11 papers
Stiefel Attention: When the Geometry of Transformer Projection Matrices Dominates Optimizer Choice---and When It Does Not
Rubén Darío Guerrero
The query and key projections $\WQ,\WK$ in attention are almost always trained by Euclidean optimizers with no constraint on their geometry. We constrain them to the Stiefel manifo…
Analytic Gradients and Nonadiabatic Couplings for Device-Resident DMRG-QD-NEVPT2 Through Conical Intersections on a Consumer GPU
Rubén Darío Guerrero
Nonadiabatic dynamics through a conical intersection requires both static and dynamic correlation and, at every geometry, an excited-state gradient and interstate nonadiabatic coup…
Spin Rotatory Strength as the Equilibrium Observable for Chiral-Induced Spin Selectivity
Rubén Dar\io Guerrero
We identify the spin rotatory strength---a chirality-odd, finite-frequency spin--dipole cross response---as the equilibrium observable for chiral-induced spin selectivity. Time-rev…
Density-Functional Excited-State Gradients and Nonadiabatic Couplings on a Consumer GPU from a Contraction-DAG
Rubén Darío Guerrero
Nonadiabatic dynamics needs an excited-state gradient and an interstate nonadiabatic coupling matrix element (NACME) at every nuclear geometry, and a double-hybrid functional's acc…
EOM-CC Excited-State Gradients and Nonadiabatic Couplings on a Consumer GPU from a Contraction-DAG with Laplace-Transform J/K Kernels
Rubén Darío Guerrero
We present a unified, memory-bounded GPU realization of equation-of-motion coupled-cluster (EOM-CC) excited-state gradients and interstate nonadiabatic couplings (NACMEs) on a sing…
Bond-dimension scaling of a local-refinement advantage over hyperoptimized tensor-network contraction on Sycamore like topologies
Rubén Darío Guerrero
We identify a missing local-refinement stage in the cotengra tensor-network contraction pipeline and show that its impact grows monotonically with bond dimension on the \emph{conne…