TURBOMOLE Users Forum

TURBOMOLE Modules => Jobex: Structure Optimization and Molecular Dynamics => Topic started by: lasermichel on April 23, 2007, 01:47:17 PM

Title: Optimization using jobex in fixed internals
Post by: lasermichel on April 23, 2007, 01:47:17 PM

Dear all,

I try to perform a structure optimization of the first excited singlet state of phenol using the jobex script of turbomole version 5.8. The internals (which are generated automatically from cartesians using iaut and imet) are all set active (#k), despite one, which I set fixed (#f) by purpose to a certain value, because I want to displace the geometry along this coordinate, relaxing all remaining ones. Nevertheless, the optimization always leads to a completely relaxed molecule i.e. also the fixed coordinate is relaxed. What am I doing wrong?

Many thanks for your help.

Michael

Title: Re: Optimization using jobex in fixed internals
Post by: antti_karttunen on April 24, 2007, 03:04:57 PM

Hi,

have you checked if Turbomole is really performing the optimization using internal coordinates? For optimization in internal coordinates, the $optimize group in control file should be something like

$optimize
   internal   on
   redundant  off
   cartesian  off
   global     off
   basis      off   logarithm

Now, if internal coordinates have some problems, Turbomole might fall back to cartesian optimization, where the fixed internals are not taken into account (just a guess, I'm not sure about this).

Note that you can also first fix the internal coordinates you want using "idef" in internals menu and then create redundant internals in the geometry menu. The "ired" command will take the fixed internals into account when creating redundant internals. In my opinion this is the most painless and robust way, because "ired" seems sometimes to manage better than "iaut". More info on constrained optimizations can be found from the Turbomole tutorial http://www.cosmologic.de/data/Turbomole_Tutorial_59.pdf (http://www.cosmologic.de/data/Turbomole_Tutorial_59.pdf)

Title: Re: Optimization using jobex in fixed internals
Post by: lasermichel on April 24, 2007, 09:28:41 PM

Thanks a lot, I will try this.

It seems to be more straightforward than the workaround that I found:
$interconversion  on
  cartesian <-- internal coordinate
and then perform a relax run. After that change back in contol to $interconversion off and perform jobex. Works, but is annoying.

Again, many thanks.

Michael