CCL: Problem freezing atoms with gaussian



 Sent to CCL by: Ramon Crehuet [rcsqtc() iiqab.csic.es]
 --Replace strange characters with the "at" sign to recover email
 address--.
 Dear CClers,
 
I am trying to perform an optimisation with gaussian having some atomic positions fixed. I use Opt=AddRedundant and then the list of atoms:
 4 F
 5 F
 23 F
 etc. After the first step, and 100 iterationsof the kind:
 Iteration  6 RMS(Cart)=  0.14928401 RMS(Int)=  0.40589165
 Iteration  7 RMS(Cart)=  0.14928401 RMS(Int)=  0.32910240
 Iteration  8 RMS(Cart)=  0.14928401 RMS(Int)=  0.25609186
 etc
 the gaussian output gives: "Curvilinear Step not converged" and stops.
 
I have two questions. First, on the theoretical side, why does gaussian need to iteratively solve a set of equations to fix the position of the atoms? I understand that this is necessary for non-linear constraints such as distances or angles, but why for fixing cartesian coordinates? Is it because in redundant internal coordinates the atomic coordinates are not linearly related? And second, can I increase the maximum number of iterations to solve this system of equations? The default is 100 and I've looked into the Opt options and some IOp's and couldn't find anything. The system is quite large, but I do not see why fixing a set of atomic positions can lead to a unresolvable system. The set of constraints does not seem incompatible to me. Alternatively, is there a different way to fix some cartensian atomic positions?
 Thanks for your consideration,
 Ramon