ExFiT: Ensemble-based Excited-State Character Fingerprinting across Temperatures

Repository Description This repository contains all data, computational outputs, and analysis scripts used for the ExFiT study: Ensemble-based Excited-State Character Fingerprinting across Temperatures. It provides equilibrium geometries, finite-temperature Wigner ensembles, quantum-chemical outputs, excited-state descriptor data, and post-processing scripts used to characterize the temperature-dependent excited-state behavior of four representative chromophores: trans-Azobenzene, (E)-Hemithioindigo (HTI), para-Nitroaniline (PNA), and Coumarin 481 (C481). The archive is organized into the following top-level components to ensure full transparency of the computational workflow: TeraChem, NewtonX, TheoDORE, and Scripts_&_CSVs (which contains seven subfolders, see below). 1. TeraChem Contains all static (equilibrium-geometry) quantum-chemical outputs, computed at the CAM-B3LYP-D3/Ahlrichs-pVDZ level using TeraChem. terachem.out — Output files for equilibrium-geometry vertical excitation (absorption) and harmonic frequency (IR) calculations, for all four molecules. Dipole moments — Ground-state, excited-state (relaxed and unrelaxed), and transition dipole moments for all systems. Raw IR data — Harmonic vibrational frequencies and IR intensities at the optimized ground-state geometry. Raw absorption data — Vertical excitation energies and oscillator strengths, provided both in eV and nm. Broadened absorption spectra — Gaussian-broadened absorption profiles (FWHM = 0.7 eV), in nm, for all four molecules. Charge density difference (CDD) figures — Excitation-resolved CDD isosurface images for the first 20 singlet excited states of all four molecules, illustrating electron accumulation (blue)/depletion (orange) patterns upon excitation. 2. NewtonX Contains the finite-temperature Wigner sampling inputs and outputs generated with Newton-X, used to build the nuclear ensembles for each molecule at 200, 300, and 400 K. Equilibrium geometry files — .xyz and .molden files for each molecule's optimized ground-state structure and normal modes, used as the reference structure and frequency input for Wigner sampling. initqp.input — Newton-X input files for Wigner ensemble generation, provided separately for each of the three temperatures (200 K, 300 K, 400 K) and each molecule. Cross-section data — Simulated nuclear-ensemble absorption cross-section data points (temperature-broadened spectra) for all three temperatures, for all four molecules. Wigner ensemble geometries — the full set of N = 501 sampled nuclear geometries (including the equilibrium geometry) for each molecule at each of the three temperatures (200, 300, 400 K), provided in XYZ format in atomic units (bohr), used as initial conditions for the TDA-TDDFT calculations. 3. TheoDORE Contains the complete excited-state descriptor analysis for every sampled geometry across the Wigner ensembles, computed with the TheoDORE package. Data is organized as compressed archives, one per molecule per temperature, each containing 501 subfolders (geom_001 – geom_501) corresponding to the individual ensemble geometries. Each geom_### subfolder contains: dens_ana.in — TheoDORE input file specifying fragment definitions and requested descriptors for that geometry. terachem.out — TeraChem vertical excitation output for that specific ensemble geometry, used as input for the TheoDORE analysis. [molecule].molden — Molden-format file for that ensemble geometry, used for orbital/NTO visualization. nto_A*.mld — Molden files of the natural transition orbitals (NTOs) for each computed excited state. nto_jmol.spt — Jmol script file for rendering the NTO isosurfaces. tden_summ.txt — Summary table of all computed transition-density-matrix descriptors (CT, RMSeh, SHE, PRNTO, COH, PR, etc.) for each excited state at this geometry (the primary per-geometry descriptor output). OmFrag.txt — Fragment-pair-resolved decomposition of the CT number, showing the contribution of each donor/bridge/acceptor fragment pair to the overall charge-transfer character. ehFrag.txt — Fragment-resolved electron and hole population analysis, quantifying how the electron and hole densities are distributed across the defined molecular fragments. theodore.log — Full log file of the TheoDORE run for this geometry. 4. Scripts_&_CSVs Contains the aggregated descriptor datasets and all analysis scripts, organized into the following subfolders: data_extractor — Scripts and outputs for extracting excited-state descriptors (CT, RMSeh, SHE, PRNTO, COH, PR) from the raw TheoDORE tden_summ.txt output across all 501 geometries and three temperatures, for all four molecules, producing the *_descriptors_all_temps.csv tables used throughout the downstream analysis: Azobenzene_descriptors_all_temps.csv, Cumarin_c12_descriptors_all_temps.csv, HTI_descriptors_all_temps.csv, PNA_descriptors_all_temps.csv. HTI_structural_analysis — Scripts and data for the HTI C–C dihedral/torsional analysis, comparing thermally sampled rotation angles against the relaxed potential-energy rotation scan (scan_HTI_CC_rotation.csv). PNA_structural_analysis — Corresponding torsional analysis for PNA, relating the S3 CT-number distribution to NO2/NH2 torsional and out-of-plane deviations (PNA_S3_merged_dihedrals_CT.csv). state_tracking — Scripts and data implementing and comparing the three excited-state assignment criteria used in this work: oscillator strength, descriptor-fingerprint distance, and NTO-pair overlap, including pairwise and three-way agreement statistics between them. statistics — Scripts and data for the ensemble-resolved statistical analysis of thermal behavior: bootstrap confidence intervals, Cohen's d effect sizes, permutation tests on the mean and Wasserstein-1 distance, Wilson confidence intervals and two-proportion/Fisher's exact tests for bright-state populations, Benjamini–Hochberg FDR correction, and seed- and ensemble-size-sensitivity checks — run separately for the fixed-label (adiabatic) and NTO-overlap-tracked state-identity analyses. terachem_runtime — Per-geometry TDA-TDDFT wall-time records from TeraChem, for all four molecules across all three temperatures. Workflow_script — Standalone scripts representative of the in-house automation used to orchestrate job submission (Newton-X → TeraChem → TheoDORE) and to collect TheoDORE outputs into the descriptor tables used for analysis on our local computing cluster, provided to illustrate the full workflow and to support adaptation to other computing environments. The top-level plotting script plot_crossmolecule.py (in data_extractor directory) uses the *_descriptors_all_temps.csv files to generate the cross-molecule comparison figures reported in the main text and SI. Scientific Scope This repository enables detailed reproduction and further analysis of: Temperature-dependent distributions of excited-state descriptors (CT number, electron-hole separation, entanglement entropy, NTO participation ratio) across finite-temperature Wigner ensembles PCA-based excited-state character fingerprints and their temperature dependence Bright-state identity and reordering statistics across thermally sampled geometries Structural (torsional) origins of descriptor variability Nuclear-ensemble-averaged absorption spectra at multiple temperatures Reproducibility All results reported in the manuscript and SI can be reproduced from the data and scripts provided: Equilibrium geometries, normal modes, and Wigner ensemble inputs are included for all four molecules. Per-geometry TDDFT and TheoDORE outputs are provided for all 501 × 3 (temperature) × 4 (molecule) ensemble calculations. Descriptor datasets and all plotting/statistical-analysis scripts used to generate the figures and statistics in the manuscript and SI are included. Software versions, hardware specifications, aggregate compute cost, job success/failure rates, and random seeds used are reported in SI Section S11. Job orchestration across the local computing cluster (Newton-X → TeraChem → TheoDORE) and the collection of results into the descriptor tables were managed by an in-house automation script; a representative version is provided in Scripts_&_CSVs/Workflow_script/.