How to improve the accuracy of the results?

[yangzw1985]

Hi ,everyone! I have a question about the accuracy of the results. I have calculate the transimission spectrum, but the coefficients are all zero. The results are 0.0000. If I want to the results are more accuracy as:0.000000000 etc, I want to know if it is possible. And what should I do in order to obtain more accuracy? Thanks in advance!

[Anders Blom]

Most likely you use a NanoLanguage script produced by VNL, in which case the part that prints the transmission spectrum contains the line

Code

if processIsMaster(): nlPrint(transmission_spectrum)

To print the values with proper accuracy, replace that line by something like

Code

energies = transmission_spectrum.energies()
coefficients = transmission_spectrum.coefficients()
print 'Energy (eV)     Transmission'
print '-----------------------------------'
for i in range(len(energies)):
    print "%g\t%g" % ( energies[i].inUnitsOf(eV),coefficients[i] )

Updated: Fixed error in code

[yangzw1985]

Thank you ! Great!
But it has an error, we have corrected it. It should be like this:


energies = transmission_spectrum.energies()
coefficients = transmission_spectrum.coefficients()
print 'Energy (eV)     Transmission'
print '-----------------------------------'
for i in range(len(energies)):
    print "%g\t%g" % ( energies.inUnitsOf(eV),coefficients )

Moderator note: This code doesn't work

[Anders Blom]

Hmm, I'm sorry, I don't think your code will work.
"energies" is an object of type TransmissionSpectrum, and has no method inUnitsOf. Besides, even if it did, your loop would print the same thing for every step in the loop (each time the whole transmission spectrum).
Try my version, I'm sure it works

Update: Now it works after I fixed it!

[yangzw1985]

But it really is !
we have check my code with li-h2-li,it is right.
but in your code ,it has an error.

[Anders Blom]

Ok, I think I know what might be the cause. Your test case probably only has a single point in the transmission spectrum? Post the script that works (Li-H2-Li) so we can confirm this, and I can update my version to work in general, for all cases Because your version will not work for a "usual" transmission spectrum with, say, 100 points from -2 to +2 eV.

[yangzw1985]

if I want to calculate the transimission spectrum of a li-znowire-li with 200 points -6--6 ev, what should I do ?

there is something with my code exhibited in the screen.
For your code, the error was occured in the second lines.

energies = transmission_spectrum.energies()
coefficients = transmission_spectrum.coefficients()


electron_transmission should be changed by transmission_spectrum

Is that right?

[Anders Blom]

Ah! I copy-pasted from the manual, and forgot to change in one place...   Thanks for noticing it. So, the correct code should be

Code: python

energies = transmission_spectrum.energies()
coefficients = transmission_spectrum.coefficients()
print 'Energy (eV)     Transmission'
print '-----------------------------------'
for j in range(len(energies)):
    print "%g\t%g" % ( energies[j].inUnitsOf(eV),coefficients[j] )

To get 201 points (better than 200 since otherwise you don't get a point at E=0) from -6 to 6 eV, you could do

Code: python

from numpy import arange
energies = arange(-6,6.01,12./200)*eV
transmission_spectrum = calculateTransmissionSpectrum(scf,energies) 

Assuming, that is, that the self-consistent object is called "scf" and that the system only needs (1x1) k-point sampling in the XY plane.

[yangzw1985]

Yes, this time is right.
Thanks!