Scalable electronic structure theory: the embedded electron-pair approximation
Lead Research Organisation:
University of Bristol
Department Name: Chemistry
Abstract
The theory of quantum mechanics provides the means to calculate the structure of molecules, and how molecules will behave. The calculations are complicated, partly because a molecule has many interacting components, and partly because of the intrinsic complications of quantum mechanics itself. Exact quantum mechanical results can be obtained for the simplest of systems, but for real problems, approximations are needed. The field that produces these approximations, then converts them into usable software tools is molecular electronic structure theory.
It turns out that the most highly cited papers in chemistry describe breakthroughs in molecular electronic structure theory. The reason is that these methods can be applied universally: they can inform us about the structure and reactivity of any molecule, so they are used by an enormous range of chemists.
Currently two approximations dominate the field, density functional theory (DFT) and coupled cluster theory (CC). The first is very efficient (ie runs quickly on computers) and the second is amazingly accurate for many problems. There has been a great deal of progress in making DFT more accurate, and CC theory more efficient; our group has been involved in some of these efforts.
In this proposal we set out a new branch of molecular electronic structure theory, based on the concept of treating the electrons one pair at a time, but with each pair embedded in a model provided by the rest of the molecule. These methods could be revolutionary, because their cost appears not much greater than that of DFT, but their accuracy could be competitive with CC theory.
Now is the right time for this research partly because of the demand for better theoretical methods; and partly because recent breakthroughs in quantum embedding theory give a remarkable opportunity to build new and potentially amazing electronic structure methods.
It turns out that the most highly cited papers in chemistry describe breakthroughs in molecular electronic structure theory. The reason is that these methods can be applied universally: they can inform us about the structure and reactivity of any molecule, so they are used by an enormous range of chemists.
Currently two approximations dominate the field, density functional theory (DFT) and coupled cluster theory (CC). The first is very efficient (ie runs quickly on computers) and the second is amazingly accurate for many problems. There has been a great deal of progress in making DFT more accurate, and CC theory more efficient; our group has been involved in some of these efforts.
In this proposal we set out a new branch of molecular electronic structure theory, based on the concept of treating the electrons one pair at a time, but with each pair embedded in a model provided by the rest of the molecule. These methods could be revolutionary, because their cost appears not much greater than that of DFT, but their accuracy could be competitive with CC theory.
Now is the right time for this research partly because of the demand for better theoretical methods; and partly because recent breakthroughs in quantum embedding theory give a remarkable opportunity to build new and potentially amazing electronic structure methods.
Planned Impact
Computers are increasingly used in all branches of science, both in Universities and in industry. In chemistry, computer modelling is used to predict structure, reactivity stability of new compounds. The design of materials and pharmaceuticals is improving all the time through the application of computational models. To remain competitive the UK needs world-leading computer facilities, but also, and crucially, world-leading theoretical models and software to run on them. This project is in the field of molecular electronic structure theory, an area that has produced an extraordinary impact throughout the chemical sciences in the past few decades. The new methods that will be investigated could revolutionize modelling of structure and reactivity in a very wide range of contexts, and the opportunity to deliver this impact will be maximized through development of robust and efficient software, and through application to chemical challenges that are otherwise intractable. All three phases - method development, implementation and application - will take place in collaboration with world leaders in the US, UK and Germany, again maximizing the opportunities to deliver lasting impact through revolutionary ideas.
Organisations
People |
ORCID iD |
Fred Manby (Principal Investigator) |
Publications
Bennie SJ
(2015)
Accelerating wavefunction in density-functional-theory embedding by truncating the active basis set.
in The Journal of chemical physics
Bennie SJ
(2016)
A Projector-Embedding Approach for Multiscale Coupled-Cluster Calculations Applied to Citrate Synthase.
in Journal of chemical theory and computation
Lee S
(2017)
Density-based errors in mixed-basis mean-field electronic structure, with implications for embedding and QM/MM methods
in Chemical Physics Letters
Pennifold RC
(2017)
Correcting density-driven errors in projection-based embedding.
in The Journal of chemical physics
Stella M
(2015)
Computational study of adsorption of cobalt on benzene and coronene
in Molecular Physics
Zhang X
(2018)
Multiscale analysis of enantioselectivity in enzyme-catalysed 'lethal synthesis' using projector-based embedding.
in Royal Society open science
Description | As a result of research conducted by Dr Simon Bennie, funded through this grant, we have been able to improve our theoretical methods for multiscale modelling of chemical reactions in complex environments. This has already being applied to challenges in spintronics and enzymology, and many future applications of this modelling technique can be expected in the future. |
Exploitation Route | The findings are already implemented in the Molpro software package, and will therefore be available to hundreds of research groups around the world to help in accurately modelling chemical processes that take place in complex environments. |
Sectors | Chemicals,Energy,Pharmaceuticals and Medical Biotechnology |
URL | http://www.molpro.net |
Description | Key findings from this research were implemented in the Molpro software package, which is used in hundreds of research groups around the world, both in academic science and in industry. Molpro (molpro.net) is a commercial software package and revenue has been raised for further investment in science. In a separate software project, we have incorporated methods developed through this research in the Entos software package (www.entos.info) which we are in the process of commercializing through a US-based spin out. This software package is specifically targeted at industrial application in the broad area of molecular design and optimization. |
First Year Of Impact | 2016 |
Sector | Chemicals,Energy,Pharmaceuticals and Medical Biotechnology |
Impact Types | Economic |
Title | Absorption spectroscopy at the ultimate quantum limit from single-photon states |
Description | Underlying data for correlated photon experiments used to demonstrate sub shot noise absorption spectroscopy measurements |
Type Of Material | Database/Collection of data |
Year Produced | 2017 |
Provided To Others? | Yes |
Title | Optical implementation of spin squeezing |
Description | Underlying data for correlated photon experiments used to demonstrate an optical implementation of spin squeezing |
Type Of Material | Database/Collection of data |
Year Produced | 2017 |
Provided To Others? | Yes |
Title | Entos Qcore software package |
Description | A platform for molecular design and optimization. Currently being commercialized through a US-based spin out (Entos Inc). |
Type Of Technology | Software |
Year Produced | 2019 |
Impact | The software is used within chemicals and pharmaceuticals companies, and we are building a commercial entity to help drive further development and sales into a range of industrial end-user settings. |
URL | https://entos.ai/qcore |
Title | Software implementation in the Molpro quantum chemistry package |
Description | Molpro is a comprehensive system of ab initio programs for advanced molecular electronic structure calculations, designed and maintained by H.-J. Werner and P. J. Knowles, and containing contributions from many other authors. It comprises efficient and well parallelized programs for standard computational chemistry applications, such as DFT with a large choice of functionals, as well as state-of-the art high-level coupled-cluster and multi-reference wave function methods. |
Type Of Technology | Software |
Year Produced | 2017 |
Impact | Developments in multiscale embedding methodologies were implemented in Molpro, and will be included in the major new release of the software, expected summer 2018. This places key advances at the hands of researchers in around 600 institutions. |
URL | http://www.molpro.net |