SpecCert

SpecCert turns ORCA TD-DFT and frequency outputs into UV-Vis and IR spectra, and VASP DOSCAR files into d-band moments, and checks the inputs on the way: whether the absorption maximum depends on how many excited states were computed, whether a frequency scaling factor exists for the method and basis set, whether there are imaginary modes, and whether the DOSCAR has d projections. A PASS means that these checks found nothing wrong; it does not validate the level of theory.

What it computes

ε follows from the oscillator strengths through f = 4.319 × 10⁻⁹ ∫ε dν̃ (ν̃ in cm⁻¹), so the broadened spectrum integrates back to Σf.

Validation (validation/)

Every number below is regenerated by the scripts in validation/.

• UV-Vis (benchmark_uv_ir.py): eight molecules (formaldehyde, acetone, acrolein, butadiene, hexatriene, benzene, pyridine, naphthalene) with ORCA 6.1.1 TD-DFT (B3LYP-D3(BJ)/def2-SVP, 15 states). ε agrees with ORCA's own orca_mapspc spectrum to 1.2 × 10⁻⁴ (relative, wherever ε exceeds 1% of its maximum); integrating the spectrum returns Σf to 8 × 10⁻¹⁰. Versions 1.0.0 and 1.1.0 used a prefactor of 2.174 × 10⁴ instead of 2.8706 × 10⁴: their ε was 0.757 times the correct value. Version 1.0.0 found no transitions in ORCA 6 outputs.
• IR (benchmark_uv_ir.py): frequencies and intensities read from the eight frequency outputs agree with the .hess files (0 and 0.005 km/mol); the broadened spectra integrate to the summed intensities.
• Scaling factors (extract_cccbdb.py): the 299 factors of the CCCBDB table for 16 methods. Versions 1.0.0 and 1.1.0 applied one factor per functional whatever the basis set (B3LYP 0.9679, while CCCBDB lists 0.892–0.975 across basis sets), matched functional names by substring (wB97X-D3 received the wB97X-D factor) and used 0.965 for unknown methods. Their command line accepted frequencies only and gave every mode an IR intensity of 50 km/mol.
• d band (fetch_nomad_doscar.py, benchmark_dos.py): nine real VASP calculations from NOMAD (CC BY 4.0; bulk Cu, Ag, Ni, Pt, Fe, Co and the spin-polarised clusters Fe₃₉, Ni₄₇, Co₁₈). The d-band centres, width and filling read from DOSCAR agree with pymatgen 2026.9.24 reading the vasprun.xml of the same calculations to 0.002 eV (0.0006 in filling). Versions 1.0.0 and 1.1.0 used the first ion only, the spin-up total DOS, and the wrong columns for spin-polarised projections: on the clusters their occupied d-band centre was off by −0.22 to −0.77 eV.

Not validated

• Gaussian TD-DFT parser (no Gaussian installation was available); Raman activities are stored but not broadened.

Installation

pip install "git+https://github.com/sircalch/speccert@v1.2.0"

From source:

git clone https://github.com/sircalch/speccert.git
cd speccert
pip install -e .[dev]

Command line

# ORCA TD-DFT and frequency outputs (level of theory read from the outputs)
speccert assess --input-uv tddft.out --input-ir freq.out --system acrolein -o report/

# a scaling factor of your own (e.g. for a basis set not in CCCBDB)
speccert assess --input-ir freq.out --scaling-factor 0.965 -o report/

# VASP DOSCAR, d projections of ions 1-4 only
speccert assess --input-dos DOSCAR --ions 1,2,3,4 -o report/

# synthetic demonstration (not a real calculation)
speccert demo -o demo/

The report directory contains report.html, figures, a summary table (CSV and LaTeX), a draft methods paragraph stating what was found, and citation.bib.

Python

from speccert.parsers.orca_tddft import parse_orca_tddft_output
from speccert.parsers.orca_freq import parse_orca_freq_output
from speccert.parsers.vasp_doscar import parse_vasp_doscar
from speccert.core.uv_vis import calculate_uv_vis_spectrum
from speccert.core.vibrational import calculate_scaled_vibrational_spectrum
from speccert.core.dos_dband import calculate_dos_and_dband_center

t = parse_orca_tddft_output("tddft.out")
uv = calculate_uv_vis_spectrum(t["energies_ev"], t["oscillator_strengths"], fwhm_ev=0.30)

f = parse_orca_freq_output("freq.out")
ir = calculate_scaled_vibrational_spectrum(f["frequencies_cm1"], f["ir_intensities_km_mol"],
                                           functional=f["method"], basis=f["basis"])

d = parse_vasp_doscar("DOSCAR")
dband = calculate_dos_and_dband_center(d["energies_ev"], d["total_dos"], d["projected_d_dos"], d["fermi_energy_ev"])
print(uv.status, ir.status, dband.status, dband.d_band_center_filled_ev)

Citation

speccert cite prints the APA and BibTeX entries for the installed version (concept DOI 10.5281/zenodo.22217588). CCCBDB: NIST Computational Chemistry Comparison and Benchmark Database, NIST Standard Reference Database 101.

License

MIT, see LICENSE.