[Pw_forum] About trust_radius

[robert.guzman]

Thanks a lot.


Then all the atoms are
On 2014-11-11 14:26, stefano de gironcoli wrote:
> well... bfgs does not do random moves but choose the search direction 
> on
> the basis of the forces and of the current estimate for the inverse
> hessian matrix... as the system is typically not in a quadratic regime
> (at least at the beginning) it does not trust the resulting step to its
> full size but limits the move to a given trust_radium that is update
> during the evolution starting from an initial value and limited to be
> between a minimum and a maximum...
> all these parametera can be set in input using the variables you are
> referring to ...
> 
> stefano
> 
> On 11/11/2014 05:19 PM, robert.guzman wrote:
> Hi everyone.
> 
> What is the meaning of trust_radius_ini, trust_radius_max,
> trust_radius_min?
> 
> I am supposing that:
> if trust_radius_ini = 1.d-2 then QE moves all the atoms of the unit 
> cell
> in 0.00529 A in a random way. And after the QE calculates the forces 
> and
> the new displacements? and if the displacement of one atom (d_a) is
> greater than "trust_radius_max" then QE makes that
> "d_a=trust_radius_max" in the direction of the force.
> 
> I didn't found this information in the web. Please someone can help me?
> 
> Best Regards.
> 
> Mgter. R. M. Guzmán A.
> Insituto Balseiro. Argentina.
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
> 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum