The SHARK integral generation and digestion system.

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Theoretical Chemistry

Background:

• Molecular integrals are fundamental in quantum chemistry calculations.
• Existing methods for integral generation can be computationally intensive.
• Efficient algorithms are crucial for advancing computational chemistry capabilities.

Purpose of the Study:

• To introduce and describe the SHARK integral generation and digestion engine.
• To present an efficient algorithm for molecular integral computation.
• To enhance quantum chemical program development workflows.

Main Methods:

• Reformulation of the McMurchie/Davidson approach for molecular integrals.
• Algorithm driven by Basic Linear Algebra Subprograms (BLAS) level 3 operations.
• Support for high angular momentum basis functions and various contracted basis sets.

Main Results:

• SHARK demonstrates high efficiency, particularly for high angular momentum basis functions.
• The engine simplifies quantum chemical program development and reduces code duplication.
• SHARK successfully generates diverse types of one- and two-electron integrals, including GIAOs and relativistic integrals.
• Timings show competitive performance against the Libint library on modern processors.

Conclusions:

• SHARK provides a significant advancement in molecular integral generation for quantum chemistry.
• Its BLAS-based approach ensures performance on current and future hardware.
• SHARK is a key component of the ORCA package, facilitating complex quantum chemical calculations.