[Columbus] Columbus Digest, Vol 128, Issue 3
Thomas Mueller
th.mueller at fz-juelich.de
Tue Mar 25 05:16:30 EDT 2025
!-------------------------------------------------------------------|
This Message Is From an External Sender
This message came from outside your organization.
|-------------------------------------------------------------------!
On Mon, 24 Mar 2025, Aleksandr Zaichenko via Columbus wrote:
Dear Aleksandr,
this looks quite weird. There is no way to produce a 1.2 TB diagonal
integral file ( this would amount to a basis of
approx. sqrt(1.2*10**12/(8))=38*10**4 ).
Since you have probably defined a reference space of 38760 CSFs,
the program should not try to use the standard dsyevx/dsyevd eigensolver
to extract the lowest root which would require (38760**2)*8 bytes of
memory plus say 20% overhead but instead use iterative reference space diagonalization.
The alternative is to just start from the unit vector with the smallest
diagonal matrix element - technically there is no difference.
CASCI is not really the target of the MR-CISD code - although of course
one can use it. The only charming feature is that the iterative CASCI
calculation is parallel (provided it is switched on).
BTW, for CASCI(20o,6e) without symmetry and S=0
the size of the CSF space should be
379050 CSFs (= 21!/(3!18!) * 21!/(4!17!) * 1/21
The alternative is to use the mcscf code, run a single macro iteration,
switch off orbital rotations (which ought also switch off the explicit
construction of the Hessian matrix components B,C and M) and set to
iterative solution of the CI problem. This should also force the AO-MO
trafo step to just compute the all-internal MOs.
Beyond about 10**6 CSFs a CASCI calculation with MR-CISD is less
useful, since the formula tape construction is getting more and more
dominant and the matrix element construction is scalar (i.e. not fast
but with sufficient patience you can go to 10**9 CSFs - but imagine
running FCI with a general MR-CISD code is not that efficient).
The MCSCF code is somewhat better positioned regarding this special
application. Despite of all shortcomings DMRG (especially CheMPS2) does
a good job here.
The documentation in $COLUMBUS/docs/fulldoc/ciudg.txt,
$COLUMBUS/docs/fulldoc/mcscf.txt hopefully clarifies the various
input options.
Best,
Thomas
> Dear Hans, Yes. Correct. It is the MR-CISD program. Excitations in the virtual space are 0, moreover all
> virtual orbitals as well as lower are frozen. The problem size from cidrtls is: total: 38760 and it seems
> correct. The growth of the memory
> Dear Hans,
>
> Yes. Correct. It is the MR-CISD program. Excitations in the virtual space are 0, moreover all virtual orbitals
> as well as lower are frozen.
>
> The problem size from cidrtls is:
>
> total: 38760
>
> and it seems correct. The growth of the memory demand occurs in the diagint file up to 1.2 Tb and calculation
> falls with SIGSEGV.
>
> It should not happen with the CI-problem of that dimension including only one root with SO, as I know. I can
> share listings, if you will need it.
>
> Best,
> Aleksandr
>
>
>
> On Mon, Mar 24, 2025 at 6:46 PM <columbus-request at lists.osc.edu> wrote:
> Send Columbus mailing list submissions to
> columbus at lists.osc.edu
>
> To subscribe or unsubscribe via the World Wide Web, visit
> https://lists.osu.edu/mailman/listinfo/columbus
> or, via email, send a message with subject or body 'help' to
> columbus-request at lists.osc.edu
>
> You can reach the person managing the list at
> columbus-owner at lists.osc.edu
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Columbus digest..."
>
>
> Today's Topics:
>
> 1. CASCI extraordinary memory demand (Aleksandr Zaichenko)
> 2. Re: CASCI extraordinary memory demand (Hans Lischka)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Mon, 24 Mar 2025 15:36:14 -0400
> From: Aleksandr Zaichenko <alex.zai90 at gmail.com>
> To: columbus at lists.osc.edu
> Subject: [Columbus] CASCI extraordinary memory demand
> Message-ID:
> <CAAJgZKtK7jg7Vij3mJuNcCrShToEQrxkgJZfQ=BKkgUz6Qvc9A at mail.gmail.com>
> Content-Type: text/plain; charset="utf-8"
>
> Dear Columbus Team,
>
> I have a question about an ressources requirement issue with the CI module
> for the CASCI calculation. A calculation with active space CAS(6,20)
> without any external excitations (all lower orbitals are frozen, all higher
> orbitals are removed) contains 38760 CI-vectors (as should be) and I should
> get an CI-only solution for only one first root. Most QC programs do this
> calculation without any problem, but the Columbus CI module requires more
> than 1 Tb memory for that and it seems very demanding. Why is it so? Do I
> understand correctly, that the program tries to construct a full-CI problem
> for all of the 38760 roots despite 1 root being required?
>
> Best,
> Aleksandr
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <http://lists.osu.edu/pipermail/columbus/attachments/20250324/b257d5cf/attachment-0001.html >
>
> ------------------------------
>
> Message: 2
> Date: Mon, 24 Mar 2025 17:45:30 -0500
> From: Hans Lischka <hans.lischka at univie.ac.at>
> To: columbus at lists.osc.edu
> Subject: Re: [Columbus] CASCI extraordinary memory demand
> Message-ID: <bc527a53-1a8d-4bea-b08a-30e2d1f800d1 at univie.ac.at>
> Content-Type: text/plain; charset="utf-8"; Format="flowed"
>
> Hi Aleksandr,
> I am not sure what kind of calculation you really did. The CI module is
> meant to do a MR-CISD calculation. I assume you took the CAS(6,20) as
> reference? How did you proceed further? The CI program will want to
> create all SD excitation from the 38760 configurations. Did you opt for
> that? This would explaining the 1TB of memory required. Look into the
> cidrtls file. There you will see the CI dimension. If you want the CASCI
> calculation you should input zero excitations (into the virtual space).
> Please let e know what you did.
> Best regards, Hans
>
> On 3/24/2025 2:36 PM, Aleksandr Zaichenko via Columbus wrote:
> > Dear Columbus Team, I have a question about an ressources requirement
> > issue with the CI module for the CASCI calculation. A calculation with
> > active space CAS(6,20) without any external excitations (all lower
> > orbitals are frozen, all higher orbitals
> > Dear Columbus Team,
> >
> > I have a question about an ressources requirement issue with the CI
> > module for the CASCI calculation. A calculation with active space
> > CAS(6,20) without any external excitations (all lower orbitals are
> > frozen, all higher orbitals are removed) contains 38760 CI-vectors (as
> > should be) and I should get an CI-only solution for only one first
> > root. Most QC programs do this calculation without any problem, but
> > the Columbus CI module requires more than 1 Tb memory for that and it
> > seems very demanding. Why is it so? Do I understand correctly, that
> > the program tries to construct a full-CI problem for all of the 38760
> > roots despite 1 root being required?
> >
> > Best,
> > Aleksandr
> >
> > _______________________________________________
> > Columbus mailing list
> > Columbus at lists.osc.edu
> > https://lists.osu.edu/mailman/listinfo/columbus
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <http://lists.osu.edu/pipermail/columbus/attachments/20250324/06c785be/attachment.html >
>
> ------------------------------
>
> Subject: Digest Footer
>
> _______________________________________________
> Columbus mailing list
> Columbus at lists.osc.edu
> https://lists.osu.edu/mailman/listinfo/columbus
>
> ------------------------------
>
> End of Columbus Digest, Vol 128, Issue 3
> ****************************************
>
>
>
-----------------------------------------------------------
Dr. Thomas Mueller
Institute for Advanced Simulation (IAS)
Juelich Supercomputing Centre (JSC)
Phone: +49-2461-61-3175
Fax: +49-2461-61-6656
E-mail: th.mueller at fz-juelich.de
WWW: https://urldefense.com/v3/__http://www.fz-juelich.de/jsc__;!!KGKeukY!zO-I_AEdCwT9hDgQtJLTGGNTW4ZquFXgNfSh1FHQg4ZrM2BGCqdvg3cUzMp472SleksI9IyfNmqdzEAmIAbJKxO1LCq2JPs$
JSC is the coordinator of the
John von Neumann Institute for Computing (NIC)
and member of the
Gauss Centre for Supercomputing (GCS)
-----------------------------------------------------------
More information about the Columbus
mailing list