Dear siesta users, I have a geometry optimization with broyden method that is stuck at a maximal force of around 0.05XX eV/A (the relaxation spent more than 400 steps around this value). Although, it is not far from the convergence criterium of my test, which I set to 0.04 eV/A, I find it annoying that the optimization is stuck at 0.05XX eV/A. I have played with the options of broyden method with no success. However, in the siesta manual under "*MD.Broyden.Cycle.On.Maxi*" input tag, it is written "*The alternative is to re-start the Broyden minimization algorithm from a first step of a diagonal inverse Jacobian (which might be useful when the minimization is stuck).*" Could anyone explain to me what is meant by this statement and which input tag setting it is referring to? Or, more generally, how can we play on the broyden method to get the optimization out of the stucking local minimum? Thanks.
For information, I am using siesta-4.1-b4 with the default settings of the Broyden option. Best, Boubacar
-- SIESTA is supported by the Spanish Research Agency (AEI) and by the European H2020 MaX Centre of Excellence (http://www.max-centre.eu/)
