Data and plotting scripts for "Machine-Learned Compact Subspace Generation for Quantum Selected Configuration Interaction within Density Matrix Embedding Framework"

This record contains the processed run logs and plotting scripts underlying the results in "Machine-Learned Compact Subspace Generation for Quantum Selected Configuration Interaction within Density Matrix Embedding Framework" (arXiv:2607.20585). carmofur/ contains the logs of the four DMET chemical-potential (mu) evaluations of DMET-QSCI-RBM applied to the QM region of Carmofur bound to the SARS-CoV-2 main protease (53 atoms, 10 fragments, cc-pVDZ). Each fragment was solved in a 16-orbital, 16-electron active space from 100,000 shots on IBM Quantum hardware (40 jobs in total). For each fragment the logs report the IBM Quantum job ID, the number of unique and symmetry-obeying bitstrings, the per-iteration QSCI-RBM energy, subspace dimension and fraction of the symmetry space accessed (SS%), the final alpha-string subspace, the fragment energy and the fragment electron number. For each mu evaluation they report the total DMET energy and mu residual. plot.py reproduces the DMET convergence figure; the DMET-SQD and DMET-FCI reference values it plots are entered directly in the script. water/ contains the log of a standalone QSCI-RBM calculation on H2O (cc-pVDZ, CAS(10,24), 48 qubits) from 250,000 shots on ibm_boston, with HF, MP2, CCSD and CCSD(T) references, transpiled circuit metrics, the physical qubit layout and 80 QSCI-RBM iterations. plot_qsci_rbm_h2o.py parses this log and reproduces the convergence figure. The logs were condensed from the raw output by removing duplicated lines and listing settings once per file; no numerical values were changed. The QSCI-RBM solver source code is not included.