[Wien] find minimum

Gavin Abo gsabo at crimson.ua.edu
Sat Jan 25 03:51:17 CET 2014


For each constant of a (a1-a5), you should only have one number.

If I remember correctly, this happens when you do not use 5 points.

In other words, you must have only 5 case_coa___*.scf files in your case 
directory. Keep the five case_coa___*.scf files closest to the minimum.  
Move (or delete) the other case_coa___*.scf files.  Then, plot E vs. c/a 
again. 
[http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg05826.html]

On 1/24/2014 7:29 PM, Lawal Mohammed wrote:
> Dear Gavin Abo, Sir,
>
> In line with the discussion above, I got when I plot E vs c/a the 
> following
>
> Fit of:  E = a1 + a2*x + a3*x^2 + a4*x^3 + a5*x^4
> a11.000
> a20.000    1.000
> a3    -0.761   -0.0001.000
> a4    -0.000   -0.930    0.000    1.000
> a5     0.6420.000    -0.972-0.000    1.000
>
> In this case what should be the value for a2, a3, a4, and a5?
>
> Thank you in advance for your response.
> With kind regards
> Mohammed Lawal
>
>
>
> On Friday, January 24, 2014 7:08 AM, Gavin Abo <gsabo at crimson.ua.edu> 
> wrote:
> Below is a summary of previous posts that should help.
>
> To plot energy vs b/a:
>
> (a) You should be able to rename the scf files and then plot with the 
> energy vs c/a option of eplot 
> [http://zeus.theochem.tuwien.ac.at/pipermail/wien/2009-January/012053.html].
> (b) You could also plot energy vs b/a in another program like Origin 
> [http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg08841.html].
>
> To get the optimized c/a (or b/a) value,  you can calculate it from 
> the curve fit equation 
> [http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg05826.html].
>
> An example calculation is given below.
>
> When you plot E vs. c/a with eplot, you should get something like:
>
> Fit of:  E = a1 + a2*x + a3*x^2 + a4*x^3 + a5*x^4
> a1              = -2201.24
> a2              = 0.00233422
> a3              = 0.000188276
> a4              = 1.33462e-05
> a5              = 9.76521e-07
>
> From dE/dx = 0, we get the equation:
>
> a2 + 2*a3*x+3*a4*x^2+4*a5*x^3 = 0
>
> Solving this equation for x, x = -7.7555, which is the optimized c/a 
> ratio in %.  I got x with Octave, but there are other software 
> programs that you can find on the internet to solve the polynomial 
> equation.
>
> Octave input:
>
> dEdx = [4*9.76521e-07,3*1.33462e-05,2*0.000188276,0.00233422]
> roots(dEdx)
>
> Octave output:
>
>   -1.2474 + 8.6889i
>   -1.2474 - 8.6889i
>   -7.7555 + 0.0000i  <= This polynomial root is chosen as x, because 
> the imaginary part is zero and it appears in the energy vs c/a plot.
>
> If you also want the minimum E, just plug the x value back into the 
> curve fit equation and solve for E (= -2201.2495).
>
> On 1/22/2014 7:22 AM, MAHDI SALMANI HIRMAND wrote:
>> Dear Prof. Blaha,
>> Please let me know why there is not a program for finding minimum of 
>> c/a or b/a ratio with Wien2k Package when we optimize c/a or b/a.
>> eplot program can show curve of Energy Vs. c/a but it doesn't find 
>> minimum of c/a.
>> Please guide me
>> With best
>> Mahdi

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20140124/dc921475/attachment.htm>


More information about the Wien mailing list