[Wien] minimization of positions

[Peter Blaha]

How big are the forces for your initial structure ?

How many geometry steps were performed ?

grep :ENE *mini
grep :FOR00x *mini

If your initial positions are really that bad (and thus the forces are 
very big), it might be that you need smaller values in case.inM (or that 
you need to restart by removing case.tmpM). 

Sometimes atoms want to come closer than allowed with the selected RMTs.
In these cases you must restart with smaller spheres (Sometimes these 
messages appear only at an intermediate step, then you can ignore them).

With a recent WIEN2k version you can use all convergence criteria at once:
  
runsp -ec 0.0001 -cc 0.0001 -fc 1 -I -i 60


> 
> Dear Wien Users, 
> 
> I am trying to optimize the positions of a ternary system.
> Below I am giving full detail about that system
> system is 
> MnCoSn ( a = 4.258 Ang, c = 5.36) (Space group 186)
> Mn (0 0 1/4)
> Co (1/3 2/3 0.44=z1)
> Ge (1/3 2/3 0.06=z2)
> 
> To optimize the  z parameters of Co (z1) and Ge (z2) atoms.
> I am using PORT option as given in the manual.
> 
> Let me tell you one thing, my system positions are far away from the
> solution, because this system is only reported as Ni2In  structure type
> (space group 194) and I would like to minimize it in LiGaGe structure 
> type (space group 186). In space group 186 each atom will have their 
> own positions as given above. These positions I took from other ternary system. 
> 
> I follow this procedure 
> 1. runsp_lapw -ec 0.0001 -i 40 -r 15
> 2. runsp_lapw -cc 0.001 -i 40 -r 15
> 3. runsp_lapw -fc 1 -i 40 -r 15
> 
> then I Have edited case.inM file like this
> 
> PORT 2.0           # PORT/NEWT;  tolf (geom.opt. stops when F<;tolf) 
> 0.0 0.0 1.0 3.0    # Mn
> 0.0 0.0 1.0 1.0    # Co #PORT: delta1,2,3=precond., 4:Bondorder 
> 0.0 0.0 1.0 1.0    # Sn
> 
> when I start with ---> min -NI -i 4 -s 1 -j 'runsp_lapw -I -i 40 -fc 1.0 '  
> 
> It is running and my forces are very far from force criterion and at the 
> end in my dayfile it writes line like this -->
> 
> ##########################dayfile
> :ENERGY convergence:  0 0 .0000525000000000
> :CHARGE convergence:  0 0.0000 .0002335
> 
> >   stop
> >   mini        (01:07:58) 0.070u 0.015s 0:00.08 100.0% 0+0k 0+0io 0pf+0w
>  :WARNING, CURVATURE CONDITION FAILED
>  :WARNING, S.Y WAS -0.304898817526764
>  :WARNING: Step size reduced due to overlapping spheres -- check RMT
>  :WARNING: Step size reduced due to overlapping spheres -- check RMT
> ############################################################
> 
> my forces are looking like --->
> 
> ###########################################
> :ENE  : ********** TOTAL ENERGY IN Ry =       -34925.629006
> 
>        TOTAL FORCE IN mRy/a.u. = |F|     Fx             Fy             Fz
> :FOR001:   1.ATOM        114.688     0.000     0.000   114.688
> :FOR002:   2.ATOM        992.576     0.000     0.000   -992.576
> :FOR003:   3.ATOM       1776.551     0.000     0.000   1776.551
> 
> #############################################
> 
> Can anyone please explain me, what is the process to do such kind of 
> minimization, am I doing the right way? and what to do if I see 
> with iterations, forces are increasing on each atom and vice versa.
> 
> What is the criterion for choosing RMT values for such complicated 
> (I mean without knowing exact positions) systems.
> 
> 
> 
> ############################
> 
> In the begining, I thought, I am doing some thing wrong with 
> my calculations. then I tried to check these steps (as given above for 
> minimization) on a well reported system 
> 
> example LiGaGe  (a = 4.175, c = 6.783 Ang) 
> 	Li (0 0 1/4)
>         Ga (1/3 2/3 0.06)
>         Ge (1/3 273 0.44)
> 
> and it coverges without any error, in my dayfile 
> 
> #############################################
> >   stop
> >   mini        (16:09:10) 0.029u 0.009s 0:00.04 50.0%  0+0k 0+0io 0pf+0w
> 
> >   stop forces small, minimization stopped
> 
> >   relaxation steps summarized in LiGaGe_opt.scf_mini
> 
> >   relaxed structure and results in LiGaGe_opt.struct, scf, clmsum
> ##############################################
> 
> 
> and in my third optimized scf file :
> 
> ############################################# 
> :ENE  : ********** TOTAL ENERGY IN Ry =       -16202.425763
> 
>        TOTAL FORCE IN mRy/a.u. = |F|     Fx             Fy             Fz
> :FOR001:   1.ATOM          1.242          0.000          0.000    1.242
> :FOR002:   2.ATOM          2.019          0.000          0.000    -2.019
> :FOR003:   3.ATOM          0.428          0.000          0.000     0.428
> 
> ############################################
> 
> 
> If it is working for such stabilized system it should work for 
> my unrelaxed system. I think somewhere I missing some points.
> 
> What would be the good option for such system either PORT 
> or NEWT. Can I also edit DELTA values (range???) in my 
> case.inM file with PORT option like Prof. Blaha expalined
> for NEWT in FAQ of Wien. 
> 
> 
> I will be grateful for your kind reply and help
> 
> Thanking you in advance 
> hem kandpal (mainz)
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> 


                                      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/
--------------------------------------------------------------------------