[molpro-user] Lagrangian multipliers in CC code

Tue May 13 13:48:43 BST 2014

Dear Rodrigo,

Yes, you are right, the first part of the output is for double amplitudes, and
the second part for single amplitudes.

The lines:
          call repcom(1)
          call print_t1 (q(iu1), 'L(IA)')
          call print_t1_reset

          call print_t2 (q(iuij), 'L(IJ, AB)', uadr, lencpy, .false.)
          call print_t2_reset
          call repcom(2)

put after all_druckm in ldip.F should do the job. Note that the Lambda 
amplitudes are used in the contravariant form, see my paper in Mol.Phys. 
http://dx.doi.org/10.1080/00268970110105424 
about details how the whole stuff is implemented.

Best wishes,

Tatiana

On Mon, 12 May 2014, ro cha wrote:

> Dear Molpro users and developers,
>
> Dear Tatiana,
>
> Thanks again for the quick response. I followed the second approach you suggested to print out the contents of the LAMBDA vector by triggering the debug mode of the all_druckm subroutine. Indeed, the output now reports:
>
> -----------------------------------------------------------------------------------
> Print all matrices for a given pair list Lambda
>
> MATRIX Lambda 101   (Symmetry=1 TYPE= 0)
> BLOCK (1,1)
> ..... (many numbers) <--- I guess these are the Lagrangian multipliers for the double amplitudes
>
> Singles part of Lambda
> OCC. SYM. BLOCK 1
> Occ.orb. 1.1
> .....(fewer numbers) <--- and for the single amplitudes
>
> ------------------------------------------------------------------------------------
>
> Could you please tell me whether or not I'm interpreting the output correctly? I would like to know which excitation is associated to each element of Lambda.
>
> I would love to have the multipliers printed out in the pretty format of the print_t1 and print_t2 routines (which would bring me back to the first approach you suggested below). Unfortunately, I haven't been able to guess which arguments to pass to these functions inside ldip.F just below the line "Saving Lambda vector..." in order to mimic this behaviour. Would you be kind enough as to guide me in this regard?
>
> Thanks in advance
>
> Rodrigo
>
>
>
>
>> Date: Sat, 10 May 2014 02:47:18 +0200
>> From: tania at tiger.chem.uw.edu.pl
>> To: crystalinflame at hotmail.com
>> CC: gyorffy at theochem.uni-stuttgart.de; molpro-user at molpro.net
>> Subject: Re: [molpro-user] Lagrangian multipliers in CC code
>>
>> Dear Rodrigo,
>>
>> The line
>>
>> Solve  a set of linear equations for Lambda...
>>
>> in the output means that you accessed my CCSD properties program, so it is my
>> turn to answer:-)
>>
>> The Lambda vector contains single (Lambda^i_a) and double (Lambda^{ij}_{ab})
>> amplitudes, and it is not possible to read it through matrop - it has the same
>> structure as the CCSD T amplitudes (T^i_a and T^{ij}_{ab}). If this is what you
>> want, then 1) you should read both T and Lambda from the same output, and 2) use
>> e.g. the subroutine print_t2 and print_t1 to look for their structure. For the
>> case of T you can use thresh,thrpri=0.001 (or a smaller threshold) to print the
>> amplitudes to the standard output, but for the case of Lambda you should add two
>> lines of code calling print_t1 and print_t2 in the file src/eom/ldip.F just
>> after the line containing "Saving Lambda vector on record" or alternatively use
>> the subroutine all_druckm just below this line by changing 'Lambda',0) to
>> 'Lambda',1)  - the latter version turns on the debug print of all Lambda
>> amplitudes and the molpro.exe compiled in this way shouldn't be used for large
>> production calculations.
>>
>> Best wishes,
>>
>> Tatiana
>>
>> On Thu, 8 May 2014, ro cha wrote:
>>
>>> Dear Molpro users and developers,
>>>
>>> Dear Werner
>>>
>>> Thanks for the quick response. I realize MOLPRO stores solutions of the CPHF equations whenever a relaxed CC properties calculation is requested as in:
>>>
>>> {ccsd
>>>      core, 0
>>>      expec, relaxed, dm
>>> }
>>>
>>> Indeed, the output file reports:
>>>
>>> Solve  a set of linear equations for Lambda...
>>> Saving Lambda vector on record 4499.8
>>>
>>> However, I haven't been able to access this record successfully using the MATROP directive. Do you know how to actually print the contents of this binary record in the output file?
>>>
>>> Best regards
>>>
>>> Rodrigo
>>>
>>>
>>>> Date: Wed, 30 Apr 2014 13:28:29 +0200
>>>> From: gyorffy at theochem.uni-stuttgart.de
>>>> To: molpro-user at molpro.net
>>>> Subject: Re: [molpro-user] Lagrangian multipliers in CC code
>>>>
>>>> Dear Rodrigo,
>>>>
>>>> Analytical energy gradients are currently available for various coupled
>>>> cluster (CC) methods in Molpro: QCISD, QCISD(T), CCD, and CCSD. One can
>>>> print solutions of the CP-HF equations, relaxed- and nonrelaxed density
>>>> matrices by using the MATROP card (see details in the manual). Solutions
>>>> of the CP-HF equations are stored automatically. Density matrices can be
>>>> stored by using the DM card (see details in the manual).
>>>>
>>>> Some options may only work by using a recent update to the current
>>>> version of Molpro ( See details on
>>>> https://www.molpro.net/download/index.php?version=2012.1 ).
>>>>
>>>> Regards,
>>>>
>>>> Werner.
>>>>
>>>> On 04/27/2014 07:52 PM, ro cha wrote:
>>>>> Dear Molpro developers and users,
>>>>>
>>>>> I need to extract the Lagrangian multipliers of the CC code to further
>>>>> post-process the relaxed densities using a back-end program developed in
>>>>> our research group. However, I have not found any reference in the
>>>>> user's manual on this regard. Does anyone know how to get MOLPRO to
>>>>> print out the lagrangian multipliers?
>>>>>
>>>>> Thanks in advance.
>>>>>
>>>>> Rodrigo Ch�vez Calvillo
>>>>> Facultad de Qu�mica, UNAM
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> 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
>>>
>>
>> Dr. Tatiana Korona http://tiger.chem.uw.edu.pl/staff/tania/index.html
>> Quantum Chemistry Laboratory, University of Warsaw, Pasteura 1, PL-02-093 Warsaw, POLAND
>

Dr. Tatiana Korona http://tiger.chem.uw.edu.pl/staff/tania/index.html
Quantum Chemistry Laboratory, University of Warsaw, Pasteura 1, PL-02-093 Warsaw, POLAND