Predictive Free Energy Simulations Through Hierarchical Distillation of Quantum Hamiltonians

2025-09-13

AI Summary

Computational cost prohibits high-accuracy quantum mechanical (QM) methods from predicting reaction free energies in condensed-phase systems. To address this, we propose a hierarchical machine learning (ML) framework based on Hamiltonian distillation: a small set of high-level QM calculations is distilled into a multi-scale, explicitly electron-resolved ML quantum Hamiltonian, enabling rigorous coupling between quantum and classical degrees of freedom and accurate description of long-range electrostatics and infinite-order quantum environmental response. Integrating free energy perturbation theory with first-principles modeling, our approach achieves fully quantum-mechanical predictions of weak-acid pKa values and enzymatic reaction rates—without empirical parameters—for the first time. Predictions attain chemical accuracy (≤1 kcal/mol) or fall within experimental uncertainty, markedly improving both accuracy and statistical convergence of reaction free energy simulations.

Technology Category

• Machine Learning: Quantum Machine Learning
• Humans and AI: Human-in-the-loop Machine Learning
• Multiagent Systems: Multiagent Learning

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Obtaining the free energies of condensed phase chemical reactions remains computationally prohibitive for high-level quantum mechanical methods. We introduce a hierarchical machine learning framework that bridges this gap by distilling knowledge from a small number of high-fidelity quantum calculations into increasingly coarse-grained, machine-learned quantum Hamiltonians. By retaining explicit electronic degrees of freedom, our approach further enables a faithful embedding of quantum and classical degrees of freedom that captures long-range electrostatics and the quantum response to a classical environment to infinite order. As validation, we compute the proton dissociation constants of weak acids and the kinetic rate of an enzymatic reaction entirely from first principles, reproducing experimental measurements within chemical accuracy or their uncertainties. Our work demonstrates a path to condensed phase simulations of reaction free energies at the highest levels of accuracy with converged statistics.

Problem

• Predicting free energies of chemical reactions computationally
• Bridging quantum accuracy with machine learning efficiency
• Enabling first-principles simulation of condensed phase reactions

Innovation

• Hierarchical machine learning framework distillation
• Explicit electronic degrees embedding
• First-principles quantum-classical embedding simulations