Author Topic: Using cc-pV6Z in mpshift?  (Read 8876 times)

Andersx

  • Newbie
  • *
  • Posts: 4
  • Karma: +0/-0
Using cc-pV6Z in mpshift?
« on: May 28, 2010, 04:51:25 pm »
Hi,

I am trying to calculate the isotropic chemical shielding of a formamide molecule (HCO-NH2)at the MP2/cc-pV6Z level of theory. However, I get the following error message:

Code: [Select]
mpshift(wunray.theory.ki.ku.dk) : TURBOMOLE V6.1 22 Sep 2009 at 17:26:03
 Copyright (C) 2009 TURBOMOLE GmbH, Karlsruhe


    2010-05-24 18:51:43.048



                        DIRECT MP2 NMR CHEMICAL SHIFTS

                 M.Kollwitz, M.Haser, J.Gauss
                 J.Chem.Phys.108(1998)8295

                 M.Kollwitz, J.Gauss, Chem.Phys.Lett.260(1996)639


                        DIRECT RHF NMR CHEMICAL SHIFTS

                 M.Haser, R.Ahlrichs, H.P.Baron, P.Weis, H.Horn
                        Theoret.Chim.Acta 83(1992)455


                        DIRECT DFT NMR CHEMICAL SHIFTS

                 U. Huniar, PhD Thesis, Univ. of Karlsruhe 1999


                          quantum chemistry group
                          universitaet  karlsruhe
                                  germany


                   >>>>> MP2 MP2 MP2 MP2 MP2 MP2 MP2 <<<<<
                      You specified the MP2 level for
                           your chemical shifts
                   >>>>> MP2 MP2 MP2 MP2 MP2 MP2 MP2 <<<<<

   *************************************************************************
   FMA_ref cc-pV6Z                                                         
   *************************************************************************




              +--------------------------------------------------+
              | Atomic coordinate, charge and isotop information |
              +--------------------------------------------------+


              atomic coordinates              atom shells charge pseudo isotop
    67.26669143   26.93804602    2.47365151    n     28    7.000    0     0
    65.63781680   23.02103322    3.17542007    o     28    8.000    0     0
    66.96633266   28.69651005    1.83106579    h     21    1.000    0     0
    65.41059005   25.17052395    2.43445527    c     28    6.000    0     0
    68.98021855   26.48148772    3.15970647    h     21    1.000    0     0
    63.62731674   25.90519140    1.63963090    h     21    1.000    0     0
 

     center of nuclear mass  :   66.14329945   25.08133026    2.69478773

     center of nuclear charge:   66.16694928   25.20170503    2.66483566


              +--------------------------------------------------+
              |               basis set information              |
              +--------------------------------------------------+


              we will work with the  1s 3p 5d  7f  9g ... basis set
              ...i.e. with spherical basis functions...


   type   atoms  prim   cont   basis
   ---------------------------------------------------------------------------
    n        1    172    140   cc-pV6Z   [7s6p5d4f3g2h1i|27s10p5d4f3g2h1i]
    o        1    172    140   cc-pV6Z   [7s6p5d4f3g2h1i|27s10p5d4f3g2h1i]
    h        3     95     91   cc-pV6Z   [6s5p4d3f2g1h|10s5p4d3f2g1h]
    c        1    172    140   cc-pV6Z   [7s6p5d4f3g2h1i|27s10p5d4f3g2h1i]
   ---------------------------------------------------------------------------
   total:    6    801    693
   ---------------------------------------------------------------------------

   total number of primitive shells          :  181
   total number of contracted shells         :  147
   total number of cartesian basis functions : 1008
   total number of SCF-basis functions       :  693


 DETECTED HIGHLY CONTRACTED SHELL(S) WITH CONTRACTION DEPTH 11 !
 THIS IS TOO MUCH, SINCE TURBOMOLE INTEGRAL ROUTINES CURRENTLY
 ARE LIMITED TO 10 PRIMITIVES WITHIN A CONTRACTED BLOCK OF BASIS FUNCTIONS

 
 ========================
  internal module stack:
 ------------------------
    molinp
 ========================

  <molinp> abend
 mpshift ended abnormally

Especially this part:
Code: [Select]
DETECTED HIGHLY CONTRACTED SHELL(S) WITH CONTRACTION DEPTH 11 !
 THIS IS TOO MUCH, SINCE TURBOMOLE INTEGRAL ROUTINES CURRENTLY
 ARE LIMITED TO 10 PRIMITIVES WITHIN A CONTRACTED BLOCK OF BASIS FUNCTIONS

Is there any way to get around this?

Thanks,
Anders Christensen

uwe

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 410
  • Karma: +0/-0
Re: Using cc-pV6Z in mpshift?
« Reply #1 on: June 01, 2010, 11:32:48 am »
Hi,

the GIAO integral routines have never been intensively tested after the removal of the limitation of the contraction depth. Therefore the old check for a maximum of 10 contractions per shell is still present. There is no way to get around this in the official Turbomole version. You can contact the Turbomole support and you will get a version without the limit, but you should first check for the correctness of the results for a couple of systems.

The only way to get around the problem is to de-contract the basis set a bit. cc-pV6Z has already 11 primitives in one shell for carbon for example. De-contracting makes the basis set larger, and the results will of course not be of cc-pV6Z quality but a bit more towards basis set limit. However, the calculation will take (much?) longer!

De-contracting is relatively easy. Make a usual input and edit the basis file. Split e.g. 11s to 6s and 5s (you do not have to change the contraction coefficients, just change the 11s to 6s and add one line after the first 6 primitives). Remove the $rundimension keyword in the control file, delete the orbitals and rerun define to make new orbitals with eht. Then of course dscf and mp2prep -c again.

Regards,

Uwe

Andersx

  • Newbie
  • *
  • Posts: 4
  • Karma: +0/-0
Re: Using cc-pV6Z in mpshift?
« Reply #2 on: June 01, 2010, 04:06:44 pm »
Thank you for the answer! I will do so and contact the support team. Unfortunately, I don't think de-contraction is a viable option in this situation, but thank you for the suggestion.

Also, are there any newer version of mpshift which works in parallel with MP2?

Thanks,
Anders
« Last Edit: June 01, 2010, 08:08:39 pm by Andersx »

Andersx

  • Newbie
  • *
  • Posts: 4
  • Karma: +0/-0
Re: Using cc-pV6Z in mpshift?
« Reply #3 on: June 25, 2010, 06:45:10 pm »
Hey, Turbomole. Thank you for all the support! I got the new version of mpshift with support for more than 10 contractions, and it seems to work fine, thanks! However, now I have a problem with mp2prep. I can only specify up to 99,999 MB (=100 GB) of disk space via this line:

mp2prep -c -p 99999 -m 5900

If I specify any higher number (e.g. -p 100000, which is just 1 higher), I'm told, that "Diskspace must be numeric." :)
So when I run my calculations, I quickly run over the max allowed diskspace for the calculation, even though I have 1.3 TB available on the cluster.

Code: [Select]
---------------------------------------------------
 expected total disk space for this pass :   99969 Mbytes
 (the files "*" and two of the three files "-" are needed)
 ---------------------------------------------------


**********************************************************************
resulting maximum disk storage requirement          :     99969 Mbyte
**********************************************************************


 mpshift statistics run ended normally
 

How do I let mpshift run with more than 100GB's of scratch space with mp2prep?

Best regards, Anders

uwe

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 410
  • Karma: +0/-0
Re: Using cc-pV6Z in mpshift?
« Reply #4 on: June 28, 2010, 11:50:24 am »
Hi,

m2prep is just a shell script in $TURBODIR/scripts -  quite an old one actually...

Edit mp2prep and search for the lines:

Quote
if [ -n "$PLATT" ]; then
   case $PLATT in
           [0-9]|[1-9][0-9]|[1-9][0-9][0-9]|[1-9][0-9][0-9][0-9]|[1-9][0-9][0-9][0-9][0-9]) : ;;
           *) echo "Diskspace must be numeric.
         $usage" >&2
              exit 1;;
   esac
fi

Either extend the [0-9] line or simply remove the complete if or the exit 1.

Regards,

Uwe

Andersx

  • Newbie
  • *
  • Posts: 4
  • Karma: +0/-0
Re: Using cc-pV6Z in mpshift?
« Reply #5 on: August 26, 2010, 09:50:43 pm »
Thank you for the previous replies! I got everything working with the mpshift you sent med which allows for the use of i-functions (such as in the cc-pV6Z basis). However, I have a calculation which I can't get to converge, any ideas as to keywords or the like?

Here is the output from mpshift - scroll to the last lines for the error message.

Code: [Select]


                        DIRECT MP2 NMR CHEMICAL SHIFTS

                 M.Kollwitz, M.Haser, J.Gauss
                 J.Chem.Phys.108(1998)8295

                 M.Kollwitz, J.Gauss, Chem.Phys.Lett.260(1996)639


                        DIRECT RHF NMR CHEMICAL SHIFTS

                 M.Haser, R.Ahlrichs, H.P.Baron, P.Weis, H.Horn
                        Theoret.Chim.Acta 83(1992)455


                        DIRECT DFT NMR CHEMICAL SHIFTS

                 U. Huniar, PhD Thesis, Univ. of Karlsruhe 1999


                          quantum chemistry group
                          universitaet  karlsruhe
                                  germany


                   >>>>> MP2 MP2 MP2 MP2 MP2 MP2 MP2 <<<<<
                      You specified the MP2 level for
                           your chemical shifts
                   >>>>> MP2 MP2 MP2 MP2 MP2 MP2 MP2 <<<<<

   *************************************************************************
   FMA_ref cc-pV6Z                                                          
   *************************************************************************




              +--------------------------------------------------+
              | Atomic coordinate, charge and isotop information |
              +--------------------------------------------------+


              atomic coordinates              atom shells charge pseudo isotop
    67.26669143   26.93804602    2.47365151    n     28    7.000    0     0
    65.63781680   23.02103322    3.17542007    o     28    8.000    0     0
    66.96633266   28.69651005    1.83106579    h     21    1.000    0     0
    65.41059005   25.17052395    2.43445527    c     28    6.000    0     0
    68.98021855   26.48148772    3.15970647    h     21    1.000    0     0
    63.62731674   25.90519140    1.63963090    h     21    1.000    0     0
 

     center of nuclear mass  :   66.14329945   25.08133026    2.69478773

     center of nuclear charge:   66.16694928   25.20170503    2.66483566


              +--------------------------------------------------+
              |               basis set information              |
              +--------------------------------------------------+


              we will work with the  1s 3p 5d  7f  9g ... basis set
              ...i.e. with spherical basis functions...


   type   atoms  prim   cont   basis
   ---------------------------------------------------------------------------
    n        1    172    140   cc-pV6Z   [7s6p5d4f3g2h1i|27s10p5d4f3g2h1i]
    o        1    172    140   cc-pV6Z   [7s6p5d4f3g2h1i|27s10p5d4f3g2h1i]
    h        3     95     91   cc-pV6Z   [6s5p4d3f2g1h|10s5p4d3f2g1h]
    c        1    172    140   cc-pV6Z   [7s6p5d4f3g2h1i|27s10p5d4f3g2h1i]
   ---------------------------------------------------------------------------
   total:    6    801    693
   ---------------------------------------------------------------------------

   total number of primitive shells          :  181
   total number of contracted shells         :  147
   total number of cartesian basis functions : 1008
   total number of SCF-basis functions       :  693


 integral neglect threshold       :  0.33E-11
 integral storage threshold THIZE :  0.10E+10
 integral storage threshold THIME :         5


 WARNING : cannot change filename for statistics

 symmetry group of the molecule :   c1

 the group has the following generators :
   c1(z)

    1 symmetry operations found

 there are 1 real representations :   a  

 maximum number of shells which are related by symmetry :  1


    mo occupation :
   irrep   mo's   occupied
    a      693       12
 
 number of basis functions   :          693
 number of occupied orbitals :           12
 
  MOs are in ASCII format !


 reading orbital data $scfmo from file mos .

 orbital characterization : scfconv=7
 

 number of non-frozen orbitals          :   693
 number of non-frozen occupied orbitals :    12
DGSTOL=   8    SCFTOL=  11    CSCONV=   2
time for          input      0.35
      time for         csplop      0.55
      time for            CSP      0.05
time for    1e matrices      0.62
time for undisturbed cs      6.52
 shell blocked total number of electrons =     24.00000000

time for   dgs 2e stuff   2728.12
Conv: HM Matrix each iteration new in cao
Conv: check all atoms in last 2 iterations

 biggest AO integral is expected to be     5.408063943

 cpu time for 2e-integral bound :      3.01 sec


  0. iteration
==============
  csxx***************  csyy***************  cszz***************
 errxx*************** erryy*************** errzz***************
  converged: none
 Inversion problem in dvdson(osinv)
 
 MODTRACE: no modules on stack

 fatal error in dvdson.
 mpshift ended abnormally