Basis set generation help

Das scpem... at lbl.gov
Wed Jul 11 05:40:34 UTC 2012


Hi all,
  I am trying to generate optimized basis sets for Lanthanide and Actinide 
elements using the gaussian basis generation program *optbas*. I tried to 
optimize a basis set for Nd using the *basis-Nd-q14-g6* file that is 
provided in the *${CP2K_HOME}/trunk/basis_sets/inpout/PADE/* directory as a 
starting point. However, the program crashes with the error
*"diag: Overlap matrix has linear dependency"*
*
*
How do I avoid this error?
I am pasting the input and output below

*========input===================================*
*
&ATOM
  NAME  Nd
  PPOT
  ZVAL  60
  ZEFF  14
  LMAX  3
  OCCU
           2        2 2
         1          6
         1          0
         1          4
  MIXI  0.2
  CONV  1.0D-12
  ITER  200
  IPPN  100
  XCFN  PADE
  OPTI  1.0D-8
&END

&BASIS
  GAUS
  6 
      2.365739530243      2.352254571258      0.766552037600     
 0.333931373408      0.115924677467      0.031752598879 
 1 2 3 4 5 6
6 
      2.365739530243      2.352254571258      0.766552037600     
 0.333931373408      0.115924677467      0.031752598879 
 1 2 3 4 5 6
6 
      2.365739530243      2.352254571258      0.766552037600     
 0.333931373408      0.115924677467      0.031752598879 
 1 2 3 4 5 6
6 
      2.365739530243      2.352254571258      0.766552037600     
 0.333931373408      0.115924677467      0.031752598879 
 1 2 3 4 5 6
6 
      2.365739530243      2.352254571258      0.766552037600     
 0.333931373408      0.115924677467      0.031752598879 
 1 2 3 4 5 6

&END
 &POTENTIAL
     GOEDECKER
 4                                    LMAX
       .529166698                                 RC
   2    17.815030430     -.594797667   #C  C1 C2
       .502999856  2   1.529109884    .834136795  -2.153731944 H(s) 11 12 
22           
       .467012905  2    .721553461    .696195481  -1.647499204 H(p) 11 12 
22           
       .325289583  1   -.543240170 H(d) 11                 
       .294742814  1 -18.520227599 H(f) 11                 
 &END
*
*===============================================*
*
*
*===============output===========================*
*
  
        PROGRAM ATOM
        ------------
 reading input file...
  LMAX =           3
  OCCU(           1 ,           0 ) =    2.00000000000000     
  OCCU(           2 ,           0 ) =    2.00000000000000     
  OCCU(           1 ,           1 ) =    6.00000000000000     
  OCCU(           1 ,           2 ) =   0.000000000000000E+000
  OCCU(           1 ,           3 ) =    4.00000000000000     
  NALPHA(           0 ) =            6
  GAUS(           1 ,           0 ) =    2.36573953024300     
  GAUS(           2 ,           0 ) =    2.35225457125800     
  GAUS(           3 ,           0 ) =   0.766552037600000     
  GAUS(           4 ,           0 ) =   0.333931373408000     
  GAUS(           5 ,           0 ) =   0.115924677467000     
  GAUS(           6 ,           0 ) =   3.175259887900000E-002
  ALPP(           1 ,           0 ) =            1
  ALPP(           2 ,           0 ) =            2
  ALPP(           3 ,           0 ) =            3
  ALPP(           4 ,           0 ) =            4
  ALPP(           5 ,           0 ) =            5
  ALPP(           6 ,           0 ) =            6
  NALPHA(           1 ) =            6
  GAUS(           1 ,           1 ) =    2.36573953024300     
  GAUS(           2 ,           1 ) =    2.35225457125800     
  GAUS(           3 ,           1 ) =   0.766552037600000     
  GAUS(           4 ,           1 ) =   0.333931373408000     
  GAUS(           5 ,           1 ) =   0.115924677467000     
  GAUS(           6 ,           1 ) =   3.175259887900000E-002
  ALPP(           1 ,           1 ) =            1
  ALPP(           2 ,           1 ) =            2
  ALPP(           3 ,           1 ) =            3
  ALPP(           4 ,           1 ) =            4
  ALPP(           5 ,           1 ) =            5
  ALPP(           6 ,           1 ) =            6
  NALPHA(           2 ) =            6
  GAUS(           1 ,           2 ) =    2.36573953024300     
  GAUS(           2 ,           2 ) =    2.35225457125800     
  GAUS(           3 ,           2 ) =   0.766552037600000     
  GAUS(           4 ,           2 ) =   0.333931373408000     
  GAUS(           5 ,           2 ) =   0.115924677467000     
  GAUS(           6 ,           2 ) =   3.175259887900000E-002
  ALPP(           1 ,           2 ) =            1
  ALPP(           2 ,           2 ) =            2
  ALPP(           3 ,           2 ) =            3
  ALPP(           4 ,           2 ) =            4
  ALPP(           5 ,           2 ) =            5
  ALPP(           6 ,           2 ) =            6
  NALPHA(           3 ) =            6
  GAUS(           1 ,           3 ) =    2.36573953024300     
  GAUS(           2 ,           3 ) =    2.35225457125800     
  GAUS(           3 ,           3 ) =   0.766552037600000     
  GAUS(           4 ,           3 ) =   0.333931373408000     
  GAUS(           5 ,           3 ) =   0.115924677467000     
  GAUS(           6 ,           3 ) =   3.175259887900000E-002
  ALPP(           1 ,           3 ) =            1
  ALPP(           2 ,           3 ) =            2
  ALPP(           3 ,           3 ) =            3
  ALPP(           4 ,           3 ) =            4
  ALPP(           5 ,           3 ) =            5
  ALPP(           6 ,           3 ) =            6
  Unknown keyword :6         
  Unknown keyword :2.36      
  Unknown keyword :2.35      
  Unknown keyword :0.76      
  Unknown keyword :0.33      
  Unknown keyword :0.11      
  Unknown keyword :0.03      
  Unknown keyword :1         
  Unknown keyword :2         
  Unknown keyword :3         
  Unknown keyword :4         
  Unknown keyword :5         
  Unknown keyword :6         
  Unknown keyword :          
 Zval =    60.0000000000000     
 finished reading input file.
  
  
 Basis Optimization for atom ND
  
 Pseudopotential calculation with  14.00 valence electrons.
 Goedecker pseudopotential
 Effective core charge is  14.0 .
  
 The exchange correlation functional is LDA .
  
  
   100 points

r(1) =       0.000000007128
r(n) =      27.592628200894
     1        14.00000000000004  -3.553E-14       -45.65611536880110
     2        14.00000000004239  -4.239E-11       -49.85166017152175
     3        13.99999999999986   1.439E-13       -52.55235239913090
     4        14.00000000004143  -4.143E-11       -52.33646383858017
     5        14.00000000001213  -1.213E-11       -52.68168868070856
     6        14.00000000001491  -1.491E-11       -52.67944455717392
     7        14.00000000001312  -1.312E-11       -52.68180921276654
     8        14.00000000001172  -1.172E-11       -52.68233730628295
     9        14.00000000001173  -1.173E-11       -52.68233759472056
    10        14.00000000001277  -1.277E-11       -52.68233759651003
    11        14.00000000001319  -1.319E-11       -52.68233759651931
    12        14.00000000001227  -1.227E-11       -52.68233759651482
    13        14.00000000001192  -1.192E-11       -52.68233759653047
    14        14.00000000001172  -1.172E-11       -52.68233759651416
    15        14.00000000001300  -1.300E-11       -52.68233759652121
    16        14.00000000001242  -1.242E-11       -52.68233759650462
    17        14.00000000001131  -1.131E-11       -52.68233759652614
    18        14.00000000001181  -1.181E-11       -52.68233759652527
   scf-loop:           18  steps        E=  -52.6823375965253     
 initializing Hessian as diagonal matrix...
           1 New direction:
     800111111
   100 points

r(1) =       0.000000007128
r(n) =      27.592628200894
     1        13.95227039006375   4.773E-02       -52.58518483142689
 ndep =           1
 Overlap matrix S
    1.000000    0.999994    0.797329    0.534326    0.274189    0.109328
    0.999994    1.000000    0.799048    0.536027    0.275242    0.109778
    0.797329    0.799048    1.000000    0.881699    0.555306    0.244363
    0.534326    0.536027    0.881699    1.000000    0.818107    0.422632
    0.274189    0.275242    0.555306    0.818107    1.000000    0.744800
    0.109328    0.109778    0.244363    0.422632    0.744800    1.000000
 Eigenvalues and eigenvectors of the overlap matrix S
    0.000001    0.705019    0.101786    0.193138    0.326513   -0.413550   
 0.421486
    0.015285   -0.709131    0.089705    0.188077    0.324076   -0.412999   
 0.422042
    0.118939    0.007650   -0.582711   -0.563991   -0.301587   -0.141783   
 0.480880
    0.511976   -0.004433    0.689136   -0.083849   -0.525255    0.176392   
 0.459386
    1.551712    0.001519   -0.390025    0.659794   -0.117125    0.506588   
 0.377094
    3.802087   -0.000358    0.122530   -0.408508    0.638591    0.592081   
 0.244436
 alpha:  
        2.3656493935        2.3523546985        0.7665441701       
 0.3339315383
        0.1159253367        0.0317525342
        2.3656493935        2.3523546985        0.7665441701       
 0.3339315383
        0.1159253367        0.0317525342
        2.3656493935        2.3523546985        0.7665441701       
 0.3339315383
        0.1159253367        0.0317525342
        2.3656493935        2.3523546985        0.7665441701       
 0.3339315383
        0.1159253367        0.0317525342
 galpha:  
        0.0018321464        0.0018715621        0.0175072478       
-0.0208608560
       -0.0272090252       -0.0215773321
       -0.0018546657       -0.0019278218       -0.0132979508       
-0.0058427766
        0.0007191953       -0.0000378422
        0.0000000000        0.0000000000        0.0000000000       
 0.0000000000
        0.0000000000        0.0000000000
        0.0000386246        0.0000381637       -0.0041959078       
 0.0267021539
        0.0264407750        0.0216793080
diag: Overlap matrix has linear dependency
*
*===============================================*
*
*
It seems that the program makes two of the exponents very similar causing 
off-diagonal elements in the overlap matrix to be very close to 1.
I see this behaviour for pretty much all of the Lanthanide elements I have 
looked at. In all cases, I've used the example basis files provided 
in ${CP2K_HOME}/trunk/basis_sets/inpout/PADE/ as inputs to optbas after 
merely adding the *OPTI 1.0D-8* flag so that optbas perfoms an optimization.

Are there any options one can pass to optbas to prevent this linear 
dependency during the optimization?

Any help is much appreciated.

Many Thanks,
-Das
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20120710/3573ac77/attachment.htm>


More information about the CP2K-user mailing list