Accurate Methods for Large Molecular Systems

Abstract

Three exciting new methods that address the accurate prediction of processes and properties of large molecular systems are discussed. The systematic fragmentation method (SFM) and the fragment molecular orbital (FMO) method both decompose a large molecular system (e.g., protein, liquid, zeolite) into small subunits (fragments) in very different ways that are designed to both retain the high accuracy of the chosen quantum mechanical level of theory while greatly reducing the demands on computational time and resources. Each of these methods is inherently scalable and is therefore eminently capable of taking advantage of massively parallel computer hardware while retaining the accuracy of the corresponding electronic structure method from which it is derived. The effective fragment potential (EFP) method is a sophisticated approach for the prediction of nonbonded and intermolecular interactions. Therefore, the EFP method provides a way to further reduce the computational effort while retaining accuracy by treating the far-field interactions in place of the full electronic structure method. The performance of the methods is demonstrated using applications to several systems, including benzene dimer, small organic species, pieces of the α helix, water, and ionic liquids. RightsWorks produced by employees of the U.S. Government as part of their official duties are not copyrighted within the U.S. The content of this document is not copyrighted. ReceiVed: December 31, 2008; ReVised Manuscript ReceiVed: February 7, 2009 Three exciting new methods that address the accurate prediction of processes and properties of large molecular systems are discussed. The systematic fragmentation method (SFM) and the fragment molecular orbital (FMO) method both decompose a large molecular system (e.g., protein, liquid, zeolite) into small subunits (fragments) in very different ways that are designed to both retain the high accuracy of the chosen quantum mechanical level of theory while greatly reducing the demands on computational time and resources. Each of these methods is inherently scalable and is therefore eminently capable of taking advantage of massively parallel computer hardware while retaining the accuracy of the corresponding electronic structure method from which it is derived. The effective fragment potential (EFP) method is a sophisticated approach for the prediction of nonbonded and intermolecular interactions. Therefore, the EFP method provides a way to further reduce the computational effort while retaining accuracy by treating the far-field interactions in place of the full electronic structure method. The performance of the methods is demonstrated using applications to several systems, including benzene dimer, small organic species, pieces of the R helix, water, and ionic liquids. Authors

Intramolecular interactions of L‐phenylalanine: Valence ionization spectra and orbital momentum distributions of its fragment molecules

“…Larger molecules such as phenylalanine can be simplified when they are reduced into smaller fragments or groups of structurally related molecules and their intramolecular interactions. 4,5 Several excellent reviews discussed a range of aromatic interactions in chemical and biological systems. [5][6][7] These interactions include aiding the stability of DNA base pair stacking and determining molecular recognition in drug docking.…”

Electronic structure calculations in arbitrary electrostatic environments

“…In this case, the system generates a negative potential on the boundary. Hence, the boundary condition (22) implies that the anion must induce a distribution of positive image charge in the environment such that the total potential vanishes on the cavity surface. Naturally, this positive charge stabilizes the negative charge of the anion.…”

“…18 The idea of partitioning a large system into subsystems to aid computation or understanding stretches back to the early works of Löwdin 19 and McWeeny, 20 where they introduced a rigorous mathematical framework for separating (localized) degrees of freedom in large-scale electronic structure calculations. Recently, there has been considerable interest in a practical, simplified version of these ideas known as the fragment molecular orbital (FMO) method [21][22][23][24] which starts with the assumption that there are localized groups of electrons which interact with the other groups (to first order) only classically. A cluster expansion is used to correct for the many-body interaction energies after the fragment wavefunctions have converged.…”

The role of the exchange in the embedding electrostatic potential for the fragment molecular orbital method

“…Recently, a large number of fragment-based methods [1][2][3][4][5][6][7][8][9][10] have been suggested in order to facilitate computing large systems with quantum-mechanical methods. The fragment molecular orbital ͑FMO͒ [11][12][13] method is based on dividing the system into pieces ͑fragments͒ and performing ab initio calculations on fragments, their pairs, and, optionally, triples.…”

“…A number of wave functions have been interfaced with FMO, [14][15][16][17][18][19] an interfragment interaction energy analysis has been developed, 20 and geometry optimizations 21 and molecular dynamics 22 can be performed; FMO has been applied to a variety of systems. 10,12 FMO from the very beginning has had the effect of the environment included in all fragment calculations by the means of the electrostatic potential ͑ESP͒ describing the environment ͑the embedding potential͒, which in FMO consists of the one and two electron parts, describing the nucleuselectron and electron-electron Coulomb interactions, respectively. It has also been the distinct general trend in the development of many fragment methods to achieve an improvement by including the effect of the environment in the fragment calculations, usually in the form of the oneelectron Coulomb terms ͑i.e., point charges͒.…”