CCL:G: freq calculation on G03 with frozen atoms

[Moreau Yohann <yohann.moreau!=!cbt.uhp-nancy.fr>]

 Sent to CCL by: Moreau Yohann [yohann.moreau]_[cbt.uhp-nancy.fr]
 Le Lundi 13 Mars 2006 10:58, Luis Simon luissimonrubio _ hotmail.com a
 écrit :
 Hi Luis,
 When computing the normal mode, you can "freeze" some atoms in
 gaussian, as
 indicated in an older CCL post :
 http://www.ccl.net/cgi-bin/ccl/message.cgi?2004+05+27+005
 The number of normal modes thus determined will be 3(N-m)*6, for a total
 number of N atoms with m frozen. However, the hessian matrix will have the
 dimension of the total system, with zero for elements corresponding to frozen
 atoms, and the checkpoint file will be usable.
 Hope this helps,
 Yohann
 > Sent to CCL by: "Luis  Simon" [luissimonrubio+*+hotmail.com]
 > I am trying to find TS structures using G03 in a large molecule, so a
 > initial (and very expensive) hessian is needed. I have observed that the
 > reaction coordinate is only related to a small part of the whole molecule,
 > so I guess that there is no use in calculating the hessian matrix
 > considering all the atoms in the molecule and it will suffice to consider
 > those atoms (and neighbour atoms) that are "active" in the
 transition
 > state.
 >
 > I have tried to use ONIOM but it fails as some imaginary frequencies are
 > obtained for the low layer. Removing atoms in a freq input files yields a
 > chk file that can not be read in the ts input file.
 >
 > s there any way of simply freezing few atoms in the hessian calculation? I
 > mean, is there some way of indicate to gaussian to skip calculation of some
 > second derivatives and put those derivatives to zero?
 >
 > Thanks in advanced:
 >
 > Luis