[Wien] accuracy test for RKMAX

Mon Jul 2 08:34:57 CEST 2007

Why do you expect the minimum energy at 0% change of the volume ?

If you like to have it for the plot then just use the optimized lattice parameter calculated from one of the EOS fits for your starting structure in the optimization.
However, I doubt that this is anything meaningfull.

Ciao
Gerhard

-----Ursprüngliche Nachricht-----
Von: wien-bounces at zeus.theochem.tuwien.ac.at im Auftrag von santhy kathirvelu
Gesendet: Mo 02.07.2007 07:50
An: A Mailing list for WIEN2k users
Betreff: Re: [Wien] accuracy test for RKMAX

Dear Prof. Blaha,
   I tried the calculation as same as in that exercise only.  I did volume optimization (-6,-3, 0, 3, 6% volume changes) with RKMAX=5 and k point=1000.  It shows that -6% volume changes gave the minimum energy value.  Then I repeated the same calculation with the RKMAX=7 and k-point=1000, this time also it shows that -6% of volume changes only gives minimum value.   By your advice I included volume changes from -9 to 9 which gave parabola with minimum energy in -6% volume changes

  I tried for RKMAX=7.2 and k-point=1000, in that case also -6% volume changes gives the minimum energy.

  ( In the case of TiC, RKMAX=7  and k-point =100 &1000, gives 0% volume changes gives minimum energy value.)

  How to choose the RKMAX in such a way that to get 0% volume changes will give the minimum energy?

  Thanks in advance.

  With regards,
  Santhy Jaiker

Peter Blaha <pblaha at theochem.tuwien.ac.at> wrote:   Why did you increase RKMAX to touching spheres ?
Of course then you cannot do a volume optimization.

If you follow the instructuions on this "exercises" it tells you you must do
these convergence tests for all volume points in your curve.

Then you have just tested k-points, but not RKMAX !

> By Prof.Blaha's advice, I added few more negative volumes and then I
> plotted the results, it gave nice parabola and shows that at -6 only
> gives the min. energy.
>
> Then I checked the accuracy test to k-point, I repeated the calculation
> with 1000, 2000 and 3000 for k point. I listed result below,
> k-point RKM Enegry Volume
> *1000 6.93 -1707.621751
> 113.36162*
> 2000 6.99 -1707.621552 ,,
> 3000 6.61 -1707.621562 ,,
>
> Therefore, for Ti-bcc system with RMT=2.5, RKMAX=7, RKM=6.93, GMAX=12
> and k-point =1000 gives the optimized structure with volume V=113.36161
> (a = 3.226757A) and E=-1707.621751Ry. Is it correct?
>
>
> In the above calculation, I used RMT of Ti is 2.5(default). From the
> mailing list (site:
> http://zeus.theochem.tuwien.ac.at/pipermail/wien/2004-April/002424.html)
> I changed the RMT of Ti from 2.5 to 2.69 (because nn-distance = 5.39080).
> I used remaining parameters as same as the above calculation (RKMAX=7,
> GMAX=12 and k-point=1000). Using this data, I am able to run SCF, but
> not volume optimization with negative volumes because RMT is close to
> nn-distance (positive volumes give higher energy and it is not possible
> to get a parabola here). This calculation is almost look like single
> point energy calculation (i.e. it gives the energy of the given
> structure) and there is not much energy difference from the above
> calculation. I listed the result below.
>
> RMT=2.69, RKMAX=7, RKM=6.89, GMAX=12 and k-point =1000
> V=120.59737 (a = 3.294) and E=-1707.621713Ry
>
> Then how can I do the optimization?
>
> Thanks in advance.
>
> With regards,
> Santhy Jaiker
>
>
>
> */Peter Blaha
/* wrote:
>
> Hi,
>
> I hope it is clear that you CANNOT use min_lapw for your bcc Ti example.
>
> With respect to the volume optimization, I don't see anything bad.
>
> Specifying RKMAX does not fix RKMAX to this value, it just guarantees
> that it does not get bigger than what you specified, but when changing
> the lattice parameter, it is natural that RKMAX changes.
>
> You shoulds add a few more negative volumes (-9 -12 maybe -15) and then
> plot the results because it seems the minimum is near -6. When you
> get a
> nice parabola it seems ok, if not it may happen because of too low
> RKMAX
> (gotro 7.5, 8, up to 9) or to few k-points (100 is very little).
>
> > I am a new user of wien2k. I had gone through the FAQ and L.D. Marks
> > notes for optimization. I started with small system like Ti-BCC.
> > Before going to mini_lapw, I tried the accuracy test as same as
> given
> > in the site http://www.wien2k.at/events/ws2006/Exercises.pdf.
> >
> > I gave lattice parameter= 3.294A°, RMT=2.5(default),
> RKMAX=7(default),
> > k-point=1000 and Gmax=12(default) for Ti-BCC system. I tried the
> same
> > step in the above site,
> > STEP 1: run x optimize and generate 5 structures (-6, -3, 0, 3, 6 %
> > volume change)
> > STEP2: edit "optimize.job".
> > STEP3: run optimize.job and plot the results.
> > I got
> > SCF File RKM Energy
> > Vol -6.0.scf *6.93 -1707.621717*
> > Vol -3.0.scf 6.85 -1707.621662
> > Vol 0.0.scf 6.79 -1707.621239
> > Vol 3.0.scf 6.72 -1707.619926
> > Vol 6.0.scf 6.65 -1707.618133
> >
> > Energy is high when RKM is 6.93. Since I changed RKMAX=6.93 in
> Ti.in1
> > file and I didn't change any other parameter. I repeated those 3
> steps.
> > I got
> >
> > SCF File RKM Energy
> > Vol -6.0.scf *6.93 -1707.620994*
> > Vol -3.0.scf 6.85 -1707.620941
> > Vol 0.0.scf 6.79 -1707.620476
> > Vol 3.0.scf 6.72 -1707.619422
> > Vol 6.0.scf 6.65 -1707.617423
> >
> > Here I observed that even though same RKMAX (=6.93) gives different
> > energy. And I couldn't keep RKMAX as constant. Since, during SCF
> > calculation it changes.
> >
> > Sorry I took more of your time. I have some experience to handle
> CASTEP
> > (in windows mode). But I couldn't correlate things in WIEN2k.
> >
> > It will be really helpful, if you tell, how to do the accuracy
> test for
> > RKMAX.
> >
> > Thanks in advance.
> >
> > Santhy Jaiker
> >
> >
> >
> ------------------------------------------------------------------------
> > Here's a new way to find what you're looking for - Yahoo! Answers
> >
> >
> >
> >
> ------------------------------------------------------------------------
> >
> > _______________________________________________
> > Wien mailing list
> > Wien at zeus.theochem.tuwien.ac.at
> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
>
>
>
> with love,
> K. Santhy
>
> **********************************************
> A truly happy person is one
> who can enjoy the scenery on a detour
> **********************************************
>
> Send free SMS to your Friends on Mobile from your Yahoo! Messenger.
> Download Now! http://messenger.yahoo.com/download.php
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> 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/
--------------------------------------------------------------------------
_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien

with love,
 K. Santhy

 **********************************************
 A truly happy person is one
 who can enjoy the scenery on a detour
 **********************************************

---------------------------------
 Here's a new way to find what you're looking for - Yahoo! Answers