Efficient and accurate fragmentation methods.

Area of Science:

• Computational Chemistry
• Electronic Structure Theory

Background:

• Accurate modeling of large molecular systems is computationally demanding.
• Existing methods like Fragment Molecular Orbital (FMO) have limitations regarding capping atoms and empirical parameters.
• Effective Fragment Potential (EFP) offers a first-principles approach to intermolecular interactions.

Purpose of the Study:

• Introduce and discuss three novel fragmentation methods available in the GAMESS program.
• Highlight the capabilities and applications of FMO, EFP, and the combined Effective Fragment Molecular Orbital (EFMO) method.
• Demonstrate the efficacy of these methods for large-scale molecular calculations.

Main Methods:

• Fragment Molecular Orbital (FMO): Scalable, no capping atoms or empirical parameters, suitable for parallel systems.
• Effective Fragment Potential (EFP): First-principles model potential for accurate intermolecular interactions without empirical fitting.
• Effective Fragment Molecular Orbital (EFMO): Merges FMO and EFP, using EFP for interfragment interactions for improved accuracy and efficiency.

Main Results:

• FMO demonstrates near-linear scalability on massive parallel systems for large clusters.
• EFP accurately describes diverse intermolecular interactions, including Coulombic, polarization, and charge transfer.
• EFMO offers enhanced accuracy and computational efficiency over standard FMO, incorporating three-body interactions via EFP.

Conclusions:

• The discussed fragmentation methods (FMO, EFP, EFMO) provide powerful tools for studying large molecular systems.
• EFMO shows particular promise for future applications like geometry optimizations and molecular dynamics simulations upon completion of analytic gradients.
• These methods advance the accurate and efficient computational modeling of complex chemical and biological systems.