UT theoretical chemistry code forum

Hi, everyone:
Recently, I encountered a problem beyond my ability. When I did the NEB calculation, I found more stable states with lower energy near the two endpoints. I optimized the two endpoints again with VASP, they all converged which mean that the endpionts were minimum or at least local minimum. I want to find all the more stable states around my "local minima", in this way, to find out their common features. It may be helpful to interprete this.
As far as I know, the Dimer can find the saddle points near a minia. But how to use it in the opposite way, that is, find a more stable states around a minima?

I would be very grateful for any suggestion!
ross

If you are just interested in finding minima I would use a method like basin hopping where you make a random kick and then minimize.

If however you want to find connected states and kinetic information you could use dimer searches to find all relevant saddles around a minima and then run a steepest decent along the negative curvature mode to find the local minima that are connected by the saddle. This method in essence is used in our (adaptive kinetic monte carlo (akmc) method to search. (JCP. 129, 114104 (2008).)

[quote="dano"]If you are just interested in finding minima I would use a method like basin hopping where you make a random kick and then minimize.

If however you want to find connected states and kinetic information you could use dimer searches to find all relevant saddles around a minima and then run a steepest decent along the negative curvature mode to find the local minima that are connected by the saddle. This method in essence is used in our (adaptive kinetic monte carlo (akmc) method to search. (JCP. 129, 114104 (2008).)[/quote]
Thank you very much ,dano. It seems that the "basin hopping" can solve my problem. However, it is a very horrible thing to deal with a large systerm(N>50) using that method. By the way,dano,how can I use the "BH", is it combined with other software?
In fact, my problem arise from a NEB run in which a energically lower IMAGE occures. As the question disscussed elsewhere in the forum(viewtopic.php?f=2&t=576),I am afraid that I have to "cut" the path into two parts.
best regards
ross