Development and Application of Digital Twin & Real-Space KS-DFT

The concept of a Digital Twin originated in industry as a “digital copy of a physical asset.” We aim to build a virtual laboratory infrastructure to address challenges in data acquisition, control, analysis, and model-driven interpretation, with a focus on X-ray Photon Science and its potential applications in chemical conversions. This approach enables a bidirectional feedback loop between theory and experiment. I will outline our key challenges and milestones, including: 1) developing quantum chemistry methods to improve XPS binding energy (BE) accuracy, 2) recognizing the universality of chemical reaction networks (CRNs) across systems like heterogeneous catalysis, and 3) introducing our user-friendly Digital Twin for Chemical Sciences (DTCS) v.01 software package, which we welcome collaboration and feedback on. The second part of my talk addresses a major bottleneck in developing and applying Digital Twin, which is the computational limits of conventional KS-DFT with a maximum cell size of a couple hundred atoms. We employed a combination of the real space finite-difference formulation and Chebyshev filtering (CheFSI) technique to solve the Kohn??"Sham equation and implemented this approach in a multi-processor, parallel environment. We derived, implemented, and achieved a ~0.15 eV desirable accuracy of ab-initio X-ray Photoelectron Spectroscopy binding energy predictions within the theoretical framework of real-space Kohn??"Sham Density Functional Theory (real-space KS-DFT). Our work provides key advantages for calculating accurate core??"electron binding energies of exascale systems with >10,000 atoms, hence bridging the knowledge gap between molecules to nanosystems of energy priorities.

Speaker: Jin Quan, Lawrence Berkeley National Labs