[molpro-user] Gradients with integral direct calculations

[Werner Gyorffy]

Dear Gabriel,

Analytical gradients for coupled cluster methods cannot currently be 
computed in an integral-direct manner in Molpro. This is due to that the 
integral-direct option is not available for the so-called CP-HF 
calculation step in the case of coupled cluster methods. You may however 
use the DIRECT directive with numerical gradients.

Regards,

Werner Gyorffy.

On 06/24/2013 03:05 AM, Gabriel Freitas wrote:
> Dear Molpro users / developers,
>
> I'm trying to perform gradient calculations using gdirect option
> (integral direct). Although single-point energies in the CCSD level runs
> just fine, the program crashes when I ask for gradients/geometry
> optimization. I've succeeded doing analytical gradients in HF or MP2
> level with gdirect.
>
> If I don't use gdirect but only use the "direct" option in the ccsd
> input line, it doesn't crash, but .log file indicates that integrals are
> calculated and stored for every new geometry. I don't know exatcly if
> the CCSD code really reads this integrals in the disk for this case, but
> my impression is that it does. In other words, it's not a integral
> direct calculation.
>
> Does anyone know if it's possible to really perform an integral direct
> gradient for CCSD/QCI method?
>
> Thanks in advance,
>
> Gabriel
>
>
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user
>