saddle point optimization

Ramon Hernandez ceeh340 at imaff.cfmac.csic.es
Mon Aug 12 09:58:23 BST 2002


Hi,

 I am trying to calculate a saddle point geometry at a level beyond
CASSCF, say RS2 or MRCI. I obtain a hessian at the CASSCF level and
then try the optimization with an input that looks like:

 ....
 ....

  rs2c;
  optg;
  root,2;
  hstart,5300.2;

 The problem I have is that after calculating the gradient numerically I

was expecting the program to read the hessian and call the algorithm
that decides the geometrical step. Instead the program went forward to
calculate the hessian numerically !! This will take a much longer
 time and of course defeats the purpose of calculating the hessian in
advance ....

 Is there a way to do this properly with Molpro ?

  I have noticed looking at the manual that exactly the same
procedure(input)
works when the method used includes analytical first derivatives(say mp2
after rhf).
I see no reason why it should work here and not in the
 preceding case, although  I guess  that the modules that are called
for the optimization are different depending on whether analytical first

 derivatives are available or not ....


 Thank you

 Ramon








More information about the Molpro-user mailing list