Teaching core-hole spectroscopy to a deep neural network Conor Rankine 1 and Thomas J. Penfold 2 1 University of York, UK, 2 Newcastle University, UK Deep neural networks (DNNs) [1] – multilayer machine-learning models that are able to extract and learn patterns represented in data without hand-coded heuristics – are transforming what we can do, and the way we do it, across the physical sciences. XANESNET [2,3] is a DNN for instantaneous simulations of X-ray absorption spectra (XAS); the XANESNET Project is about addressing the challenge of delivering detailed, high-level theoretical simulations that can capture the complex underlying physics of these experiments but that are – at the same time – fast, affordable, and accessible enough to appeal to beamline users. [4] Using DNNs like XANESNET, we can reduce the time taken to simulate XAS from hours/days to a fraction of a second, democratise data analysis, and enable beamline users to better plan beamtime allocations by facilitating ‘on-the-fly’, ‘real-time’ analysis. [4] We’ve already deployed XANESNET in the practical arena to take on open questions in physical [5] and materials [6] chemistry with success. This talk will showcase how XANESNET delivers ‘black-box’, qualitative predictions of XAS at the transition metal K-edges using nothing more than the local geometries of arbitrary absorption sites, bypassing time- and resource- intensive quantum-chemical calculations. In benchmarks, XANESNET reproduces peak positions to sub-eV accuracy wrt . reference XAS data and peak intensities with errors an order of magnitude smaller than the spectral variation in our reference XAS data set consistently across the transition metal K-edges. Behavior in line with the expected physics is also learned, and this talk will touch on how we can 'lift the lid' on the model and see this in action. References 1. Y. LeCun, Y. Bengio, G. Hinton, Nature , 2015, 521 , 436.C. D. Rankine, T. J. Penfold, J. Chem. Phys. , 2022, 156 , 164102. 2. C. D. Rankine, M. M. M. Madkhali, T. J. Penfold, J. Phys. Chem. A , 2020, 124 , 4263. 3. C. D. Rankine, T. J. Penfold, J. Phys. Chem. A , 2021, 125 , 4276.M. M. M. Madkhali, C. D. Rankine, T. J. Penfold, Phys. Chem. Chem. Phys. , 2021, 23 , 9259. 4. E. Falbo, C. D. Rankine, T. J. Penfold, Chem. Phys. , 2021, 780 , 138893.
DC06
© The Author(s), 2023
Made with FlippingBook Learn more on our blog