How to get robust convergence performance?

[Ke Zhou]

Dear All, 

Now I am simulating the strained (about 0.15, close the fracture strain) single-layer graphene. 

But it cannot get well convergence after running for about ~100 or ~ 1000 steps.

I used OT method (with DIIS or CG minimizer) or direct diagonalization method. But both methods are not working.

Can anyone give me some suggestions?

Best,

Justin

##############

####### for OT #######

      CHOLESKY OFF

      &OT

        PRECONDITIONER FULL_ALL

        ! the most robust choice (DIIS might sometimes be faster, but not as stable).

        MINIMIZER DIIS ## CG

        N_DIIS 5

       ENERGY_GAP 1.0E-5

      &END OT

###### for diagonalization #######

      ADDED_MOS 100

      CHOLESKY OFF #INVERSE

      &SMEAR ON

        METHOD FERMI_DIRAC

        ELECTRONIC_TEMPERATURE [K] 300

      &END SMEAR

      &DIAGONALIZATION

        ALGORITHM STANDARD

        EPS_ADAPT 0.01

      &END DIAGONALIZATION

....

      &XC_GRID

             XC_SMOOTH_RHO NN50

             XC_DERIV NN50_SMOOTH

      &END

[Ke Zhou]

More information. 

The graphene is physically stable in this strained field.

Best,

Ke 

在 2020年7月18日星期六 UTC+8下午11:05:22，Ke Zhou写道：