### Author Topic: partial geometry optimization in cartesians?  (Read 6597 times)

#### resofidentity

• Full Member
• Posts: 65
• Karma: +0/-0
##### partial geometry optimization in cartesians?
« on: January 18, 2007, 09:13:37 am »
Hello TM forum members,

is there any way to optimize, lets say the position of one atom in a large (200 Atoms) containig structure. I have problems, since:
1. iaut fails and manual completeion of internals is too complicated (for me)
2. ired is too cryptic, the atom of interest is part of too many redundant internals
3. i am not aware of a way to fix certain cartesian coordinates and to relieve others for gradient calculations (like k,f,d ...)

Is there any easy way out?

PS.: I welcome the idea of TURBOMOLE forum very much!

#### Marek Sierka

• Developers
• Jr. Member
• Posts: 14
• Karma: +0/-0
##### Re: partial geometry optimization in cartesians?
« Reply #1 on: January 18, 2007, 09:42:22 am »
Hello and thank you for the first post!

As far as I know it is not possible to freeze Cartesian coordinates of atoms when optimizing
using internal coordinates (it is still on To Do list). However, if you use Jobex/Relax combination
for structure optimization you can freeze Cartesian coordinates when optimizing in Cartesian
coordiantes. You can do it by specifying "f" at the end of the coodinate line in the "coord" file, i.e.

\$coord
0.0 0.0 0.0  h  f
....
\$end

I know it is slower, but you can preoptimize structure using some cheaper method (RI-DFT)
and/or smaller basis set, and do the refining optimization with your final method.

In the new release of T-Mole (5.9) it is also possible to freeze only chosen Cartesian components
using Statpt/Jobex combination. This can be done by specifying three integers at the end of the
coordinate line (0 = frozen, 1 = optimized), e.g.

\$coord
0.0 0.0 0.0  h  1 0 1
....
\$end

will freeze only the y coordinate of the atom.

Best Wishes

mas
« Last Edit: January 18, 2007, 03:05:45 pm by turbomaster »

#### resofidentity

• Full Member
• Posts: 65
• Karma: +0/-0
##### Re: partial geometry optimization in cartesians?
« Reply #2 on: January 29, 2007, 09:45:37 am »
Hello,

thank You very much, good to know d,f,... works also for cartesians! It is not always easy, to find certain subjects in the manual by simply searching terms or checking the table of contents and the index. Another example for such a case is the very useful tool screwer. I tried to search "jump" or "saddle point" but did't succed until I read almost half of chapter 3 again. In this respect the fascinatingly efficient TM may be a little bit user unfriendly. Thanks for help!

R