[molpro-user] Getting rid of limit of 32 active orbitals in molpro2012.1
Alexander Mitrushchenkov
Alexander.Mitrushchenkov at u-pem.fr
Thu Jun 1 15:06:00 CEST 2017
Hi Evgeniy,
As I wrote to Nike, I think it is more complicated than it seems as it
is hardwired in many places, and also included through some efficient
bit routines in C which I guess can not be done int*16. As far as I can
see, nobody from Molpro team (which is not that big) will do this soon.
Also because we can not really see the practical use of it. Already 32
active orbitals is far too heavy/undoable (many billions) for MRCI kind
of treatmeant with any imaginable computer in near future [unless you
have very few active electrons but this is not the case normally for
chemical species]. The only use would be perhaps for CASSCF with
different CI part, as FCIQMC or DMRG, but that would require full
rewriting of CASSCF code anyway, as the nonlinear optimization
algorithms of CASSCF are strongly bound to classical CI approach. If
this new CASSCF will ever be done, there will be no restrictions of course.
CIC code is being actively developed, and I suppose will hopefully be
much more functional in the new official release, including many-state
calculation via projection, and properties.
But if you have the source code license, of course you can try, and
eventually share your experience on that.
Best,
-Sasha
Le 01/06/2017 à 14:44, Evgeniy Gromov a écrit :
> Dear Sasha,
>
> Thanks for your detailed reply. It is annoying that the new MRCIC code
> has at present very limited
> functionalities. I am just curious if one could try the suggestion of
> Nike, which is to replace
> int8 by int16 in the code, in order to overcome the 32-barrier of the
> old MRCI code?
> Didn't someone try this before?
>
> Kind regards,
> Evgeniy
>
> On 05/31/2017 11:52 AM, Alexander Mitrushchenkov wrote:
>> Hi,
>>
>> The limit of 32 active orbitals is intrinsic for all orbital
>> configuration space algorithms implemented in Molpro as it is based
>> on integer*8 bit level routines (so it is 16 for integer*4
>> compilations), and as such it will NEVER be changed, unless the codes
>> are fully rewritten. By, now new codes are being developed which
>> allow any nulber of CLOSED orbitals (but the number of ACTIVE is
>> always limited to 32). This currently only applies to RS2C code, and
>> new MRCIC (CIC) codes. These, however, still have more limited
>> functionalities as compared to RS2 or MRCI codes. For example, CIC
>> code can calculate only one (ground) state, there is no multi-state
>> RS2C either, etc. All codes, based on MRCI, like RS2,RS3,ACPF, etc,
>> have the limit of VALENCE (ACTIVE+CLOSED) orbitals of 32, and I say I
>> do not believe this will ever change.
>>
>> Best,
>>
>> -Sasha
>>
>> Le 26/05/2017 à 16:50, Evgeniy Gromov a écrit :
>>> Dear Sham,
>>>
>>> Thanks very much for your reply. Yes, I meant the previously
>>> existing limit for the number
>>> of closed shell inactive but correlated orbitals in MRCI. This is
>>> supposed to have been removed in molpro2012.
>>> However, I still get this message "TOO MANY ACTIVE ORBITALS: ...",
>>> when running a mrci calculation
>>> with molpro2012.1.
>>>
>>> My input for mrci is:
>>>
>>> {ci;wf,86,1,0;save,3010.1;state,6;noexc}
>>> {ci;wf,86,1,2;save,3020.1;state,6;noexc}
>>> !
>>> lsint
>>> !
>>> {ci;hlsmat,ls,3010.1,3020.1}
>>>
>>> I tried to replace "ci" with "mrcic" as you suggested but the job
>>> crashed with an error,
>>> which indicated that it doesn't "know" the keyword "mrcic". Also in
>>> the manual
>>> I didn't find such a keyword, though it is mentioned that there is
>>> "a new internally
>>> contracted MRCI code: MRCIC ...". Do you have any example of input
>>> with "mrcic"
>>> keyword?
>>>
>>> Best regards,
>>> Evgeniy
>>>
>>>
>>> On 05/24/2017 11:29 AM, Sham K wrote:
>>>> Dear Evgeniy,
>>>>
>>>> There is no compiler option for this. The limit of 32 correlated
>>>> internal orbitals (closed + active) exist for methods RS2 and MRCI
>>>> in molpro. Two new related methods, RS2C and MRCIC, have been also
>>>> available in molpro. For these programs, the limit is now 32 active
>>>> orbitals and there is no limit on number of closed orbitals. The
>>>> new MRCIC program works only for one state (typically ground state
>>>> with multi-reference character).
>>>>
>>>> You can use them with {rs2c;...} or {mrcic;...} just like old programs.
>>>>
>>>> Best,
>>>> Sham
>>>>
>>>>
>>>> On Tue, May 23, 2017 at 7:34 PM, Evgeniy Gromov
>>>> <evgeniy.gromov at pci.uni-heidelberg.de
>>>> <mailto:evgeniy.gromov at pci.uni-heidelberg.de>> wrote:
>>>>
>>>> Dear Developers and Users of Molpro,
>>>>
>>>> If I am not mistaken, the limit of 32 active orbitals is
>>>> supposed to have been (quasi) removed in molpro 2012.1.
>>>> I vaguely remember that some trick is however needed before or
>>>> during the compilation to actually get rid
>>>> of that limit. I will very much appreciate a hint for that trick.
>>>> Many thanks!
>>>>
>>>> Best regards,
>>>> Evgeniy
>>>>
>>>> --
>>>>
>>>> Dr. Evgeniy Gromov
>>>> Theoretische Chemie
>>>> Physikalisch-Chemisches Institut
>>>> Im Neuenheimer Feld 229
>>>> D-69120 Heidelberg
>>>> Germany
>>>>
>>>> Telefon: +49/(0)6221/545213 <tel:%2B49%2F%280%296221%2F545213>
>>>> Fax: +49/(0)6221/545221 <tel:%2B49%2F%280%296221%2F545221>
>>>> E-mail: evgeniy.gromov at pci.uni-heidelberg.de
>>>> <mailto:evgeniy.gromov at pci.uni-heidelberg.de>
>>>>
>>>> _______________________________________________
>>>> Molpro-user mailing list
>>>> Molpro-user at molpro.net <mailto:Molpro-user at molpro.net>
>>>> http://www.molpro.net/mailman/listinfo/molpro-user
>>>> <http://www.molpro.net/mailman/listinfo/molpro-user>
>>>>
>>>>
>>>
>>> --
>>> Dr. Evgeniy Gromov
>>> Theoretische Chemie
>>> Physikalisch-Chemisches Institut
>>> Im Neuenheimer Feld 229
>>> D-69120 Heidelberg
>>> Germany
>>>
>>> Telefon: +49/(0)6221/545213
>>> Fax: +49/(0)6221/545221
>>> E-mail:evgeniy.gromov at pci.uni-heidelberg.de
>>>
>>>
>>> _______________________________________________
>>> Molpro-user mailing list
>>> Molpro-user at molpro.net
>>> http://www.molpro.net/mailman/listinfo/molpro-user
>>
>>
>> --
>> Dr. Alexander Mitrushchenkov, IGR
>> Laboratoire de Modélisation et Simulation Multi Echelle
>> UMR 8208 CNRS
>> Université Paris-Est Marne-la-Vallée
>> 5 Bd Descartes
>> 77454 Marne la Vallée, Cedex 2, France
>>
>> Phone: +33(0)160957316
>> Fax: +33(0)160957320
>> e-mail:Alexander.Mitrushchenkov at u-pem.fr
>
> --
> Dr. Evgeniy Gromov
> Theoretische Chemie
> Physikalisch-Chemisches Institut
> Im Neuenheimer Feld 229
> D-69120 Heidelberg
> Germany
>
> Telefon: +49/(0)6221/545213
> Fax: +49/(0)6221/545221
> E-mail:evgeniy.gromov at pci.uni-heidelberg.de
--
Dr. Alexander Mitrushchenkov, IGR
Laboratoire de Modélisation et Simulation Multi Echelle
UMR 8208 CNRS
Université Paris-Est Marne-la-Vallée
5 Bd Descartes
77454 Marne la Vallée, Cedex 2, France
Phone: +33(0)160957316
Fax: +33(0)160957320
e-mail: Alexander.Mitrushchenkov at u-pem.fr
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20170601/99d37c7b/attachment-0001.html>
More information about the Molpro-user
mailing list