CPOS Seminar: "A Stochastic Framework for Efficient Simulation of Solvated Reaction Energetics"

Speaker: Annabelle Canestraight, PhD Candidate, Vlcek Group / Department of Chemical Engineering, UC Santa Barbara

Electronic structure theory plays a central role in determining the energetics of reaction pathways and identifying the most probable reaction mechanisms. For many industrially and biologically relevant processes, accurate modeling requires the inclusion of a solvated environment, as solvent molecules can strongly correlate with reacting species and mediate critical reaction steps. However, systems that include both solute and solvent often contain a prohibitive number of electrons, making accurate correlated calculations computationally challenging. Mean-field approaches such as Hartree–Fock and density functional theory (DFT) offer affordable routes to obtain ground-state energies but fail to fully capture electron correlation, leading to errors along the reaction coordinate. High-level correlated methods such as full configuration interaction (FCI), coupled cluster (CC), and the density matrix renormalization group (DMRG) provide systematic improvements but are limited by unfavorable, often exponential, computational scaling.

In this work, we present a stochastic framework for obtaining the total correlation energy with the accuracy of DMRG at a fraction of the computational cost. The method samples single-particle orbitals and repeatedly solves a low-dimensional effective Hamiltonian, using the resulting correlation energies to reconstruct the total correlation energy of the full system. We demonstrate this approach for a molecule evolving dynamically in solvent and for a prototypical chemical reaction, showing excellent agreement with deterministic high-level benchmarks. Finally, we illustrate how this stochastic methodology offers a new quantitative metric for assessing electronic “innocence” in condensed-phase reactions.