tce mrcc readref

From NWChem

Viewed 1245 times, With a total of 2 Posts
Jump to: navigation, search

Clicked A Few Times
Threads 18
Posts 44
Hi,

I'm giving the MRCC capability of TCE a try. I've set up a job configured with

           General Information
-------------------
Number of processors : 144
Wavefunction type : Restricted open-shell Hartree-Fock
No. of electrons : 240
Alpha electrons : 120
Beta electrons : 120
No. of orbitals : 3234
Alpha orbitals : 1617
Beta orbitals : 1617
Alpha frozen cores : 108
Beta frozen cores : 108
Alpha frozen virtuals : 1350
Beta frozen virtuals : 1350
Spin multiplicity : singlet
Number of AO functions : 1620
Number of AO shells : 504
Use of symmetry is : on
Symmetry adaption is : on
Schwarz screening : 0.10D-09

The idea is to freeze 90% of the lowest occupied and highest virtual orbitals, and look at memory requirements as I successively unfreeze more. As configured, the reference active space should be (4,4), and the CC active space should have 1617-1350-108 = 159 orbitals and (0.1*120 - 4/2) = 10 electrons outside of the reference active space, for each spin manifold.

The input mrccdata information looks like
mrccdata
  root 2
  cas 4 4
  subgroupsize 4
  improvetiling
end

Per the documentation, the subgroupsize was set based on 36 references and 144 MPI ranks --> 144/36 = 4.
I can understand the error from the output, but I don't know what the output is telling me:
  !! WARNING !! The number of MO is less than the number of AO

          Correlation Information
          -----------------------
          Calculation type : State Specific Multireference Brillouin-Wigner CCSD theory
   Perturbative correction : none
            Max iterations :      100
        Residual threshold : 0.10D-06
     T(0) DIIS level shift : 0.00D+00
     L(0) DIIS level shift : 0.00D+00
     T(1) DIIS level shift : 0.00D+00
     L(1) DIIS level shift : 0.00D+00
     T(R) DIIS level shift : 0.00D+00
     T(I) DIIS level shift : 0.00D+00
   CC-T/L Amplitude update :  5-th order DIIS
                I/O scheme : Global Array Library

Active elec :   4
Active orbs :   4


Alpha 1100
Beta  1100

Total no. of references:    0</b>

MRCC module - reference analysis

Fermi vacuum according to SCF wave function  has been requested.
 ------------------------------------------------------------------------
 tce_mrcc_readref                   3
 ------------------------------------------------------------------------
 ------------------------------------------------------------------------


So, it seems like there are no "references" (I assume this is the CAS part of the calculation), but the calculation identified the active space correctly. I also don't know what the "1100" numbers mean for alpha and beta.

Any clues?

Thanks; Chris

Clicked A Few Times
Threads 18
Posts 44
OK, seems like targetsym needs to be set. Now getting a nonsense number:

Fermi vacuum according to SCF wave function  has been requested.
 Inconsistent number of electrons in bwcc input n:                   447 
  ne.nocc(s):                   120  n0:   2097867219951104560
 ------------------------------------------------------------------------
 tce_mrcc_readref                   4
 ------------------------------------------------------------------------

Coming from code
      do s = 1,2
         do j = 1, nref
            n  = 0
            n0 = 0
            do i = 1, nmo(s)
               n  = n  + iocc(i,j,s)
               n0 = n0 + iocc0(i,s)
            end do
            if (n.ne.nocc(s) .or. n0.ne.nocc(s)) then
               write(LuOut,*)
     &            'Inconsistent number of electrons in bwcc input'
     1            ,' n:',n,' ne.nocc(s):',nocc(s),' n0:',n0
               call errquit('tce_mrcc_readref',4,UNKNOWN_ERR)
            end if
         end do
      end do


So if I read this correctly,

n should be the number of occupied orbitals in configuration j, spin s
n0 should be the number of occupied orbitals in the reference HF determinant, spin s

I turned on the MRCC debug output, and I can't tell whether the output is just interleaved from multiple processes dumping at the same time, or if the references are really coming out different lengths. It looks a little like the latter--I see labels "Reference #" in order, but e.g., reference 1 has 4651 digits (all 0 or 1, presumably occupations) while reference 14 has 1400. They look like concatenations of a shorter occupation string. If anyone responds, I'll be happy to send output offline.

This is occurring with or without the "subgroupsize" and "improvetiling" options set.

  • Karol Forum:Admin, Forum:Mod, NWChemDeveloper, bureaucrat, sysop
    Profile
    Send PM
Clicked A Few Times
Threads 1
Posts 36
Hi Chris,
Yes, targetsym should be defined.
You can use the subgroup option (so far) only for the BW-MRCCSD runs.
Please send me your input - I will have a look.

Best,
Karol


Forum >> NWChem's corner >> Running NWChem



Who's here now Members 0 Guests 1 Bots/Crawler 0


AWC's: 2.5.10 MediaWiki - Stand Alone Forum Extension
Forum theme style by: AWC