GeinDFT
Home › Accuracy and validation

Accuracy and validation

What the numbers have been checked against, how closely they agree, and what is deliberately not supported.

Every quantity below has been compared against an established reference program at matched settings, on a set of molecules spanning closed and open shells, neutral and charged species and several basis sets. The agreement is quoted as the largest difference observed, not the average.

Measured agreement

QuantityAgreementConditions
Hartree-Fock total energyBetter than 1e-8 hartreeMatched grids and thresholds
DFT total energy, local, gradient-corrected, meta and hybrid functionalsBetter than 1e-6 hartreeConverged integration grid
Valence orbital energiesBetter than 1e-5 hartree
Nuclear gradientsBetter than 1e-6 hartree per bohr
Optimised bond lengthsBetter than 0.0001 angstrom
Harmonic frequenciesBetter than 1 reciprocal centimetre
Dipole momentBetter than 0.0001 debye
Mulliken and Loewdin chargesBetter than 0.0001 electrons
CHELPG chargesBetter than 0.02 electronsSame radii and grid on both sides
Electrostatic potential on the density surfaceBetter than 0.002 kcal/mol

Reproducibility

The same input gives bit-identical results on any number of processor cores, from one to twenty. This is not automatic: it requires that every parallel sum be accumulated in a fixed order rather than in whatever order the threads happen to finish. It means a number you publish can be reproduced exactly on a different machine.

Internal consistency checks

Settings matter

The agreement above is measured at converged integration grids. At the default accuracy preset, which is chosen to be fast enough for everyday work, differences are larger. Where a number matters, use the Tight or Very Tight preset and say which was used.

Not supported, and why

ItemStatus
Minnesota functionalsAvailable, but the agreement with the reference does not reach 1e-6 hartree. Treat results as indicative.
wB97X-DRefused. Its definition includes a dispersion term that is not implemented, so a result under that name would not be that functional. Use wB97X or CAM-B3LYP.
Double hybridsSingle-point energies only. There is no analytic gradient, so they cannot drive an optimisation or a frequency calculation.
ROHF and ROKSNot supported. Use the unrestricted forms for open-shell systems; the spin contamination is reported.
Hirshfeld and CHELPG charges compared across programsThe definitions differ between programs. Our values are internally consistent and documented; comparisons across programs need matched definitions.

Dispersion

The dispersion correction supplied with GeinDFT is parameterised for B3LYP, PBE0, PBE, BLYP, TPSS and Hartree-Fock, and covers elements from hydrogen to krypton. On a halogen benchmark set that played no part in fitting it, the mean absolute error is between 0.26 and 0.40 kcal/mol, against 0.92 to 2.29 kcal/mol with no dispersion correction at all.

A dispersion correction that cannot be applied to the chosen functional or to an element in the molecule is refused with a reason. It is never silently omitted, because an energy that claims a correction it did not receive is worse than an error message.