TURBOMOLE Users Forum

TURBOMOLE Modules => Aoforce and Numforce => Topic started by: quant_m on November 07, 2007, 12:45:12 PM

Title: How to force TM to obtain Hessian on unoptimized structures?
Post by: quant_m on November 07, 2007, 12:45:12 PM
Dear Colleagues,

I have faced a weird problem calculating Hessian with the numforce module (we have TURBOMOLE V5-7-1).
I need a Hessian for the transition search optimization, of course the starting guess geometry is not a stationary point.
The message in the numforce.out  states:

"""CHECKING GRADIENT NORM
   gradient norm                  : 0.13390370E-01
No stationary point of energy, therefore no 2nd derivatives, ...!
''''

To tackle this problem I reduced convergence criteria in the control as well as in the run script files according to

"""$statpt
   itrvec      0
   threchange  1.000000E-02
''''

(control file) and

"""jobex -ex -energy 2 -gcart 0 -c 1  -keep > jobex.out
NumForce -ex 1 -level rpa -central > numforce.out
'''''

(run script).

However, the result was zero (the message is just given above).

Any ideas how to force TM to calculate Hessian on unoptimized structures?


Kind regards,

quant_m








Title: Re: How to force TM to obtain Hessian on unoptimized structures?
Post by: uwe on November 07, 2007, 04:34:42 PM
Hello,

in the NumForce version of Turbomole 5.8 and newer there is a flag (-c) that switches off the gradient check and the threshold for the gradient norm check can be changed by using the option -thrgrd <real> (with Turbomole 5.9).

If you want NumForce of Turbomole 5.7 to ignore the gradient norm, you can change the NumForce script:

search NumForce for the lines:

Code: [Select]
grep 'gradient norm' out.getgrad.test
echo "No stationary point of energy, therefore no 2nd derivatives, "`whoami` "!"
 exit 2

and remove the

exit 2

line.

Regards,

Uwe
Title: Re: How to force TM to obtain Hessian on unoptimized structures?
Post by: quant_m on November 07, 2007, 05:40:57 PM
Thanks, I will try.