CamCASP/Bugs/6

From CUC3
Jump to navigation Jump to search

CamCASP => Bugs

Bugs that arise when rotation are performed

Error because molecule reference is not correctly set

See Rotations for more on this bug. It is present in 5.4.00. Basically what happens is that the molecular geometry and orbitals have been updated to be consistent, but the rotation (and translation) field is not re-set to NULL. So, at some point in the code, the rotation could be re-applied and this would make the geometry and orbitals inconsistent.

This has been fixed as follows:

  • Subroutine set_reference is called when the molecule is created in the MOLECULE block. This does the updating and sets the reference correctly. The molecular rotation and translation fields are then nullified so everything is consistent.
  • The EDIT block is provided with a REFERENCE command that allows the user to set the reference manually. So, if geometry and orbitals are updated in this block, the REFERENCE command must be called.

Error in <math>E^{(1)}_{\rm elst}</math> when dimer DF is not done

  • Version: 5.4.00 and earlier
  • Basis: Spherical auxiliary basis.
  • Effects: S___,_A_B and any integral and energy that depends on this.

Whether of not you do the dimer DF should have nothing to do with <math>E^{(1)}_{\rm elst}</math>, but version 5.4.00 of the code has an error in j_matrix.F90 that results in an error if the dimer DF is not done before calculating the electrostatic energy (or any energy that uses the S___,_A_B integrals). This occurred because norm_integrals2 was not called even though it is needed if the auxiliary basis uses shperical GTOs.

FIXED: I've now added code to make the call to norm_integrals2. Seems to work.