Denis Zavelev
denis.zavelev@gmail.com
Thanks for the answer!
This is really sad... From one hand side, it's of course known that any geometry optimization with partially frozen coordinates will not lead to true extremum, the same is with IRC...
But nevertheless, we do optimization as well as TS search freezing some atoms, so why don't do this in IRC? Anyway, if so, I should examine my hypotheses more thoroughly, perhaps the cluster became stable after its correction...