A new approach for second-order perturbation theory.

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Theoretical Chemistry

Background:

• Second-order perturbation theory (MP2) is crucial for accurate electronic structure calculations.
• Existing parallel MP2 implementations can be memory-intensive and computationally expensive.

Purpose of the Study:

• To develop a novel, highly efficient MP2 algorithm for closed-shell energy evaluations.
• To reduce computational resource requirements (memory, FLOPs, time) compared to existing methods.

Main Methods:

• Developed a new MP2 algorithm with a reduced memory footprint and FLOP count.
• Implemented an adaptive strategy for storing MP2 amplitudes across disk and distributed memory.
• Adapted the algorithm for graphical processing unit (GPU) architecture.

Main Results:

• The new MP2 approach demonstrates significantly lower memory usage, FLOP count, and time to solution.
• The algorithm shows excellent scalability on single workstations, small clusters, and large supercomputing systems.
• Large-scale calculations with thousands of basis functions are achievable in hours on a single workstation.
• The GPU-adapted algorithm shows promising performance.

Conclusions:

• The presented MP2 algorithm offers a substantial improvement in efficiency and performance for electronic structure calculations.
• This method enables larger and more complex quantum chemistry computations on accessible hardware.
• The adaptive memory strategy and GPU adaptation enhance the practicality and applicability of MP2 theory.