From hinsen@ibs.ibs.fr Thu Oct 24 09:31:46 1996 Received: from ibs.ibs.fr for hinsen@ibs.ibs.fr by www.ccl.net (8.8.0/950822.1) id IAA23252; Thu, 24 Oct 1996 08:40:46 -0400 (EDT) Received: from lmshp2.ibs.fr (lmshp2.ibs.fr [192.134.36.228]) by ibs.ibs.fr (8.6.12/8.6.12) with SMTP id OAA14921; Thu, 24 Oct 1996 14:42:16 +0100 Message-Id: <199610241342.OAA14921@ibs.ibs.fr> Received: by lmshp2.ibs.fr (1.37.109.4/16.2) id AA06005; Thu, 24 Oct 96 14:36:36 +0100 Date: Thu, 24 Oct 96 14:36:36 +0100 From: Konrad Hinsen To: Schiffer@MSMWIA.hoechst.hoechst-ag.d400.de Cc: chemistry@www.ccl.net In-Reply-To: <004DAC30.MAI*/S=Schiffer/OU=MSMWIA/O=HOECHST/PRMD=hoechst-ag/ADMD=dbp/C=de/@MHS> Subject: Re: CCL:G:Charges > postulates.) are most appropriate. The fitting of the electrostatic > potential of a > molecule is a highly non-trivial subject ( see e.g. Michelle Miller Francl > et al., > J. Comput. Chem., Vol. 17, 1996, pp. 367-383, and the very recent review by > Sarah L. Price, J. Chem. Soc., Faraday Trans., Vol. 92, 1996, pp. 2997-3008 > ) > and in my view not a really solved problem, especially for large molecules. I don't agree. The charge fitting problem is not essentially different from other fitting problems in all branches of science and engineering, and lots of articles and books have been written about it. There are well established and robust techniques to find optimal charges and to judge to what degree they are well determined by the reference data. Much of this is not generally known in the computational chemistry community, but that's another problem. There remains of course the problem that the electrostatic potential in a reasonable region may not be sufficient to determine all charges uniquely. But that is not really fatal: if after adding all reference data you can reasonably get (e.g. from multiple conigurations) some charges still remain illdetermined, then they probably don't matter much in real life. If they do, then probably the fixed point charge model is not good enough. > are quite unknown to the community. Hall et al. fit the calculated electron > densities > of molecules to atom-centered gaussians as it is done by many DFT-programs > ( "density fitting with an auxillary basis" ) and got a very good ( in my > view an > excellent ) representation of the electrostatic potential of a molecule. Essentially that means choosing a more complicated model (gaussian densities rather than point charges), and it is not surprising that you can get better results. The question is whether people are willing to work with such more complicated models in their force fields. It is certainly worthwhile exploring in which situations such models can give better results. -- ------------------------------------------------------------------------------- Konrad Hinsen | E-Mail: hinsen@ibs.ibs.fr Laboratoire de Dynamique Moleculaire | Tel.: +33-76.88.99.28 Institut de Biologie Structurale | Fax: +33-76.88.54.94 41, av. des Martyrs | Deutsch/Esperanto/English/ 38027 Grenoble Cedex 1, France | Nederlands/Francais -------------------------------------------------------------------------------