Microkinetically informed library generation for modeling effective rate laws from kinetic data
The Journal of Chemical Physics · 21.11.2026 · 165 (19)
DOI: 10.1063/5.0355371 ↗Kurz gefasst
Fachartikel in der Zeitschrift The Journal of Chemical Physics, online erschienen am 20.11.2026. Verfasst von Maryke Kouyate und 5 weiteren Autor:innen. Die Arbeit nennt 63 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Deutsche Forschungsgemeinschaft, Bundesministerium für Forschung, Technologie und Raumfahrt, Deutscher Akademischer Austauschdienst.
Abstract (Originalsprache)
A profound methodological divide separates top-down macroscopic rate law identification from bottom-up, first-principles microkinetic modeling. While top-down macroscopic models often oversimplify underlying chemistry and fail to capture localized kinetic variations, bottom-up microkinetic simulations provide an exceptionally detailed look at elementary reaction pathways. However, these complex microscopic networks are frequently bottlenecked by uncertainties in both their kinetic parameters and topologies and remain too computationally demanding for full-scale reactor engineering. To bridge this gap, we present a methodology that leverages first-principles simulations for automated library generation, circumventing the bottleneck of scarce experimental data. Central to this pipeline is the novel Linearized Ensemble Motif Operator Network framework, which extracts the underlying effective kinetic information by filtering key kinetic motifs from symbolic expressions, effectively distilling intricate microkinetic networks down to simple, chemically interpretable macroscopic rate laws. This pre-compiled motif library is integrated into a sparse model identification architecture with derivative ensembles (SMIDE). To robustly handle experimental noise, SMIDE generates an ensemble of splines from heteroscedastic noise realizations to capture derivative uncertainty. Through multi-objective optimization, the framework maps out a comprehensive Pareto front of candidate models. This approach transforms rate law discovery from a simple curve-fitting exercise into a rigorous statistical analysis of the kinetic mechanisms supported by the physical system, eliminating the need for manually predefined structural libraries.
Förderung
- Deutsche Forschungsgemeinschaft
- Bundesministerium für Forschung, Technologie und Raumfahrt
- Deutscher Akademischer Austauschdienst
- U.S. Department of Energy, Office of Basic Energy Science
Zitieren
APA
Kouyate, M., Badger, K., Ducci, G., Goldsmith, C. F., Reuter, K., & Scheurer, C. (2026). Microkinetically informed library generation for modeling effective rate laws from kinetic data. The Journal of Chemical Physics, 165(19). https://doi.org/10.1063/5.0355371
Vancouver
Kouyate M, Badger K, Ducci G, Goldsmith CF, Reuter K, Scheurer C. Microkinetically informed library generation for modeling effective rate laws from kinetic data. The Journal of Chemical Physics. 2026;165(19). doi:10.1063/5.0355371
BibTeX
@article{Kouyate2026,
author = {Kouyate, Maryke and Badger, Kirk and Ducci, Gianmarco and Goldsmith, C. Franklin and Reuter, Karsten and Scheurer, Christoph},
title = {Microkinetically informed library generation for modeling effective rate laws from kinetic data},
journal = {The Journal of Chemical Physics},
year = {2026},
volume = {165},
number = {19},
doi = {10.1063/5.0355371}
}
EUVIMED