Fast and Flexible Coupled Cluster Implementation.

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• High-Performance Computing

Background:

• Coupled cluster singles and doubles with triples correction (CCSD(T)) is a high-accuracy quantum chemistry method.
• Traditional CCSD(T) implementations face challenges with computational cost and memory requirements.
• Efficient algorithms are crucial for applying accurate methods to larger molecular systems.

Purpose of the Study:

• To present a new, optimized coupled cluster singles and doubles with triples correction (CCSD(T)) algorithm.
• To implement the algorithm in an efficient and flexible C++ codebase.
• To benchmark the performance of the new algorithm on various hardware architectures.

Main Methods:

• Object-oriented C++ implementation of the CCSD(T) algorithm.
• Flexible storage backend supporting distributed memory and file systems.
• Performance evaluation on single workstations, small clusters, and high-end Cray systems.
• GPU acceleration was explored, with modest performance gains.

Main Results:

• The new CCSD(T) algorithm demonstrates low memory footprint, fast execution, and low I/O overhead.
• Complex calculations (hundreds of basis functions, dozens of occupied orbitals) complete in under a day on a single workstation.
• The implementation shows good scalability across different computing environments.
• GPU implementation provides a modest performance improvement.

Conclusions:

• The developed CCSD(T) algorithm significantly enhances computational efficiency for high-accuracy electronic structure calculations.
• This advancement makes advanced quantum chemistry methods more accessible on standard and high-performance computing resources.
• The flexible design facilitates broader application in computational chemistry research.