric 3D-FFT significantly relieved parallelization limitations.
• The program effectively analyzed the hydration properties of a large biomolecular system.

Conclusions:

• The developed massively parallel 3D-RISM program is highly effective for large-scale biomolecular simulations.
• This advancement enables more efficient analysis of hydration in complex biological systems.
• The combination of 3D-RISM and molecular dynamics provides a powerful approach for studying biomolecular processes.