Phase-Space Natural Orbital Functional Theory

点击翻译按钮获取中文摘要

Computational chemistry has become an important tool for predicting, confirming, and explaining experimental results in not only all fields of chemistry, but in areas such as molecular biology and nanotechnology. High quality, mathematically rigorous, and computationally expensive calculations provide chemical properties within the errors of experimental measurements, but are currently unfeasible for large systems. Therefore when studying systems such as proteins, surfaces of materials, or bulk properties of molecular ensembles (eg. compounds dissolved in water), it is necessary to employ more approximate computational chemistry methods.The current approximate method of choice for the computational study of large molecular systems is density functional theory (DFT), due to its combination of accuracy and efficiency. Unfortunately, DFT is unable to properly describe bond breaking or formation, as well as transition metals and excited states in general, phenomena that are key to understanding chemistry. Recent research has revealed that the basis of many DFT failures is static correlation. I propose to develop a new computational chemistry method based on natural orbital functional theory (NOFT). A key feature of NOFT is the occupation of the natural orbitals by fractions of electrons which effectively treats the static correlation problem encountered by DFT. We will make use of the successful concepts of DFT, within an NOFT framework, to design a universal and accurate natural orbital functional. We will combine our functional with the latest theoretical advances for calculations on large systems to create a computational chemistry method capable of providing results within experimental accuracy for large and complex molecular systems. The development of such a method will allow computational and experimental researchers to reliably determine molecular structures, reaction pathways, and other properties of interest of systems for which current computational methods are unsuitable.