[Wien] ENE and DIS oscillations in case of NM case

[Laurence Marks]

Reducing the GREED (aka mixing factor) does little; that it helps is a dft
myth for poor coding/models.

Probably nm does not converge because it is bad physics.

You have to do instgen, lstart then dstart; doing instgen later does
nothing. And, you need to read case.scfm to see what the spins really are.

On Sat, Jun 2, 2018, 12:52 Dr. K. C. Bhamu <kcbhamu85 at gmail.com> wrote:

> Dear Wien2k Users,
>
> For a A3B2C9 type system, I am running AFM, FM and NM cases with TEMPS
> 0.0018. TEMPS is always recommended with MSR1a for complex case by Prof.
> Marks.
>
> In literature, AFM structure is reported in the ground state while authors
> reported NM case to save computational cost.
>
> I am curious to see the stable phase among all three types with Wien2k and
> running all three cases (AFM, FM and NM) and have some queries for the same:
>
> 1. How the calculations may affect the :ENE or properties if we create
> case.inst file after initialization? In my case I see AFM and FM are
> looking exactly same with no difference.  What I did is:  init_lapw -b
> followed by instgen -ask for AFM and instgen for FM (instgen -nm for NM).
> does dstart make any trouble if I create these files after initialization?
>
>  Should I keep case.instgen file before initialization of can create after
> initialization?
>
> 2. AFM and FM are conversing withing ~110 scf cycles at -6% reduction and
> ~20 scf cycles for -4% reduction and so on but my NM calculations (instgen
> -nm with runsp_c_lapw and MSR1a) are going on even after ~250 scf cycles
> and still I do not see any hope for convergence as the ENE and DIS are
> oscillating.
>
> Should I reduce mixing factor or I have to worry for something elese?
> I reduced rmt 5% and then fixed the rmt of heavies element "Cs" to 2.35 to
> avoid any warnings of QTLB errors.
>
>
>
> Regards
> Bhamu
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20180602/4052c55f/attachment.html>