Id: arxiv/2611.16080 Url: https://arxiv.org/abs/2611.16080 Tags: quant-ph, cs.ET, physics.chem-ph Score: 0 Title: TGen-SQD: Physics-Informed Transferable Configuration Generator for Sample-Based Quantum Diagonalization Domain: quant-ph Source: arxiv Authors: Niloy Deb Roy Mishu, Lei Jiang, Mohsen Heidari, Ruizhe Zhang, Fan Chen Reviews: 0 Subject: quant-ph Category: research Comments: 0 Disputed: false Division: physics Mentions: 0 Discussed: 0 Published: 2026-11-11 Page: https://delphi.science/r/arxiv/2611.16080 Sample-based quantum diagonalization (SQD) avoids iterative variational circuit optimization, but its accuracy and cost depend on selecting a compact determinant subspace that captures the ground state. Existing machine-learning selectors use task-specific physical cues and typically retrain for each Hamiltonian. We introduce TGen-SQD, a physics-informed generator jointly trained across multiple Hamiltonians. It learns from configuration-interaction-weighted data and uses orbital- and system-level descriptors to guide generation. After training, its weights remain frozen for inference on…