[Wien] Volume optimization to monoclinic structure

[Gavin Abo]

Sorry, I didn't check the output of "parabolfit_lapw -h".  After 
generating "case.ene" and "case.latparam" with parabolfit_lapw, WIEN2k's 
non-linear least squares fitting tool "x eosfit6" is used.  I think you 
are right, that gnuplot's non linear fitting tool could be used instead 
of eosfit6 on the data in the files. Also, Origin's Nonlinear Curve 
Fitting [ https://www.originlab.com/doc/Origin-Help/Nonlinear-Curve-Fit 
], Matlab's Nonlinear Least Squares [ 
https://www.mathworks.com/help/optim/nonlinear-least-squares-curve-fitting.html 
], or another nonlinear data fitting program would likely work too.  
Though, eosfit6 should be working, while the other tools may take some 
additional effort to read, format, and/or fit the data.

username at computername:~/wiendata/TiC$ parabolfit_lapw -h
parabolfit_lapw is an interface for fitting E vs. 2-4-dim lattice 
parameters
by a non-linear least squares fit (eosfit6) using PORT routines.
Once you have several scf calculations at different volumes (usually 
generated
with "optimize.job") it generates the required "TiC.ene" and "TiC.latparam"

parabolfit_lapw [ -t 2/3/4 ]  [ -f FILEHEAD ] [ -scf '*xxx*.scf' ] [-a/b/g]

-a/b/g allows to specify alpha,beta or gamma in 4D fit.

On 5/8/2018 7:34 AM, Víctor Luaña Cabal wrote:
> On Tue, May 08, 2018 at 06:22:25AM -0600, Gavin Abo wrote:
>> Currently, I believe parabolfit_lapw is the best way available at this
>> time.  I'm not currently aware of any better fitting methods. If you
>> find a better way, feel free to share.
> I don't even know parabolfit_lapw, but I ever try the simplest tool
> available, and the non linear fitting tool of gnuplot works for almost
> anything. You create first a data file
>
> # comment
> # data description
> # var1   var2   var3  function
> #   a      b      c       E
>    1.234  1.342  1.785  -764.456
> .....
>
> then create in gnuplot the mathematical description of the f(v1,v2,v3)
> and then, ggain in gnuplot
> E(V,k1,k2,k3) = ......
>
> fit f(x,y,z) 'datafile' using ($1*$2*$3):4 via  k1,k2,k3
>
> whithin gnuplot a 'help fit' will provide you a detailed description
> and examples.
>
> It will not be the best method but an all-purpose tool.
>
> Best regards,
>               Víctor Luaña
> --
>      .  .    "Never let your sense of morals prevent you from
>     / `' \   doing what is right."
>    /(o)(o)\  -- Salvor Hardin, "Foundation"
>   /`. \/ .'\
> /   '`'`   \ "Freedom!, freedom!, freedom! After that put whatever
> |  \'`'`/  | term you like"
> |  |'`'`|  |  --Largo Caballero (socialist, spanish president in the exile,
>   \/`'`'`'\/   shortly before dying)
> ==(((==)))===================================+===========================
> ! Dr.Víctor Luaña, in silico chemist & prof. ! A person is slave of his words
> ! Departamento de Química Física y Analítica ! and owner of his silences.
> ! Universidad de Oviedo, 33006-Oviedo, Spain !
> ! e-mail:  <victor at fluor.quimica.uniovi.es>  ! What has been seen cannot
> ! phone: +34-984080927    fax: +34-985103125 ! be unseen, ¿are you sure?
> +--------------------------------------------+
>   GroupPage: <http://azufre.quimica.uniovi.es/>
>   Articles:  <http://scholar.google.com/citations?user=Ibl1BWAAAAAJ&hl=es>
>   git-hub:   <https://github.com/aoterodelaroza>
>   ORCID: 0000-0003-4585-4627; RID: H-2045-2015