[molpro-user] Extended Multi-State CASPT2
Alan Chien
alandc at umich.edu
Thu Jul 3 21:32:00 BST 2014
Just a follow-up to my previous post.
Based on a number of runs, and looking at the number of CSF's that come
out, I've convinced myself that the double excitations are from the
CLOSED-CORE orbitals to all virtuals.
The definition of "active space" in Molpro's CASPT2 (RS2) program seems to
just be specific to the RS2 program itself.
The definition of active space for RS2 and RS2C is laid out in slightly
clearer language in the "quickstart" portion of the online manual.
http://www.molpro.net/info/2012.1/doc/quickstart/node24.html
Quoted below in case the online manual changes in the future
> In case of rs2 and rs3, exactly the same configuration spaces as in the
> MRCI are used. In this case the excitations with two electrons in the
> external orbital space are *internally contracted*. The total number of
> correlated orbitals is restricted to 16 for machines with 32-bit integers
> and to 32 for machines with 64-bit integers.
>
> In the rs2c case certain additional configuration classes involving
> internal and semi-internal excitations are also internally contracted (see
> J. Chem. Phys. *112*, 5546 (2000)). This is exactly as in the mrcic case
> (see above). This method is much more efficient than rs2 and more
> suitable for large cases. In particular, in this case only the number of
> *active* orbitals is restricted to 16 or 32 on 32 and 64 bit machines,
> respectively, and any number of closed-shell (inactive) orbitals can be
> used (up to a maximum as defined by a program parameter).
>
I've pretty much exhausted my questions on the limitations/capabilities of
Molpro's CASPT2 program.
Final question: Why is there a limit of 32 correlated orbitals in RS2? I
fiddled with the source code a bit, and it doesn't seem to be just an
arbitrary number set when the program was written - but the reason is not
quite transparent from a brief look at the code in src/mrci/.
On Wed, Jul 2, 2014 at 12:51 PM, Alan Chien <alandc at umich.edu> wrote:
> Thank you for your reply Werner, its cleared up a lot of my initial
> questions.
> I had actually determined that there was a difference between CORE and
> FROZEN, but was still trying to figure things out a bit better.
>
> I have a few clarification questions. Please forgive the length, but I'm
> trying to be as precise as possible so that no further clarifications will
> be necessary, heh.
>
> For the CAS portion of CASPT2 (separate from the MULTI calc), the
> reference determinants are generated using the orbitals defined by OCC -
> CLOSED.
> For the PT2 portion, we have to define a space from which we can do
> doubles excitations, i.e. correlate certain orbitals. This is defined as
> those orbitals in CLOSED - CORE. *But how do we define where those double
> excitations go to*? Do they only go to the empty orbitals that we
> included in OCC? Or are they automatically allowed to be excited to any/all
> virtuals (Which seems unlikely since virtuals usually make up a large
> portion of the MOs).
>
> CASPT2 (RS2) defines "active space" as both the space used to create the
> CAS determinants *as well as* the orbitals which we allow to be
> correlated.
> I find this strange as before now, I've only ever worked with CAS/RAS, and
> I'm used to thinking of active spaces as just the orbital set used to
> generate the set of reference determinants. Is this definition of active
> space a quirk of Molpro or is it the accepted standard when discussing
> CASPT2?
>
> Finally, does RS2C treat the CORE,CLOSED,OCC cards differently when making
> its definition of the "active space"? This link (which I posted earlier),
> http://www.molpro.net/pipermail/molpro-user/2006-April/001759.html, seems
> to suggest that it does. But I am unclear as to why that may be as the
> manual states that the RS2 and RS2C programs are essentially the same, just
> one is faster or has different features implemented compared to the other.
> Perhaps it is just a difference in the code of each program instead of a
> difference having any substantive meaning?
>
> I understand that some of these questions I can answer myself (namely the
> last one) with a couple of Molpro test jobs, and in lieu of provided
> answers, I may well determine them myself.
>
> But any help is welcome,
> Alan Chien
>
>
>
> On Wed, Jul 2, 2014 at 6:12 AM, Werner Győrffy <
> gyorffy at theochem.uni-stuttgart.de> wrote:
>
>> Dear Alan,
>>
>> CLOSED orbitals are doubly occupied in all reference configurations but
>> excitations are allowed from them in the CASPT2 calculations if they are
>> not frozen. If you want to define orbitals which are frozen (not
>> correlated) in CASPT2, you should use CORE orbitals. CORE orbitals are not
>> counted for the 32 orbital limitation in the rs2 program.
>>
>> Active orbitals in CASSCF are defined as OCC minus CLOSED orbitals.
>>
>> An (X)MS-CASPT2 calculation is still possible only with the rs2 program
>> not with rs2c.
>>
>> Regards,
>>
>> Werner Gyorffy.
>>
>>
>>
>>
>> On 07/01/2014 11:56 PM, Alan Chien wrote:
>>
>>> Interesting nugget of information. The example given in the multi-state
>>> CASPT2 section of the manual DOES change its active space when you
>>> change the numbers in the closed card.
>>> So it seems that for that particular example it can recognize closed
>>> input.
>>>
>>> Example: http://www.molpro.net/info/current/examples/lif_mr_mscaspt2.com
>>> (I edited it so that it only did r=3 ang to reduce the amt of output I
>>> had to look at. Also removed everything below RS2 section)
>>>
>>> Trying to replicate it now. In fact, been trying for a while.
>>>
>>>
>>>
>>>
>>> On Tue, Jul 1, 2014 at 3:48 PM, Alan Chien <alandc at umich.edu
>>> <mailto:alandc at umich.edu>> wrote:
>>>
>>> Dear Molpro Community,
>>>
>>> I was wondering if there was something wrong with my following input
>>> for an extended multi-state CASPT2 calculation. My input is not
>>> recognizing the CLOSED card, and so is throwing the error "TOO MANY
>>> ACTIVE ORBITALS: 48 THIS VERSION ALLOWS 32." An old thread mentions
>>> that I should use RS2C -
>>> http://www.molpro.net/pipermail/molpro-user/2006-April/001759.html -
>>> which should recognize closed orbitals.
>>>
>>> But it's a rather old thread, and I want XMS-CASPT2 which isn't
>>> implemented for RSC2, and so I thought it worth mentioning to see if
>>> they had updated the RS2 code and perhaps my input was just off. The
>>> manual makes no mention of RS2 being unable to recognize the CLOSED
>>> card. I also tried the same input with FROZEN and that threw the
>>> same error.
>>>
>>> Any input welcome,
>>> Alan Chien
>>>
>>>
>>> ***,title
>>> memory,400,m
>>> symmetry
>>> geometry={
>>> S 1.9692714536 0.9837700026 0.0000000000
>>> C 0.6745300892 -0.1941173367 0.0000000000
>>> C 1.1941024748 -1.4906632561 0.0000000000
>>> C 2.5920412491 -1.5265017306 0.0000000000
>>> C 3.1677860531 -0.2809638443 0.0000000000
>>> C -0.6745300892 0.1941173367 0.0000000000
>>> C -1.1941024748 1.4906632561 0.0000000000
>>> C -2.5920412491 1.5265017306 0.0000000000
>>> C -3.1677860531 0.2809638443 0.0000000000
>>> S -1.9692714536 -0.9837700026 0.0000000000
>>> C -4.5811338904 -0.0064991780 0.0000000000
>>> C -5.5229897499 1.0602944736 0.0000000000
>>> N -6.2538051640 <tel:2538051640> 1.9321608642
>>>
>>> 0.0000000000
>>> C 4.5811338904 0.0064991780 0.0000000000
>>> C 5.5229897499 -1.0602944736 0.0000000000
>>> N 6.2538051640 <tel:2538051640> -1.9321608642
>>>
>>> 0.0000000000
>>> C 5.0766433381 1.3388671764 0.0000000000
>>> N 5.4513787354 2.4133206959 <tel:4133206959>
>>>
>>> 0.0000000000
>>> C -5.0766433381 -1.3388671764 0.0000000000
>>> N -5.4513787354 -2.4133206959 <tel:4133206959>
>>> 0.0000000000
>>> H -3.1636666579 2.4343293653 <tel:4343293653>
>>> 0.0000000000
>>> H -0.5803308458 <tel:5803308458> 2.3697711304
>>> 0.0000000000
>>> H 0.5803308458 <tel:5803308458> -2.3697711304
>>> 0.0000000000
>>> H 3.1636666579 -2.4343293653 <tel:4343293653>
>>>
>>> 0.0000000000
>>> }
>>>
>>> basis=6-31G*
>>> hf
>>> {casscf;
>>> closed,31,6,30,5 !core orbs, total 72
>>> occ,31,8,30,7 !active orbs, total 76
>>> wf,148,1,0;state,2
>>> wf,148,3,0
>>> pspace,100.0
>>> }
>>>
>>> {rs2,xms=1,mix=2;
>>> state,2,2,3
>>> }
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> 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
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20140703/2f480dbd/attachment-0001.html>
More information about the Molpro-user
mailing list