ccsd(t) freq errors

[Dave Wilson]

I've been running calcs on weakly bound systems but the freq's I'm getting
for ccsd(t) calcs have obvious errors. I've tried using both uccsd(t) and
rccsd(t) (open shell molecules), tried using numerical gradients as well
as central differences etc but without success. Adjusting the step size
sometimes improves the results but it seems a lot of trial and error - a
lot of wasted cpu for such calcs. 

I've tried using different basis sets but I'm still getting rubbish freqs.

Has anybody had similar experiences with ccsd(t) freq calcs? Hopefully
there is a solution to this.

As an example of the type of output I'm getting, (for a diatomic molecule)

 Imaginary Vibration  Wavenumber
        Nr             [1/cm]
        1            15981.09

   Low Vibration      Wavenumber
        Nr             [1/cm]
        1               23.97
        2               24.25
        3              113.02

     Vibration        Wavenumber
        Nr             [1/cm]
        1              532.35
        2            15980.62


The 532.35 vibration is the normal mode vibration, which is about the
expected value. 

Any advice would be gratefully accepted.

Cheers
Dave
___________________________________________________________
David Wilson		        Theoretical Chemistry Group

Chemistry Department	    	Phone  Work +61 2 4921 5467
University of Newcastle     	Fax    Work +61 2 4921 5472 
Australia		    		
	 http://quantum.newcastle.edu.au/david