quantum chemistry software developers

Abstract: Second-quantized neural-network quantum states have achieved accurate molecular energies, but extending them across molecular geometries requires a shared representation of the geometry-dependent wavefunction coefficients. We introduce geometry-conditioned foundation neural-network quantum states for molecular electronic structure in second quantization. A single autoregressive model learns a family of ground states from sparse anchor geometries and provides wavefunctions at untrained geometries without further optimization. Orbital alignment matches orbital identities and transports their phases, establishing an aligned orbital basis across geometries. Frozen energies reach chemical accuracy at every untrained query geometry for N$_2$, CO, and H$_4$. On additional molecular paths, the energy-trained wavefunctions yield dipoles, quadrupoles, and natural occupations without property labels. Across three paired N$_2$ training seeds, orbital alignment lowers the mean absolute energy error over all untrained query geometries from 34-37 mHa to 0.049-0.085 mHa. At approximately 1 mHa mean absolute error, frozen evaluation reduces the per-geometry cost by $986\times$ relative to independent optimization, yielding an estimated $25.8\times$ end-to-end GPU-cost reduction on a 161-point N$_2$ grid.

Abstract: Nonadiabatic dynamics through a conical intersection needs both static and dynamic correlation and, at every geometry, an excited-state gradient and interstate nonadiabatic coupling (NACME); analytic multireference derivatives at this level have been cluster-scale. We report device-resident \mbox{DMRG-QD-NEVPT2} with analytic derivatives, on B2PLYP double-hybrid Kohn--Sham orbitals: a density-matrix-renormalization-group reference supplies the static correlation, and quasi-degenerate NEVPT2 adds the dynamic correlation through a multi-state effective Hamiltonian that stays valid \emph{through} the intersection, where single-state perturbation theory fails. Each gradient and NACME is one reverse-mode transpose of a contraction graph that carries the DMRG sweep as a gauge-free, tape-free node and reduces to active-space RDM contractions with Cholesky factors. The whole construction runs on a consumer 8\,GB GPU. The DMRG reference is FCI-in-active-space to $10^{-15}$ and the single-precision leg is spectroscopically inert ($<10^{-6}$\,eV); the correction yields smooth adiabats and a non-vanishing, $1/ΔE$-divergent NACME through the twisted-ethene intersection---where adiabatic linear-response TDDFT returns zero---and restores the dynamic correlation the ionic $ππ^*$ states require. DMRG-NEVPT2 and quasi-degenerate NEVPT2 are each established; the contribution is the entire stack---multi-state through a conical intersection, with analytic gradients and couplings---made device-resident on commodity hardware. Multireference companion to our density-functional realization of the same engine.