[Wien] about mini

[Peter Blaha]

>        I have trouble in minimizing 40-atom supercell containing two interface. I relaxed the atomic positions only in the direction normal to the interface.
a) You can put as many iterations as you like (i.e. a maximum of 999 is still
   present). The comments about 99 iterations apply to an older WIEN version.

b) Before running a minimization, one should try to understand why the
convergence is so poor. Usually, nonmagnetic systems should converge more
reasonable, in particular once you have a converged solution and make a
small movement of some atoms.
Do you have transition metals in your structure ?
How many k-points ? (maybe increased mesh helps)
What mixing ? (reduced factor ?)
Any improvements when using TEMP 0.008  ?


c) In principle you can simply repeat your  "min -command", maybe with more
iterations. BUT some cases may NOT converge in 1000 iterations when you do not
adapt some parameters as indicated above.

I input the command:min_lapw -I -s 1 -j ' run_lapw -I -i 97
> -fc 2 ', but after doing 3 structure changes, it stopped. In *.dayfile it contains the following information:
>
> cycle 97  (Fri Apr 16 04:17:20 UTC 2004)
>     cycle 97  (1/4 to go)
> >   lapw0  (04:17:20) 111.810u 0.170s 1:51.98 100.0% 0+0k 0+0io 169pf+0w
> :FORCE convergence:  0 2 0 XCO 6.8100000000000000 ZCO 8.4055000000000000 ZCO 0 XCO 10.9405000000000000 ZCO 2.8575000000000000 ZCO 6.5700000000000000 ZCO 5.8840000000000000 ZCO 3.7310000000000000 ZCO 5.3135000000000000 ZCO 5.2575000000000000 ZCO 3.1515000000000000 ZCO 0 XCO 0 XCO 0 XCO 3.8530000000000000 ZCO 5.0085000000000000 ZCO
> >   lapw1    (04:19:15) 1066.230u 9.080s 17:55.34 99.9% 0+0k 0+0io 290pf+0w
> >   lapw2  (04:37:10) 412.870u 7.850s 7:00.74 99.9% 0+0k 0+0io 223pf+0w
> >   lcore (04:44:11) 0.450u 0.000s 0:00.44 102.2% 0+0k 0+0io 93pf+0w
> >   mixer (04:44:12) 9.820u 0.180s 0:10.01 99.9% 0+0k 0+0io 110pf+0w
> :ENERGY convergence:  0 0 .0004860000000000
> :CHARGE convergence:  0 0 .0104920
>     cycle 98  (Fri Apr 16 04:44:22 UTC 2004)
>     cycle 98  (0/3 to go)
> >   FORCES NOT CONVERGED

                                      P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------