[molpro-user] CCSD natural orbitals

[Werner Győrffy]

I have fixed the bug. The correction will be available in Molpro soon. 
Thank you for reporting the problem.

Regards,

Werner.

On 15.03.2017 10:07, Werner Győrffy wrote:
> Hi Alan,
>
> I have just realized that the CORE option in NATORB seems to be not
> working properly in case of MP2 or CCSD. Please do not use that until
> the problem is fixed.
>
> I would however recommend to always include core orbitals in the natural
> orbital calculations. When they are frozen in the correlated
> calculation, the natural orbitals and the canonical orbitals are
> identical with occupation number 2 in the case of using nonrelaxed
> density matrices. The presence of them in the natural orbital
> calculation does not affect any other natural orbitals. By using the
> relaxed density matrices, it is important to include the frozen core
> orbitals in the natural orbital calculation, as their occupation number
> is not 2 anymore.
>
> To print all natural orbitals, one should use the maximum number of
> virtual (external) orbitals in any irrep. In the given example, that
> would be PRINT=26, but you can give any larger number, too. For example
> PRINT=100.
>
> Regards,
>
> Werner.
>
> On 14.03.2017 18:37, Alan Chien wrote:
>> I'm toying with CCSD natural orbitals for butadiene as a starting point
>> for higher level calculations. Full input included at the end.
>>
>> {ccsd;
>> expec,type=norelax;
>> dm,2141.2;
>> natorb,record=2141.2,print=23,_core_=0;
>> }
>>
>> I'm confused as to the function of "_core_". The manual
>> (https://www.molpro.net/info/2015.1/doc/manual/node341.html#sec:natorb_mp2
>>
>> <https://www.molpro.net/info/2015.1/doc/manual/node341.html#sec:natorb_mp2>)
>>
>> states that core=1 freezes the core orbitals when constructing the
>> density matrix.
>>
>> If I'm already using the frozen-core approximation in my CCSD
>> calculation, which I am, why does setting core=1 change my NO occupancy
>> results? The core orbitals should not be coupled to the rest of the
>> orbitals at all in a non-relaxed density matrix, so it should not matter
>> if the core orbitals are included or not in the density matrix.
>>
>> Any insight appreciated,
>> Alan Chien
>>
>>
>>
>> _Input below_
>> _
>> _
>> ***,buta
>> geometry={
>>   C         0.6009903066   -0.4020172170    0.0000000000
>>   C        -0.6009903066    0.4020172170    0.0000000000
>>   C         1.8329498097    0.1103686968    0.0000000000
>>   C        -1.8329498097   -0.1103686968    0.0000000000
>>   H         1.9876059086    1.1777308428    0.0000000000
>>   H        -1.9876059086   -1.1777308428    0.0000000000
>>   H         0.4702284110   -1.4752930006    0.0000000000
>>   H        -0.4702284110    1.4752930006    0.0000000000
>>   H         2.7040838288   -0.5215644347    0.0000000000
>>   H        -2.7040838288    0.5215644347    0.0000000000
>> }
>> basis=roos(3s2p1d/2s1p) !anos-l-pvdz
>> {hf;
>> orbprint,5;
>> }
>> {ccsd;
>> expec,type=norelax;
>> dm,2141.2;
>> natorb,record=2141.2,print=23,core=0;
>> }
>> {put,molden,ccsd.molden;
>> orbital,record=2141.2,type=natural
>> }
>>
>>
>>
>>
>>
>>
>> _______________________________________________
>> Molpro-user mailing list
>> Molpro-user at molpro.net
>> http://www.molpro.net/mailman/listinfo/molpro-user
>>
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user