physics.chem-phNov 7, 2026

Toward universal excited-state machine-learned potentials: a local excitation Hamiltonian for size consistency and transferability

Authors: Pascal Friederich

Organizations: Institute of Nanotechnology, Karlsruhe Institute of Technology, Kaiserstr. 12, 76131 Karlsruhe, Germany · Institute for Anthropomatics and Robotics, Karlsruhe Institute of Technology, Kaiserstr. 12, 76131 Karlsruhe, Germany

Abstract

Universal machine-learned interatomic potentials could make excited-state simulations across chemical space routine, but excitation energies, unlike ground-state energies, are not sums of atomic contributions. The excited states of noninteracting molecules are the union of their states, which pooled multistate readouts reproduce at best for the lowest state. We introduce LEH-MACE, whose excitation energies are the eigenvalues of a local excitation Hamiltonian assembled from atom-centred blocks, so this union holds exactly. We generated TD-DFT energies and forces for about 300,000 equilibrium, thermal, and excited-state geometries of over 100,000 molecules. LEH-MACE's advantage over direct readouts grew with the distance from the training data and was largest above S1S_1, with about half their S2S_2 and S3S_3 errors for long conjugated oligomers. For molecule pairs, only LEH-MACE kept its single-molecule accuracy, while a softmin readout remained superior for S1S_1 dynamics of larger chromophores. Exact size consistency is thus one step toward such potentials.

Figures & tables

Explore similar work

CardsList
  1. On the Theoretical Limits of Learning Unconstrained Interatomic Potentials

    Oct 10, 2026Prajwal Ananth, Shuwen YueMachine Learning Interatomic PotentialsMolecular Dynamics Simulation

  2. MACE-POLAR-1: A Polarisable Electrostatic Foundation Model for Molecular Chemistry

    Feb 23, 2026Ilyes Batatia, William J. Baldwin, Domantas Kuryla +10Molecular Foundation ModelsMachine Learning Interatomic Potentials

  3. Physics-Aligned Electronic Ground-State Learning Improves Generalization

    Oct 7, 2026Eike S. Eberhard, Xaver Kainz, Viktor Kotsev +2Physics-Informed MLOOD Generalization