[molpro-user] Numerical Freq calculations using F12-CCSD(T)

[Werner Győrffy]

Dear John,

There might be some problem with the choice of the basis sets you use. 
It is recommended to use the optri basis sets, for example, for 
basis=avtz use ccsd(t)-f12b,ri_basis=avtz/optri. Could you please 
provide an input for your calculation failed?

Regards,

Werner.

On 01/16/2017 05:41 PM, John Travers wrote:
> Hi
>
> I was trying to do a numerical frequency calculation using f12-ccsd(t)
> in c1 symmetry for a small molecule. The program always produces some
> warning messages like : "Poor overlap with frozen basis set. May have
> deleted the wrong vectors!
>  ? The problem occurs in SvdConstructOrthoBasis"
>
> When checked the detailed output,  I also found some possible numerical
> errors, something looks like :
>
>  Construction of CABS:
>  Pseudo-inverse stability               NaN
>  Smallest eigenvalue of S         -3.71E-13  (threshold= 1.00E-08)
>  Ratio eigmin/eigmax              -3.71E-13  (threshold= 1.00E-09)
>  Smallest eigenvalue of S kept    -3.71E-13  (threshold=-3.71E-13, 0
> functions deleted, 658 kept)
>  Maximum overlap with frozen deleted vectors: (nDel=  44)
>     Vector:         1       2       3       4       5       6       7
>     8       9      10      11      12      13      14      15      16
>   17      18      51      25      67      44      36      35      43
>   32      40      38      33      39      55      52      34      22
>   37      21      42      50      20      41      74      19      53
>   54      26      23     231      28      24      95      93      27
> 158      75      61     123      68     108      31      63     163
> 121     203     152      62      29     157      45      94      72
> 97      77      65     204     103      30     115     126      66
> 250      73     131     101     162     116      91     200     122
>     Overlap:       NaN     NaN     NaN     NaN     NaN     NaN     NaN
>   NaN     NaN     NaN     NaN     NaN     NaN     NaN     NaN     NaN
>   NaN     NaN   1.436   1.248   1.102   1.000   0.999   0.997   0.996
> 0.994   0.985   0.983   0.973   0.964   0.962   0.912   0.896   0.871
> 0.858   0.853   0.827   0.825   0.742   0.596   0.594   0.537   0.523
> 0.485   0.465   0.386   0.375   0.373   0.346   0.327   0.321   0.310
> 0.307   0.268   0.259   0.257   0.249   0.247   0.216   0.204   0.194
> 0.191   0.183   0.166   0.164   0.159   0.135   0.134   0.130   0.126
> 0.118   0.114   0.110   0.100   0.095   0.093   0.092   0.092   0.091
> 0.089   0.080   0.079   0.079   0.078   0.076   0.075   0.072   0.070
>
>
> My questions are related to these "NaN" and warnings. Do these matter
> and how to avoid this? Thanks!
>
> Best wishes
>
>
> John
>
>
>
>
>
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user
>