[molpro-user] multiple core specifications in one input

[Peter Knowles]

This is indeed a bug which was fixed in the development version of 
Molpro, but not backported onto 2002.6. This has now been attended to, 
and a fully patched 2002.6 should no longer show the problem.
Peter

Kirk Peterson said the following on 12/08/05 19:04:
> Brian,
> 
> I've also reproduced this.  This does not occur with the 2002.9 
> development version and I think the
> problem is that the integral transformation is not carried out in the 
> 3rd ccsd(t) and the operators
> are re-used from the core,0 calculation (as noted in the output) .  It 
> looks like the program has no problem if the number of
> core orbitals just decreases in subsequent calls, e.g., core,2 then 
> core,1 and then core,0.  It doesn't
> seem to expect the number of cores to go back and forth.  A quick hack 
> is to redo the rhf step before the 3rd ccsd(t)
> calculation.
> 
> -Kirk
> 
> On Aug 10, 2005, at 12:56 PM, Brian wrote:
> 
>> If you try something along the lines of
>>  
>> hf
>> ccsd(t)
>> core,1
>> ccsd(t)
>> core,0
>> ccsd(t)
>> core,1
>>  
>> for water, you will find that the first and last ccsd(t) values are 
>> different.  In fact, the second enery is lower than the correct 
>> energy.  Is this a known bug?
>>  
>> -Brian
> 
> 
> 

-- 
Prof. Peter J. Knowles
School of Chemistry, Cardiff University, Main Building, Park Place, 
Cardiff CF10 3AT, UK
Telephone: +44 29208 79182 Fax: +44 2920874030
Email KnowlesPJ at Cardiff.ac.uk  WWW 
http://www.cardiff.ac.uk/chemy/staff/knowles.html