Rodrigo Wang

A tractable approach to solving the exact many-body electronic wavefunction has long remained an elusive goal in quantum chemistry. Accurate computation of the electronic structure and related properties of molecules would unlock a trove of precise information, but the exponential scaling of exact methods has impeded this possibility. In this work, we report that the combination of an incremental expansion of electronic correlation with effective utilization of modern cloud compute environments can overcome the long-standing barrier of intractability. As a demonstration of viability, we investigate the bond breaking of per- and polyfluoroalkyl substances (PFAS), a class of strongly correlated systems with pressing environmental relevance. Using the incremental Full Configuration Interaction (iFCI) method, we decompose the many-body wavefunction into independently computable units and distribute them across a record-breaking one million simultaneous cloud vCPUs. Calculations for perfluorooctanoic acid, the largest PFAS studied, involve correlating 150 electrons in 330 orbitals, corresponding to a wavefunction dimension of ∽10¹⁵¹ configurations and producing the most accurate correlation energies and electron densities to date. iFCI reveals an electron localization transition in the electronic state during bond dissociation, which standard quantum chemical methods (e.g., DFT) fail to capture due to insufficient treatment of correlation. The union of a highly parallel cloud-enabled infrastructure with a polynomial-scaling method that treats static and dynamical correlation on equal footing lays the foundation for further development of novel protocols for the degradation of PFAS, alongside potential for characterization and screening of advanced molecules and materials in a wide range of chemistries.