Specify nuclear model.
Gaussian charge distribution (default):
The point nucleus model is useful to compare Lévy-Leblond type calculations with regular nonrelativistic calculations done with another code, e.g. Dalton. The two methods should give precisely the same energies.
For the Gaussian charge distribution the default exponents are in accordance with values proposed by Visscher and Dyall [Visscher1997b].
Restrict range of nuclei in one-electron integrals involving single atomic centers, for example electric field gradients.
Example: Restrict range to four nuclei (first number), the nuclei 1, 3, 7, and 8:
.SELECT 4 1 3 7 8
Print the symmetrized nuclear magnetic moments. This corresponds to taking symmetry combinations of rotations, not coordinates, at each nuclear center. The numbering is used in labels of various magnetic integrals.
General print level in the integral module. Default:
This subsection gives directives for the generation of one-electron integrals. Based on input in the other section, the program will determine what integrals to calculate.
Print level in one-electron integral routines. By default print level is taken from **INTEGRALS.
The mol file¶
This subsection allows changes of defaults in the reading of the mol file.
Decontract basis sets specified as contracted (when being read from the library, for example). In the case of using non-relativistically contracted basis set the decontraction is necessary for heavy elements. Two-component quasirelativistic Hamiltonians (like X2C) work only with decontracted basis set. By default only the small component is decontracted.
Print level in the reading of the mol file. By default print level is taken from **INTEGRALS.
Maximum number of primitive functions in a given block in basis file. Default:
This subsection gives directives for the generation of two-electron integrals. It also gives directives for the construction of Fock matrices, such as screening.
Screening threshold for integral direct calculations of Fock matrices [Saue1997]. Default:
Note that the screening threshold may influence the convergence. In general, the screening threshold must be about three orders of magnitude smaller than the desired norm of the electronic gradient at convergence.
Choosing a negative value for the screening turns the integral screening completely off.
Separate screening of Coulomb and exchange contributions [Saue1997]. Useful for fine regulation of the convergence process. Default: Coulomb and exchange on (1 = on; off = 0):
.ICEDIF 1 1
Adjust the integral thresholds for SL and SS integrals. For conventional integral calculations only integrals above the threshold given in the mol file are written to disk. The thresholds for the SL and SS integrals are divided by the factors given here. Default:
.THRFAC 1.0 1.0
Do direct Fock matrix construction in non-symmetry-adapted basis (AO basis).
The direct Fock matrix construction is performed in AO basis using the skeleton matrix approach. This may give better screening, and does give more tasks for better parallelization, but AOFOCK is more memory intensive than SOFOCK. Default is AOFOCK if 25 or more MPI nodes.
Do direct Fock matrix construction in symmetry-adapted basis (SO basis). This is the default setting.
AOFOCK may give better screening, and does give more tasks for better parallelization, but AOFOCK is more memory intensive than SOFOCK. Default is SOFOCK if at most 24 MPI nodes.
Set the print level in two-electron integral routines for the calculation of a particular shell quadruplet. The print level is changed only for the given shell quadruplet. A zero matches all shells, thus:
.PRINT 4 0 0 0 0
sets the print level to 4 for all shell quadruplets.
Use with care to avoid massive output! At print level 15 the individual integrals are printed.
Give detailed timing for integral calculation.