[Wien] some queries from 2Doptimization
Peter Blaha
pblaha at theochem.tuwien.ac.at
Thu May 31 07:44:39 CEST 2018
When doing a c/a variation, it is usually nonsense without additional
optimization of the internal positions (force minimization) (except for
the few cases like an hcp crystal, where there are no forces).
So repeat the whole !!! procedure, but switch on "-min" in the running
scripts for every data point.
PS: I guess your results are inconsistent for some numerical issues and
inconsistent data.
Your lowest energy data point is at:
> lowest data point: -197507.047435900 14.3863200000000
> 35.5987000000000
but the minimum from 2Doptimize is far away from that. You must always
look at the individual E vs. c/a curves (for each volume). If the fit
deviates too much from the calculated values, something went wrong and
you can forget this.
You also get a plot min-c/a vs. volume. Is this a smooth curve or are
there irregularities ? And finally, check the final plot energy vs.
Volume, if all data points fall on the fitted line, or if there are
large irregularities. Don't just trust some numbers which a fitting
procedure would provide.
Your E-variations (I can see numbers varying by 160 mRy) is too large,
you are probably outside the "parabolic regime". Make smaller steps.
Am 30.05.2018 um 17:46 schrieb Dr. K. C. Bhamu:
> Dear Wien2k Users,
>
> I finished lattice parameter optimization with 2Doptimize (set2D) script
> for AFM case.
>
> I gave me a nice parabola fit for MB Eqn.
>
> At the end of ana2D_lapw I see below parameters
>
>
>
>
> #####################################
> Vol-opt= 6477.2543 , coa-opt= 2.4690
>
> a0= 13.7918 , c0= 34.0519 bohr
> a*0= 7.2983 , c0= 18.019 Ang
> *
> *
> *
> *or a0=13.79178818 c0= 34.050975
> Bohr--------------------------------------------------------------------
> Say this is Eqn.-(1)*
>
>
>
>
> then "parabolfit_lapw -t 2 -scf '*.scf'" gave me below parameters (the
> pink and blue may be "a" and "c", respectively):
>
>
> AFM.ene and AFM.latparam generated
> Enter dimension of fit (number of variable lattice parameters, 1-6):
> 2 fitcase 6 parameter
> lowest data point: -197507.047435900 14.3863200000000
> 35.5987000000000
> 36 -197506.997315480 -197506.995255490
> -197506.995195780 -197506.989177860 -197506.989866030
> -197506.979961790 -197507.046392950 -197507.044306660
> -197507.045024960 -197507.038881620 -197507.040400960
> ************** skipping some data here************
> -197506.841649710 -197507.039826230 -197507.041358040
> -197506.834684780 -197506.829326540 -197507.031120700
> -197506.824849460
>
> I INITIAL X(I) D(I)
>
> 1 -0.197507D+06 0.100D+01
> 2 0.100000D+00 0.102D+01
> 3 0.143863D+02 0.600D+00
> 4 0.100000D+00 0.761D+01
> 5 0.355987D+02 0.608D+00
> 6 0.100000D+00 0.155D+01
>
> IT NF F RELDF PRELDF RELDX MODEL STPPAR D*STEP
> NPRELDF
>
> 0 1 0.848D+00
> 1 2 0.261D-02 0.10D+01 0.10D+01 0.2D-05 G 0.0D+00 0.7D+00
> 0.10D+01
> 2 3 0.121D-03 0.95D+00 0.96D+00 0.2D-06 G 0.0D+00 0.1D+00
> 0.96D+00
> 3 4 0.109D-03 0.10D+00 0.10D+00 0.2D-08 G 0.0D+00 0.9D-03
> 0.10D+00
>
> ***** X-CONVERGENCE *****
>
> FUNCTION 0.108736D-03 RELDX 0.210D-08
> FUNC. EVALS 4 GRAD. EVALS 24
> PRELDF 0.100D+00 NPRELDF 0.100D+00
>
> I FINAL X(I) D(I) G(I)
>
> 1 -0.197507D+06 0.100D+01 -0.319D-06
> 2 0.172755D+00 0.129D+01 0.251D-04
> 3 0.144346D+02 0.411D+00 0.223D-05
> 4 0.102224D-01 0.822D+01 -0.376D-06
> 5 0.357076D+02 0.608D+00 -0.462D-08
> 6 0.382409D-01 0.187D+01 -0.262D-07
> Parabolic equation of state: info 3
> E = x1 + x2(a-x3)^2
> + x4(b-x5)^2 + x6(a-x3)(b-x5)
> Fitparameter are
> -197507.051608 0.172755 14.434559 0.010222
> ------------------ *Say this is Eqn.-(2)*
>
> 35.707596 0.038241
> lattic parameters energy de(EOS)
> 14.021960 34.697070 -197506.997315 0.001500
> 13.929700 35.158170 -197506.995255 0.001373
> ************** skipping some data here************
>
>
> My queries are:
>
> 1. I did not relax the forces for optimization as there always be some
> forces due to lattice strains if we optimize the structure. So I wish to
> relax the structure with these parameters and then wish to do further
> calculations. Am I right?
>
> 2. Why there is the difference between the lattice parameters obtained
> in Eqn.(1) and Eqn. (2), and which data I should consider for further
> calculations/
>
> 3. The B-M Eqn fit gave me below parameters:
> V0,B(GPa),BP,E0 6477.0172 25.2686 5.0558
> -197507.048851
>
> while in graph obtained from B-M. Eqn I see a slight variation in these
> parameters:
> V0,B(GPa),BP,E0 6477.6387, 24.7964, 5.7605, -197507.048792
>
>
> 4. In the link (provided below) for 2Doptimization, it is mentioned for
> clmextrapol_lapw that "Convert it to #clmextrapol_lapw for SP case"
>
> As my case is AFM and I did not follow this instruction, how it may
> influence mine optimizes lattice parameters?
>
>
> http://susi.theochem.tuwien.ac.at/reg_user/textbooks/2Doptimize.pdf
> <http://susi.theochem.tuwien.ac.at/reg_user/textbooks/2Doptimize.pdf>
>
>
> It would be nice for me if I get inlined reply.
>
>
>
> Kind regards
>
> Bhamu
>
>
>
>
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
>
--
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300 FAX: +43-1-58801-165982
Email: blaha at theochem.tuwien.ac.at WIEN2k: http://www.wien2k.at
WWW:
http://www.imc.tuwien.ac.at/tc_blaha-------------------------------------------------------------------------
More information about the Wien
mailing list