Quantum Computing based density functionals for fast and accurate materials and chemistry simulations

量子计算机有可能导致可用计算能力的变革性增长。当今可用的量子计算机的大小和质量正在迅速增加，实现量子纠错能力的进步使我们更接近于拥有量子计算机，甚至能够超越现有的最好的经典计算机。量子计算机提供的最有前途的改进之一是能够准确模拟原子尺度的物理系统，这对经典计算机来说是一项具有挑战性的任务。密度泛函理论（DFT）是一种经典的计算方法，适用于制药、化工和材料等行业的产品开发。例如，DFT可用于建模和评估新材料，药物，催化剂等的性质，以补充实验室中昂贵的开发。与其他方法相比，DFT是唯一一种足够快的方法，适用于许多应用中所需的包含金属和分子的大型系统。然而，近似的DFT算法中使用的今天，使他们不适合许多重要的问题。DFT的改进可以导致开发更好的电池，更高效的催化剂，更快，更可靠的药物和疫苗候选评估以及更绿色的化学生产。最近，机器学习（ML）方法已被证明可以改进DFT。然而，即使是ML + DFT方法也依赖于对物理系统的大量精确模拟，这对于经典计算机来说是具有挑战性的。在这个项目中，我们的目标是利用量子计算机的上级能力来执行精确的模拟，以改善DFT。我们的方法将联合收割机提供的量子计算机和ML的优势，以显着改善DFT方法，广泛应用于各个行业。该项目结合了机器学习，量子计算，DFT，和建模在工业环境中的InstaDeep，NPL，Atos UK和约翰逊Matthey的专业知识。

Quantum computers have the potential to lead to a transformative increase in the available computational power. The size and quality of the quantum computers available today are increasing rapidly, and advancements in ability to implement quantum error correction are getting us closer to have quantum computers which will be able to outperform even the best existing classical computers. One of the most promising improvements offered by quantum computers is the ability to accurately simulate atomic scale physical systems, which is a challenging task for classical computers. Various algorithms are being developed today that aim to make use of this advantage that is offered by quantum computers.Density functional theory (DFT) is a core classical computing method for product development across industries such as pharmaceuticals, chemicals, and materials. For instance, DFT can be used to model and evaluate the properties of new materials, drugs, catalysts, etc. to complement expensive developments in the lab. Compared to alternative methods, DFT is the only approach fast enough for large systems comprising metals and molecules as required in many applications. However, approximations made in the DFT algorithms used today make them unsuitable for a lot of important problems. Improvements in DFT can lead to the development of better batteries, more efficient catalysts, faster and more reliable drug and vaccine candidate assessments, and greener chemical production.Recently, machine learning (ML) methods have been demonstrated to lead to improvements in DFT. However, even the ML + DFT methods rely on a large number of accurate simulations of the physical systems, which are challenging to do with classical computers. In this project, we aim to harness the superior ability of quantum computers to perform accurate simulations in order to improve DFT. Our approach will combine the advantages provided by quantum computers and ML in order to significantly improve the DFT method that is widely used in various industries.This project combines the expertise in machine learning, quantum computing, DFT, and modelling in industrial settings of InstaDeep, NPL, Atos UK, and Johnson Matthey.