[Wien] Force calculations: problem solved

[Peter Blaha]

		$ro = 0.000005;
		if ($z <= 71.) { $ro = 0.00001 }
		if ($z <= 36.) { $ro = 0.00005 }
		if ($z <= 18.) { $ro = 0.0001  }

I'll put some hints into UG and lstart.

For the forces we need dV/dR at R=0
The wavefunctions (density) vary much stronger and closer to the nucleus for heavier
elements, so you need a small R0 to have a good description near the nucleus.

mazin schrieb:
> keywords: Force calculations, R0, Forces and total energy
> 
> Thank you Peter,
> 
> This solved the problem.
> 
> This is one parameter I never ever have paid any attention to, and, 
> sorry to say, the Users Guide all but ignores it as well.
> 
> The w2web by itself does not reset this parameter either, it copies it 
> from the existing struct file (may I suggest that in the future releases 
> this be corrected?). So I had to delete the old struct file and start 
> from scratch. At this point, w2web reduced R0 for Bi greatly (from 
> 0.0001 to 0.000005), and for Cu twice (0.00005), while regenerating the 
> same values for the rest.
> 
> To be honest, I do not fully understand why heavy elements should 
> require such drastically finer radial mesh (what recipe is actually used 
> in w2web?), nor why the forces are so sensitive to it. But in any event 
> this appears to be a fact and I suspect a fact that is not well known. 
> Would it be difficult to issue a warning in lstart if the radial mesh 
> looks insufficient for the element in question?
> 
> 
> Thanks again, I owe you, as usually,
> 
> Igor
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien

-- 

                                       P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------