AI-Driven Adaptive Spherical-Harmonic Basis Sets Anchored on the Bohr Radius for Qubit-Efficient Molecular Simulation
AbstractThe deployment of quantum chemistry on near-term quantum processors is severely con-strained by the qubit overhead imposed by conventional Gaussian-type orbital (GTO) basissets. Correlation-consistent bases such as cc-pVDZ, while chemically reliable, map even smallmolecules like H2O onto 48 qubits—a scale that strains the coherence budgets of currentnoisy intermediate-scale quantum (NISQ) hardware. We introduce an alternative paradigm:an AI-driven adaptive spherical-harmonic basis anchored directly on the Bohr radius (a0),in which radial scaling is modulated by a neural-network-predicted effective nuclear chargeZeff(n, l) that captures electron shielding across angular momenta. Angular polarization—whether arising from solvent fields or protein active-site electrostatics—is encoded throughequivariant tensor coupling using Clebsch–Gordan coefficients. In a PySCF benchmark onH2O, the compact basis reduces the qubit count from 48 to 14 (a 70.8% compression) while aresidual neural correction layer recovers the correlation energy gap without additional qubitoverhead. We outline the mathematical framework, present the benchmark, and discusshigh-value applications in pharmaceutical discovery and solid-state electrolyte design.
Authors
- Alwin Roth (ORCID: https://orcid.org/0009-0009-9467-0956)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-20
- DOI
- https://doi.org/10.5281/zenodo.24387205
- Primary Topic
- Quantum Computing Algorithms and Architecture
- Type
- article
- Field-Weighted Citation Impact
- 0.00