guest ::
| Home > Publications database > Data‐Driven Spectral Prediction for Accelerating Large‐Scale Electronic Structure Calculations |
| Home > Publications database > Data‐Driven Spectral Prediction for Accelerating Large‐Scale Electronic Structure Calculations |
| Typ | Amount | VAT | Currency | Share | Status | Cost centre |
| Hybrid-OA | 0.00 | 0.00 | EUR | (DEAL) | ZB | |
| Sum | 0.00 | 0.00 | EUR | |||
| Total | 0.00 |
| Journal Article | FZJ-2026-04616 |
; ; ; ; ; ;
2026
Wiley-VCH
Weinheim
This record in other databases:
Please use a persistent id in citations: doi:10.1002/pamm.70204 doi:10.34734/FZJ-2026-04616
Abstract: Simulating large molecular systems comprising thousands of atoms requires highly scalable methodologies. While modern Density Functional Theory (DFT) codes exhibit linear scaling, solving the associated large, sparse generalized eigenproblems remains a critical computational bottleneck on exascale architectures. In the context of the LimitX project, we propose a data-driven framework to accelerate these calculations. By shifting the machine learning target from discrete eigenvalues to the coefficients of an interpolating Chebyshev polynomial, and by comparing both all-atom and fragment-based structural representations, we successfully overcome the dimensionality constraints of large-scale spectral prediction. We investigate three machine learning models (kernel ridge regression, graph neural networks, and Random Forests) trained on a novel 2 TB dataset of protein dimers. The predicted spectra provide initial guesses that effectively bypass early self-consistent field (SCF) iterations in BigDFT. Ultimately, these spectral predictors will be deployed to dynamically optimize upcoming rational filter-based eigensolvers, such as FrASE, which is currently in initial development.
; 
; 
; DEAL Wiley ; Ebsco Academic Search
|
The record appears in these collections: |
PDF