SEARCH
TOOLBOX
LANGUAGES
Forum Menu

Counterpoise corrected MP2 optimization errors

From NWChem

Viewed 139 times, With a total of 0 Posts
Jump to: navigation, search

Just Got Here
Threads 1
Posts 1

EDIT: RESOLVED THE ISSUE BY SWITCHING TO A NON-MKL COMPILED VERSION ON THE CLUSTER





Computer details:
Ohio Supercomputer Cluster, operating on 32 processors, 64 bit, Linux-based

Compilation details:
NWChem v 6.1, with Intel compilers and MKL

Necessary computational details:
Attempting an MP2 geometry optimization with counterpoise correction on a simple fluoride and water dimer with the aug-cc-pvtz basis set.

Additional information:
I am performing the same EXACT calculation for 5 other dimers. These optimizations work with an input file of the same form as the following... (obviously with different ions). If I remove the request to perform a counterpoise correction in the optimization, the calculation works. In fact, I use those coordinates resulting from a non-corrected optimization for my input to the corrected one.

Further additional note:
Changing fluoride basis set to double zeta quality from triple... optimization appears to head to completion. However, for what I'm doing the triple zeta basis set (or higher) is absolutely essential. Using the double zeta basis on water and triple zeta on the fluoride works too.

start f_wat
charge -1
driver
  tight
  maxiter 50
end
geometry 
  symmetry c1
  f    -1.22898319    -0.04847559     0.00000000
  o     1.19722843    -0.04930316     0.00000000
  h     0.13592221    -0.07434392     0.00000000
  h     1.37509402     0.89146226     0.00000000
end       
basis spherical
  f   library   aug-cc-pvtz   
  o   library   aug-cc-pvtz
  h   library   aug-cc-pvtz
  bqf library f aug-cc-pvtz 
  bqo library o aug-cc-pvtz
  bqh library h aug-cc-pvtz
end 
bsse
  mon ion 1
    charge -1
  mon wat 2 3 4
end 
mp2 
 tight 
end
task mp2 optimize



The error:
 
 Renormalizing density from       9.00 to     10
      Non-variational initial energy
      ------------------------------
 Total energy =    -105.099546
 1-e energy   =    -154.710543
 2-e energy   =      49.610997
 HOMO         =       0.167295
 LUMO         =       0.168697
 Starting SCF solution at      24.5s
 ----------------------------------------------
         Quadratically convergent ROHF
 Convergence threshold     :          1.000E-08
 Maximum no. of iterations :           30
 Final Fock-matrix accuracy:          1.000E-10
 ----------------------------------------------
Integral file          = ./f_wat.aoints.00
 Record size in doubles =    65536    No. of integs per rec  =    43688
 Max. records in memory =       46    Max. records in file   =  5937499
 No. of bits per label  =        8    No. of bits per value  =       64
 #quartets = 5.833D+05 #integrals = 2.329D+07 #direct =  0.0% #cached =100.0%
File balance: exchanges=    16  moved=    25  time=   0.0
              iter       energy          gnorm     gmax       time
             ----- ------------------- --------- --------- --------
                 1      -97.1949604777  5.61D+00  1.58D+00     22.4
  Setting level-shift to   7.32 to force positive preconditioner
 ------------------------------------------------------------------------
''' ga_iter_lsolve: dgesv failed*********'''
 ------------------------------------------------------------------------
 ------------------------------------------------------------------------
  current input line :
    37: task mp2 optimize
 ------------------------------------------------------------------------
 ------------------------------------------------------------------------
 ------------------------------------------------------------------------
 For more information see the NWChem manual at
 http://www.nwchem-sw.org/index.php/NWChem_Documentation
 For further details see manual section:
'''0:0:ga_iter_lsolve: dgesv failed:: 0'''
(rank:0 hostname:opt2085.ten.osc.edu pid:31220):ARMCI DASSERT fail. ../../ga-5-1/armci/src/common/armci.c:ARMCI_Error():208 cond:0




If I grep around this calculation, I see the error occurs in the 3rd round...
grep -A40 "Quadratically" <my logfile>

         Quadratically convergent ROHF
 Convergence threshold     :          1.000E-08
 Maximum no. of iterations :           30
 Final Fock-matrix accuracy:          1.000E-10
 ----------------------------------------------
 Integral file          = ./f_wat.aoints.00
 Record size in doubles =    65536    No. of integs per rec  =    43688
 Max. records in memory =       46    Max. records in file   =  5937499
 No. of bits per label  =        8    No. of bits per value  =       64
 #quartets = 5.791D+05 #integrals = 2.259D+07 #direct =  0.0% #cached =100.0%
File balance: exchanges=    16  moved=    18  time=   0.0
              iter       energy          gnorm     gmax       time
             ----- ------------------- --------- --------- --------
                 1     -175.4725399419  1.50D+00  3.26D-01      2.1
                 2     -175.5441736076  2.51D-01  6.25D-02      3.7
                 3     -175.5480210637  2.20D-02  4.89D-03      4.5
                 4     -175.5480719685  1.73D-04  4.36D-05      5.0
                 5     -175.5480719717  2.06D-08  3.43D-09      5.8
                 6     -175.5480719717  7.13D-10  1.56D-10      6.2
       Final RHF  results 
       ------------------ 
         Total SCF energy =   -175.548071971704
      One-electron energy =   -309.837546958955
      Two-electron energy =    104.630478111790
 Nuclear repulsion energy =     29.658996875461
        Time for solution =      4.9s
--
         Quadratically convergent ROHF
 Convergence threshold     :          1.000E-08
 Maximum no. of iterations :           30
 Final Fock-matrix accuracy:          1.000E-10
 ----------------------------------------------
 #quartets = 5.565D+03 #integrals = 7.297D+04 #direct =  0.0% #cached =100.0%
 Integral file          = ./f_wat.aoints.00
 Record size in doubles =    65536    No. of integs per rec  =    43688
 Max. records in memory =        2    Max. records in file   =  5937499
 No. of bits per label  =        8    No. of bits per value  =       64
File balance: exchanges=     0  moved=     0  time=   0.0
              iter       energy          gnorm     gmax       time
             ----- ------------------- --------- --------- --------
                 1      -98.9649726015  3.61D+00  1.22D+00     19.8
                 2      -99.4169373627  9.97D-01  4.11D-01     19.8
                 3      -99.4446561758  4.72D-01  2.04D-01     19.8
                 4      -99.4502293438  6.12D-02  3.37D-02     19.8
                 5      -99.4508067949  1.57D-03  6.23D-04     19.9
                 6      -99.4508070224  8.62D-07  3.69D-07     19.9
                 7      -99.4508070224  2.48D-10  1.15D-10     20.0
       Final RHF  results 
       ------------------ 
         Total SCF energy =    -99.450807022410
      One-electron energy =   -143.996531394389
      Two-electron energy =     44.545724371979
 Nuclear repulsion energy =      0.000000000000
        Time for solution =      0.3s
--
         Quadratically convergent ROHF
 Convergence threshold     :          1.000E-08
 Maximum no. of iterations :           30
 Final Fock-matrix accuracy:          1.000E-10
 ----------------------------------------------
 Integral file          = ./f_wat.aoints.00
 Record size in doubles =    65536    No. of integs per rec  =    43688
 Max. records in memory =       46    Max. records in file   =  5937499
 No. of bits per label  =        8    No. of bits per value  =       64
 #quartets = 5.833D+05 #integrals = 2.329D+07 #direct =  0.0% #cached =100.0%
File balance: exchanges=    16  moved=    25  time=   0.0
              iter       energy          gnorm     gmax       time
             ----- ------------------- --------- --------- --------
                 1      -97.1949604777  5.61D+00  1.58D+00     22.4
  Setting level-shift to   7.32 to force positive preconditioner
 ------------------------------------------------------------------------
 ga_iter_lsolve: dgesv failed*********
 ------------------------------------------------------------------------
 ------------------------------------------------------------------------
  current input line : 
    37: task mp2 optimize
 ------------------------------------------------------------------------
 ------------------------------------------------------------------------
 ------------------------------------------------------------------------
 For more information see the NWChem manual at 
 http://www.nwchem-sw.org/index.php/NWChem_Documentation
 For further details see manual section: 


Hmmm... noticed this after the grep bit:

" Setting level-shift to 7.32 to force positive preconditioner"

I'm not sure what this means.
Edited On 10:57:05 AM PDT - Mon, Jun 25th 2012 by Pollartp


Forum >> NWChem's corner >> Running NWChem



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


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