This thread is locked.Only browsing is available.
Top Page > Browsing
A tip for geometry optimization
Date: 2007/04/18 04:27
Name: T.Ozaki

Dear All,

In the geometry optimization by the steepest decent method, the pre-factor is
automatically determined so that the initial maximum step of atomic position
can be 0.02 (Ang). So, when you start the optimization with almost a converged
geometrical structure, it can be possible that '0.02' (Ang) is too large,
and the geometrical structure does not converge in some cases.
(Dr. Dam at JAIST reported this behaviour.)

For such a case, the initial maximum step can be tuned by the following keyword:

MD.Initial.MaxStep 0.02 # default=0.02 (Ang)

With a smaller step, it would be possible to obtain the convergent structure
even if you start from almost a converged geometrical structure.

Best regards,

TO
ƒƒ“ƒe
Page: [1]