Local Møller-Plesset Perturbation Theory:  A Massively Parallel Algorithm

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• High-Performance Computing

Background:

• Accurate electronic structure calculations are crucial for understanding molecular properties.
• Second-order Møller-Plesset perturbation theory (MP2) provides a good balance of accuracy and cost.
• Local approximations (LMP2) reduce the computational scaling of MP2, but efficient parallelization remains a challenge.

Purpose of the Study:

• To develop a massively parallel algorithm for local second-order Møller-Plesset (LMP2) energy computations.
• To achieve linear scaling of computational time and storage with molecular size.
• To design a scalable algorithm suitable for high-performance computing environments.

Main Methods:

• Development of a massively parallel algorithm for LMP2 energy calculations.
• Implementation of a distributed data scheme for two-electron integrals.
• Utilization of sparse data representation and generalized contraction routines.
• Employing distributed sparse multidimensional arrays for efficient computation.

Main Results:

• The algorithm demonstrates linear scaling of both storage requirements and computational time with molecular size.
• The parallel implementation effectively avoids communication bottlenecks.
• High parallel efficiency was achieved using up to 100 processors.
• The method allows for efficient computation of LMP2 energies on large molecular systems.

Conclusions:

• The presented massively parallel algorithm offers an efficient and scalable solution for LMP2 energy calculations.
• The linear scaling properties make it suitable for large molecular systems.
• The developed techniques pave the way for routine application of LMP2 theory in high-performance computing.