Dear Lefteris,

 From calculations on small and medium-sized molecules at low
frequencies, we observed that
this g_xc term (which is related to the second derivative of the xc
potential) usually contributes
about 5% to the hyperpolarizability at the ALDA level.

Because this term is also numerically sensitive [the second derivative
of v_xc
tends to infinity in the outer region], I decided to print a warning
message in case this g_xc
term is a large portion of the total hyperpolarizability tensor
component.

This in itself does not mean that there is anything wrong with the
calculation, especially not
if this particular tensor component is small compared to the other
ones, or if you are approaching
a pole where the hyperpolarizability tends to infinity.

If you have reasons to distrust the g_xc term, you can also decide to
set it to zero.
In Time-Dependent Hartree-Fock theory this term does not occur, and not
much research has yet
been done on appropriate approximations for g_xc in DFT. The ALDA
approach seems a very good
first approximation though.

There have been several publications showing the validity of
frequency-dependent hyperpolarizability
calculations with ADF, both for small and medium-sized molecules, also
at various frequencies.
It is important to include diffuse functions in your basis set though
(cf. $ADFRESOURCES/ET)
AND to use the DEPENDENCY keyword.

I hope this sheds some light on the topic you raised.

Best regards,
Stan

On Thursday, November 13, 2003, at 04:32 PM, Lefteris Danos wrote:

> Hello
>  
> I have been doing frequency hyperpolarisability calculations on the
> Toluene molecule at the LB94/Vdiff level and at some frequencies (500,
> 515, 400 nm) I get an error in the hyperpolarizability program part
> saying:
>  
> " Extra DFT term (g_xc) too large
>  beta (diff,total)=  0.22033618275712E+00  0.49362407245446E-01
>  Extra DFT term (g_xc) too large
>  beta (diff,total)=  0.22033618275712E+00  0.49362407245446E-01"
>  
> and at the end of the program it prints
> "<Jun06-2003> <18:01:21>  CAUTION! Extra DFT term related to g_xc
> suspiciously large
>  <Jun06-2003> <18:01:21>  CAUTION! Extra DFT term related to g_xc
> suspiciously large"
>  
> The program calulates the hyperpolarisability tensors but I am not
> sure how reliable they are, especially the frequency dependent. Any
> suggestions why is this message appearing will be helpfull. I get the
> same message for other molecules and other approximating methods
> (GRACLB) as well near their absorption frequencies in the harmonic
> wavelength. My suspicion is it has to do something with the size of
> the basis set.
>  
> Thanks
> Best Regards
> Lefteris
>  
>  
> *************************************************
> Lefteris Danos
> School of Chemistry
> Highfield
> University of Southampton
> Southampton
> SO17 1BJ
> UK
>  
> Tel : +44 (0) 2380 59 4150
> Fax : +44 (0) 2380 59 3781
>  
> Email : ld4_at_soton.ac.uk
> http://www.soton.ac.uk/~jgfchem
> *************************************************

Dr. S.J.A. van Gisbergen      
Scientific Computing & Modelling NV
Theoretical Chemistry, Vrije Universiteit
De Boelelaan 1083
1081 HV Amsterdam
The Netherlands                              
T: +31-20-4447626    
F: +31-20-4447629        
vangisbergen_at_scm.com  
http://www.scm.com
Received on 2003-11-14 09:13:03