QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: Roc on January 1, 2009, 03:47

Title: some questions about the geometry optimizing
Post by: Roc on January 1, 2009, 03:47
Dear everyone,

First, Please accept my sincere wishes for the New Year. I hope you will continue to enjoy good health.

Second, I have some questions to consult with you about the geometry optimizing.
As we know, in some other softwares, i.e., Gaussian03,Dmol, after optimizing one molecular or cluster geometry, we could judge the stability of the molucular or cluster through analyzing the vibration frequency. However, if we optimize one molecular or two-probe system in ATK , how do we verify the stability of the optimized structure.

Thanks!
Title: Re: some questions about the geometry optimizing
Post by: kstokbro on January 1, 2009, 10:15
Hi,
Generally, we are interested in the global minimum
When a structure is relaxed the forces are zero, and it can be either the global minimum, or a local minimum, or a saddle point.
In general to ensure that it is the global minimum, the only option is to start the relaxation from different initial configurations.

How to ensure that a relaxed geometry is not a saddle point?
The only way a relaxed structure can end up as an unstable geometry (a saddle point) is if the initial geometry has certain symmetries and these symmetries are not broken during the relaxation. Usually the relaxation will break the symmetries due to numerical errors, introduced for instance by the real space grid, and it is usually a very rare event to end up in an unstable geometry. However, to be completely safe the initial geometry must not have any symmetry. I.e. add a small number to some of the initial coordinates to break the symmetry. Fo rinstance the geometry

H 0.0 0.0 -5.0
H 0.0 0.0 0.0
H 0.0 0.0 5.0

can have a broken symmetry by using:

H 0.0 0.0 -5.02
H 0.0 0.0 0.0
H 0.0  0.0 4.99

Title: Re: some questions about the geometry optimizing
Post by: Roc on January 1, 2009, 13:12
okay, Thanks for your concern!

I think our quantum wise software could be more powerful than other softwares if someday it could be more convenient to optimize any molecular which we desier by using the VNL tool. Of course, it's going to be a huge task or one challenge.

Thanks again, Sir!


Title: Re: some questions about the geometry optimizing
Post by: Nordland on January 1, 2009, 17:40
Hey Anyipeng.

Are you using the newest version of VNL?
Title: Re: some questions about the geometry optimizing
Post by: Roc on January 2, 2009, 03:20
I want to have a try for the latest version, but my boss has no time to upgrade it recently. So.........(*^__^*)  ;D

Thanks.