Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2611.00483

PLUMED-JAX: GPU-native enhanced sampling for atomistic simulations

Venkat Kapil
View original source ↗Read source document ↗Export metadata ↓

Abstract

We present PLUMED-JAX, an AI-assisted adaptation of selected PLUMED collective variables, umbrella sampling, and metadynamics for JAX-based atomistic simulations, developed using gpt-6.1-sol with reasoning effort set to high within the Codex harness. Integration with JAX-MD supports compiled molecular dynamics, batched replicas, walkers sharing a bias, and execution across multiple graphics processing units. We validate the framework against PLUMED and exact free energy benchmarks, and demonstrate applications to butane, silicon crystallization, and lithium transport, with severalfold improvements in computational throughput for both classical forcefields and machine learning interatomic potentials.

Explore related subjects

physics.chem-ph

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Venkat Kapil. 2026-10-31. PLUMED-JAX: GPU-native enhanced sampling for atomistic simulations. https://arxiv.org/abs/2611.00483

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Disorder-Induced Eigenstate Localization of Molecular Polaritons Despite Spectroscopic Strong Coupling

Molecular polaritons are hybrid light--matter states whose collective character is associated with a molecular excitation extended across many molecules. However, clearly resolved polariton peaks do not by themselves establish that the corresponding eigenstates retain this molecular delocalization. Here, we use a disordered Tavis--Cummings model to distinguish spectral signatures of strong coupling from eigenstate-level molecular delocalization of electronic polaritons. Static energetic disorder can strongly reduce the molecular extent of optically visible polariton states while resolved polariton peaks persist and their apparent separation can even increase. We quantify this crossover using complementary molecular localization measures applied to individual optically visible eigenstates and the photon-weighted bright sector. Finite-size scaling shows a stable large-ensemble crossover; for Gaussian disorder, retaining approximately half of the homogeneous molecular extent requires a collective coupling roughly seven times the Gaussian disorder standard deviation $σ$, although the numerical scale depends on the disorder distribution. Including cavity and molecular linewidths through a non Hermitian Hamiltonian reveals a contrasting spectral and eigenstate response: finite linewidths reduce spectral resolvability, while cavity--molecule linewidth imbalance shifts the localization crossover to larger static disorder by suppressing the disproportionate contribution of near-resonant molecular transitions. Spectroscopic strong coupling and eigenstate-level molecular delocalization therefore characterize distinct properties and need not evolve in parallel under disorder and dissipation.

physics.chem-ph↗
arXiv2026

Q-Score: A Quantum-Native Scoring Function for Molecular Docking

Molecular docking predicts how a small molecule binds to a protein and is a key bottleneck in drug discovery. Classical scoring functions sum empirical pairwise contacts, blind to quantum-mechanical effects like orbital charge transfer that govern binding specificity. We introduce Q-Score, encoding GNN-predicted orbital donor-acceptor energies into a weighted graph and scoring binding by solving a maximum-weight vertex clique problem via Digitized-Counterdiabatic QAOA. Each interaction anchor maps to one qubit and compatibility constraints become edges. Across 11 protein targets, DC-QAOA recovers the exact optimum on 8 at 10 qubits. On 1000 AI-generated molecules, Q-Score is orthogonal to classical scoring with Spearman rho of 0.05, driven by orbital quality with rho of 0.90, and free of molecular-weight bias, enriching for strong orbital interactions at twice the random rate. DC-QAOA achieves a mean approximation ratio of 0.94 with 52 percent exact. Execution of 1000 circuits on IBM Eagle confirms 6-qubit solvability on NISQ hardware.

physics.chem-ph↗
arXiv2026

An Interacting-Response Constraint for Photon Density Functionals

When the coupling to a quantized field becomes weak, the leading term of the photon exchange--correlation action depends on the fluctuations of the interacting electronic system, not those of a density-matched noninteracting reference. We establish this requirement by proposing an adiabatic connection formulation of electron-photon coupling at fixed electronic density and controlling the density derivatives. Approximating the response in the integral yields a ring functional in terms of the Kohn--Sham response and electronic kernel. Exact model tests verify fourth-order rather than second-order errors. In a two-electron Hubbard dimer, the functional correctly predicts a correlation-induced polarization reversal, while two Kohn--Sham photon prescriptions don't. In an interacting quantum-dot resonator model with electron injection and escape, the functional recovers resonant photon-assisted transport missed by two Kohn--Sham photon prescriptions.

physics.chem-ph↗