Why don't the energies I obtained from Jaguar exactly match those from other programs?

By default, Jaguar uses the pseudospectral method for converging the SCF and calculating the quantum mechanical energy of the system. The pseudospectral method is very fast, and is unique to Jaguar. It is a grid-based method, and all grid-based methods (including most DFT implementations) will necessarily give energies slightly different from methods which do not employ grids (or use differing grids). Energy discrepancies should be less than 1 kcal/mol, however, and they are generally less than 0.1 kcal/mol. When comparing DFT energies, be aware that different grid sizes and densities and different "pruning" schemes will necessarily lead to different energies.

When comparing energies from two different programs, you must also ensure that both programs are using the same number of d functions in the basis set. Some programs and/or basis sets will default to the use of 6 d functions, while others will use 5 d functions. You can control what Jaguar does by setting numd=5 or numd=6 in the input file.

Jaguar also removes basis functions if they cause the basis set to become linearly dependent, because this would lead to numerical problems in the calculations. Other programs might remove a different number of basis functions, or none at all. You can see whether Jaguar has removed any basis functions by looking in the output file for occurrences of the "number of basis functions:" and "number of canonical orbitals:" headings. If any basis functions have been removed, then the number of canonical orbitals is less than the number of basis functions.

Keywords: mismatch, mismatched, Jaguar, SPE, energy

Back to Search Results