[molpro-user] Is it possible to do geometry optimization "by hands" via a do-loop

[Evgeniy Gromov]

Dear Professor Werner,

Thank you for your response to my question. There is however
some uncertainty left concerning updating the (model) Hessian.
If I use the following cycle for geometry optimization:

do i=1,20
label:
...
(here are procedures to calculate the energy to be (numerically)
differentiated)
...
forces,startcmd=label:, ... (and so on)
...
optg,maxit=1
end do
---

will the hessian be updated in this case?
Sorry my re-asking you but I am concerning that OPTG
will generate a new hessian in each do-loop step (because it is
always like the first optimization step for it), won't it?
Many Thanks!

Best regards,
Evgeniy

Hans-Joachim Werner wrote:
> The OPT command does not exist any more. Of course the hessian is updated
> in each step, and it is also possible to use more accurate hessians (either
> by computing it during the optimization or using a previously computed 
> hessian).
> These hessians can be computed at lower level than the energy/gradient 
> (see manual).
> You can also use the STARTCMD or PROCEDURE options to construct your own
> more complicated procedure for the energy calculation (this is more or less
> like using a do-loop). In case of convergence problems it is sometimes
> helpful to use {optg;method,diis}
> 
> Am 08.11.2009 um 10:48 schrieb Evgeniy Gromov:
> 
>> Dear Developers and Users of Molpro,
>>
>> I am wondering whether it is possible to do geometry optimization
>> via a do-loop, without using "optg". I mean, in some previous version of
>> molpro there was the command "opt" which just calculated next geometry
>> from (already) available gradients calculated before "opt" with
>> "forces". In recent versions there is only the "optg" command which
>> crashes if one evaluates gradients inside the optg loop. One could
>> of course specify maxit=1 for optg, i.e.
>>
>> optg,maxit=1
>>
>> and then organize a loop using "do". In this case
>> I however suspect that "optg" might use a new crude model
>> Hessian at each optimization step, so that the Hessian
>> will be not updated which will result in poor convergence.
>> So, I will very appreciate if someone knows how to do what
>> geometry optimization "by hands", without automatic "optg".
>> Many thanks!
>>
>> Best regards,
>> Evgeniy
>>
>> --_______________________________________
>> Dr. Evgeniy Gromov
>> Theoretische Chemie
>> Physikalisch-Chemisches Institut
>> Im Neuenheimer Feld 229
>> D-69120 Heidelberg
>> Germany
>>
>> Telefon: +49/(0)6221/545263
>> Fax: +49/(0)6221/545221
>> E-mail: evgeniy at pci.uni-heidelberg.de
>> _______________________________________
>>
>>
>>
>> _______________________________________________
>> Molpro-user mailing list
>> Molpro-user at molpro.net
>> http://www.molpro.net/mailman/listinfo/molpro-user
> 
> 


-- 
_______________________________________
Dr. Evgeniy Gromov
Theoretische Chemie
Physikalisch-Chemisches Institut
Im Neuenheimer Feld 229
D-69120 Heidelberg
Germany

Telefon: +49/(0)6221/545263
Fax: +49/(0)6221/545221
E-mail: evgeniy at pci.uni-heidelberg.de
_______________________________________