[molpro-user] Molpro CCSD(T) convergence problem
Kirk Peterson
kipeters at wsu.edu
Wed Jul 8 07:37:52 BST 2009
Hi,
you're using a 60 electron ECP yet you're telling Molpro that you want
a calculation with 79 electrons. Only
19 electrons are actually present due to the ECP. So just try wf,
19,1,1. In the case of your occ card, of course this also means
that all orbitals associated with the 1s2s2p3s3p3d4s4p4d4f electrons
are not present either.
Hope that helps.
regards,
Kirk
On Jul 7, 2009, at 8:17 PM, manti027 at umn.edu wrote:
>
> Hi there,
>
> I am trying to calculate the single point energy of Au atom (Z=79)
> using
> CCSD(T). I have the effective core-potential for 60 electrons and
> valence
> basis set for the 19 electrons of outermost orbitals in the basis
> set file
> included for the calculation.
>
> A calculation with default allocation of core electrons has been
> initially
> conducted with the following input:
>
> ************************
> gprint,basis,orbitals
> include /home/bc2/mantinam/basis/AUCO.mbas
>
> text, Calculate ccsd(t) energy of Au
> geometry={Au}
> int;thr,16,14
> rhf;wf,79,1,1;accu,13;maxit,60
> uccsd(t);thresh,energy=13;option,maxit=80
> e1=energy;
> ************************
>
> The error listed in the .out file from the above job is as follows:
>
> *********************************
> *********************************
> Reference energy: 312.55476997
>
> ITER. NORM CORR.ENERGY TOTAL ENERGY ENERGY CHANGE DEN1 VAR(S) VAR(P)
> DIIS
> TIME
> 1 1.23024784 -1.11859719 311.43617278 -1.11859719 -0.09156600 0.16D-02
> 0.29D-01 0 0 10.31
> 2 1.34431797 -1.20404576 311.35072421 -0.08544856 -0.01802645 0.36D-03
> 0.21D-01 0 0 20.89
> 3 1.48078009 -1.19261578 311.36215419 0.01142997 -0.01336297 0.46D-03
> 0.27D-01 1 1 31.69
> 4 1.62134698 -1.20226709 311.35250288 -0.00965131 -0.01675052 0.50D-03
> 0.38D-01 2 2 41.54
> 5 1.29361641 -1.18545411 311.36931586 0.01681298 -0.00652391 0.38D-03
> 0.97D-02 3 3 51.28
> 6 1.41802572 -1.19623641 311.35853356 -0.01078230 -0.00482888 0.56D-03
> 0.87D-02 4 4 61.09
> 7 1.40106649 -1.19538985 311.35938012 0.00084656 -0.00450589 0.51D-03
> 0.80D-02 5 5 70.01
> 8 1.38253148 -1.19453674 311.36023323 0.00085312 -0.00243274 0.39D-03
> 0.36D-02 6 6 79.92
> 9 1.42296530 -1.19530916 311.35946081 -0.00077243 -0.00268666 0.50D-03
> 0.43D-02 6 1 90.31
> 10 1.51944935 -1.19869637 311.35607360 -0.00338721 -0.04073747
> 0.63D-03
> 0.53D-01 6 0 99.66
> 11 126.06692364 -1.24279237 311.31197760 -0.04409600 -81.76602138
> 0.77D-03
> 0.12D+03 6 0 109.13
>
> UNREASONABLE NORM. CALCULATION STOPPED .........
>
>
> .... ?ERROR: NO CONVERGENCE. TRIPLES WILL NOT BE DONE. THIS CAN BE
> AVOIDED
> USING THE NOCHECK OPTION.
>
> Spin contamination <S**2-Sz**2-Sz> 0.91672515
>
>
> RESULTS
> =======
>
> Reference energy 312.554769970745
> Correlation energy -1.242792366583
> !RHF-UCCSD STATE 1.1 ENERGY 311.311977604161
>
> Program statistics:
>
> Available memory in ccsd: 199999900
> Min. memory needed in ccsd: 4117444
> Max. memory used in ccsd: 6207380
> Max. memory used in cckext: 8893506 (11 integral passes)
> Max. memory used in cckint: 2908899 ( 1 integral passes)
>
>
> ? Error
> ? No convergence. This error exit can be avoided using the NOCHECK
> option
> ? The problem occurs in ccsd
>
> GA ERROR fehler on processor 0 tmp =
> /home/bc2/mantinam/pdir//usr/local/molpro/2006_mpp/lib/molpro-mpp-
> Linux-x86_64-i8-2006.1/molprop_2006_1_i8_x86_64_tcgmsg.exe.p
> Creating: host=blade224, user=mantinam,
>
> file=/usr/local/molpro/2006_mpp/lib/molpro-mpp-Linux-x86_64-
> i8-2006.1/molprop_2006_1_i8_x86_64_tcgmsg.exe,
> port=35013
>
> *********************************
> *********************************
>
> To obtain an initial guess for the occupancy of electrons, an initial
> Hartree-Fock calculation has been conducted using the following input:
>
> ***************************
> gprint,basis,orbitals
> include /home/bc2/mantinam/basis/AUCO.mbas
>
> text, Calculate ccsd(t) energy of Au
> geometry={Au}
> int; thr,16,14
> rhf;wf,79,1,1;accu,13;maxit,60
> ***************************
>
> This calculation converged. Although, the electron orbitals
> occupancy is
> quite mixed and is difficult to be interpreted as to which orbital
> is being
> occupied even for the expectedly most stable orbitals. Main part of
> the Output obtained from this calculation is attached.
>
> Using the orbital occupancy from the output of the HF calculation, the
> error statement "ERROR: NO CONVERGENCE. TRIPLES WILL NOT BE DONE.
> THIS CAN
> BE AVOIDED USING THE NOCHECK OPTION" is obtained again. The input
> for this
> calculation and the error listed lines from the output is as follows:
>
> *****************************
> gprint,basis,orbitals
> include /home/bc2/mantinam/basis/AUCO.mbas
>
> text, Calculate ccsd(t) energy of Au
> geometry={Au}
> int;thr,16,14
> rhf;occ,14,5,5,3,4,5,3,1;wf,79,1,1;accu,13;maxit,60
> uccsd(t);core,10,4,4,2,3,4,2,1;thresh,energy=13;option,maxit=80
> e1=energy;
> ******************************
>
> LINES WITH ERROR IN .OUT FILE:
>
> *******************************
> *******************************
> Reference energy: 312.48747691
>
> ITER. NORM CORR.ENERGY TOTAL ENERGY ENERGY CHANGE DEN1 VAR(S) VAR(P)
> DIIS
> TIME
> 1 1.12333348 -0.32484798 312.16262893 -0.32484798 -0.01593550 0.21D-03
> 0.18D-01 0 0 0.41
> 2 1.22084300 -0.34039411 312.14708280 -0.01554613 -0.01302992 0.25D-04
> 0.22D-01 0 0 0.64
> 3 1.38734108 -0.35177500 312.13570191 -0.01138089 -0.01784110 0.20D-04
> 0.33D-01 1 1 0.86
> 4 1.13587345 -0.33124247 312.15623444 0.02053253 -0.00879974 0.40D-04
> 0.12D-01 2 2 1.08
> 5 1.28517020 -0.34756285 312.13991406 -0.01632038 -0.00623722 0.33D-04
> 0.10D-01 3 3 1.30
> 6 1.25941048 -0.34607371 312.14140320 0.00148914 -0.00571449 0.35D-04
> 0.90D-02 4 4 1.54
> 7 1.26525317 -0.34506948 312.14240743 0.00100423 -0.00397328 0.46D-04
> 0.57D-02 5 5 1.78
> 8 1.30428486 -0.34749359 312.13998332 -0.00242411 -0.00415983 0.36D-04
> 0.63D-02 6 6 2.01
> 9 1.31253180 -0.34769388 312.13978303 -0.00020029 -0.00425153 0.34D-04
> 0.65D-02 6 2 2.24
> 10 1.29012937 -0.34728499 312.14019192 0.00040889 -0.00349707 0.50D-04
> 0.50D-02 6 3 2.47
> 11 1.40022531 -0.35575824 312.13171866 -0.00847325 -0.05671379
> 0.32D-04
> 0.77D-01 6 0 2.68
> 12 174.19082786 -0.61280116 311.87467575 -0.25704292 -123.04838902
> 0.20D-04 0.17D+03 6 0 2.88
>
> UNREASONABLE NORM. CALCULATION STOPPED ......
>
> ....... ?ERROR: NO CONVERGENCE. TRIPLES WILL NOT BE DONE. THIS CAN BE
> AVOIDED USING THE NOCHECK OPTION.
>
> Spin contamination <S**2-Sz**2-Sz> 0.49831182
>
>
> RESULTS
> =======
>
> Reference energy 312.487476907880
> Correlation energy -0.612801161430
> !RHF-UCCSD STATE 1.1 ENERGY 311.874675746450
>
> Program statistics:
>
> Available memory in ccsd: 199999900
> Min. memory needed in ccsd: 313673
> Max. memory used in ccsd: 450233
> Max. memory used in cckext: 661777 (12 integral passes)
> Max. memory used in cckint: 709286 ( 1 integral passes)
>
>
> ? Error
> ? No convergence. This error exit can be avoided using the NOCHECK
> option
> ? The problem occurs in ccsd
>
> GA ERROR fehler on processor 0 tmp =
> /home/bc2/mantinam/pdir//usr/local/molpro/2006_mpp/lib/molpro-mpp-
> Linux-x86_64-i8-2006.1/molprop_2006_1_i8_x86_64_tcgmsg.exe.p
> Creating: host=blade209, user=mantinam,
>
> file=/usr/local/molpro/2006_mpp/lib/molpro-mpp-Linux-x86_64-
> i8-2006.1/molprop_2006_1_i8_x86_64_tcgmsg.exe,
> port=34946
>
> *******************************
> *******************************
>
> I doubt the values I am feeding for the option "occ" are incorrect.
> If this
> is the case, I am unable to figure out a way out of it as the HF
> calculation output is not helpful in understanding the orbital
> occupancy
> correctly due to its nature of highly mixed states. I have no
> information
> of what else could be wrong with the input OR an option that would
> fix the
> situation.
>
> Any suggestions on a way to fix the convergence problem I am facing in
> obtaining the total energy of Au atom using CCSD(T) would be greatly
> helpful.
>
> Thanks and Regards,
> Manjeera
> <Au.out>_______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user
More information about the Molpro-user
mailing list