Fragme∩t: An Open-Source Framework for Multiscale Quantum Chemistry Based on Fragmentation

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Software Development

Background:

• Conventional electronic structure calculations face computational scaling challenges.
• Fragment-based methods offer a solution by partitioning large systems into smaller subsystems.
• Interoperability between quantum chemistry codes is crucial for these methods.

Purpose of the Study:

• Introduce "Fragme∩t", an open-source software for fragment-based quantum chemistry.
• Provide a tool for community validation and development of new approximations.
• Enable analysis of many-body interactions in large molecular systems.

Main Methods:

• Developed algorithms for automatic fragment generation and structure modification.
• Implemented distance- and energy-based screening for subsystem selection.
• Integrated checkpointing, database management, and parallelization using Python, SQLite, and Ray.

Main Results:

• Fragme∩t demonstrates high parallel efficiency (~96%) on over 1,000 processors.
• The software effectively handles large-scale protein fragmentation on workstation hardware.
• Provides interfaces to numerous quantum chemistry engines (Q-Chem, PySCF, xTB, etc.).

Conclusions:

• Fragme∩t addresses the need for specialized software in fragment-based calculations.
• The application promotes community validation and the development of novel approximations.
• Offers a user-friendly framework for complex quantum chemistry computations.