Contents: Table of Contents   Detailed Control of Optimisations  Chapter 4   Finding Minima

Constants, Variables and Partial Optimisations

The ability to declare certain geometric parameters to be CONSTANTS or VARIABLES was mentioned in the chapter on geometry input. The difference only matters when one is doing an optimisation. For example, suppose the HF dimer example in the previous section were modified so that the dataset looked like,

TITLE
HF-HF  OPT
SYMMETRY
CS
END
BASIS 631G**
CONSTANTS
RFH1   0.912 A
RFH2   0.912 A
END
VARIABLES
RFF 2.75 A
THETA1  5.0 D
THETA2  -110.0  D
END
ATOMS
F1 9 0 0 0
F2 9 0 0 RFF
H1 1 POL RFH1 THETA1 0
H2 1 PTC F2 F1 H1 RFH2 THETA2
END
OPTIMISE
START
FINISH

This is the same initial geometry as previously except that the two HF bond lengths are declared to be CONSTANTS. In the geometry optimisation these then are constrained to their initial values. The final set of parameters looks like this.

    Parameter                 value                 gradient
      name          (bohr or        (angstrom
                     radian)        or degree)
      RFF           5.15686438      2.72889531      0.00008308
      THETA1        0.24558172     14.07079633      0.00002883
      THETA2       -1.77701001   -101.81517368     -0.00002775

Note that the values of the two CONSTANTS are not given as the corresponding gradient elements are zero. A partial optimisation like this can be useful on occasions.