[molpro-user] how to use orbital from displaced geometry with frozen card

[Kirk Peterson]

I agree with Wanyi.  If your HF doesn't converge it doesn't make sense to use these as core orbitals in the CASSCF.   Either used "closed" in your CASSCF (and then "core" in the subsequent CASPT2) or design a smaller active space CASSCF to generate orbitals that can be used as frozen core in your larger CASSCF calc (then you can save the former to a record like 2100.2 and specify those on your frozen card in the latter).

best,

-Kirk


On Aug 22, 2012, at 10:36 AM, Wanyi Jiang <wanyi.jiang at gmail.com> wrote:

> Amit,
> 
> If HF does not converge and you have to use multi even if it takes
> longer time. In the caspt2 calculations, you may freeze these closed
> orbitals (keyword: core).
> 
> Wanyi
> 
> On Wed, Aug 22, 2012 at 12:25 PM, amit sharma <amit0206.s at gmail.com> wrote:
>> Kirk,
>> I don't want to use HF orbitals, HF does not converge in my case (which
>> should be resolved separately). When I explicitly give record number on
>> frozen card then I already should have the orbitals to be read in. How can I
>> get orbitals in record 2100.2 (say) at displaced geometry?
>> I tested it by performing a MULTI step with closed card, followed by MULTI
>> with frozen card, it works, but there is no point in doing that, is there
>> another way?
>> 
>> Wanyi,
>> closed orbitals are optimized (if I remember correctly) meaning longer
>> runtime, 100's of seconds vs few second for each MCSCF iteration.  I was
>> hoping to runs caspt2 geometry optimization where the displaced geometry
>> used pervious orbitals.
>> 
>> Amit
>> 
>> 
>> On Wed, Aug 22, 2012 at 7:15 AM, Wanyi Jiang <wanyi.jiang at gmail.com> wrote:
>>> 
>>> Dear Amit,
>>> 
>>> I am not sure if this is a good answer to your question. But why not
>>> using" closed" instead of "frozen" for the core orbitals?
>>> 
>>> Wanyi
>>> 
>>> On Tue, Aug 21, 2012 at 2:56 PM, amit sharma <amit0206.s at gmail.com> wrote:
>>>> Hi all,
>>>> I am doing MCSCF calculation inside a do-loop on a series of displaced
>>>> geometries. The orbital are
>>>> {multi
>>>> forzen,42,6,
>>>> occ,42,8
>>>> wf,98,1,0
>>>> state,2}
>>>> The first calculation runs fine, however for the second one, I get error
>>>> ?CORE ORBITAL RECORD  2140.2 CORRESPONDS TO DIFFERENT GEOMETRY
>>>> Can anyone please tell me how to fix this.  Molpro manual talks a bit on
>>>> this but it is not very clear to me
>>>> 
>>>> 2) Is there a way to start with atomic density guess orbital without
>>>> doing
>>>> HF;  in other words can I use
>>>> start,atdens in multi ?
>>>> 
>>>> I tired deleting file 1 and 2 as before the end of the loop, but next
>>>> iteration was still looking for pervious orbital file and I get this
>>>> error.
>>>> 
>>>> ?CORE ORBITAL RECORD  2140.2 NOT FOUND
>>>> 
>>>> Thanks
>>>> Amit
>>>> 
>>>> 
>>>> _______________________________________________
>>>> Molpro-user mailing list
>>>> Molpro-user at molpro.net
>>>> http://www.molpro.net/mailman/listinfo/molpro-user
>> 
>>