finite-difference frequencies

[Charles Doubleday]

Molpro apparently has a bug in its finite-difference frequency code.
The Molpro 2000.1 manual lists the following example in which the input
file h2o_freqdft.com produces an output file h2o_freqdft.out.

"h2o_freqdft.com":

geometry={O;                 !Z-matrix for water
          H1,O,R;
          H2,O,R,H1,THETA}
R=0.96 Ang                   !start bond distance
Theta=104                    !start bond angle
basis=6-31g**                !Pople basis set
functional=b3lyp             !define fucntional (optional, b3lyp is
default)
freqdft                      !run frequency calculation

frequencies from "h2o_freqdft.out"

 Imaginary Vibration  Wavenumber
        Nr             [1/cm]
        1               62.30
        2               62.30
        3               23.57

   Low Vibration      Wavenumber
        Nr             [1/cm]
        1                1.60

     Vibration        Wavenumber
        Nr             [1/cm]
        1             1052.45
        2             1660.40
        3             2322.20
        4             2964.69
        5             3794.43


"freqdft" is evidently a macro that first optimizes H2O (gradmax =
2.6e-4 au) and computes frequencies at the optimized geometry.  The
program appears to have a bug because it gives 5 frequencies.  In
addition, it does not project out translations and rotations.  Also,
where is "freqdft" documented?

Thanks,
Charles Doubleday
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20010509/dde221c7/attachment.html>