Page 1 of 1

NaN sup. gap in Pade approximation

Posted: Sun Aug 13, 2023 1:37 pm
by timonmosko
Dear epw experts,

recently I have been trying to calculate superconducting properties of NbSe2 using epw code. It seems that there is no problem to calculate the imaginary part of gap , but the real one (in Pade approx.) always ends as NaN regardless the npade parameter (I have tried different values, as it was suggested in some other discussions at this forum, e.g. 80,40,30). Then the job crashes or turns into zombie.

Electron-phonon coupling strength = 0.7024506

Estimated Allen-Dynes Tc = 4.641897 K for muc = 0.16000

Estimated w_log in Allen-Dynes Tc = 20.310191 meV

Estimated BCS superconducting gap = 0.704014 meV


temp( 1) = 0.30000 K

Solve anisotropic Eliashberg equations on imaginary-axis

Total number of frequency points nsiw( 1) = 6156
Cutoff frequency wscut = 1.0000


Size of allocated memory per pool: ~= 0.1061 Gb
iter ethr znormi deltai [meV]
1 1.614501E+00 1.698695E+00 4.172803E-01
2 3.059249E+01 1.699975E+00 8.009037E-01
3 9.824474E-01 1.699304E+00 6.930678E-01
4 7.359392E-02 1.698898E+00 7.185265E-01
5 2.392814E-02 1.698728E+00 7.366301E-01
6 1.842136E-03 1.698693E+00 7.427669E-01
Min. value of superconducting gap = 0.000000 meV
Convergence was reached in nsiter = 6

iaxis_imag : 4791.63s CPU 4834.98s WALL ( 1 calls)


Pade approximant of anisotropic Eliashberg equations from imaginary-axis to real-axis
Cutoff frequency wscut = 1.0000

pade Re[znorm] Re[delta] [meV]
4924 NaN NaN

Convergence was reached for N = 4924 Pade approximants

Min. value of superconducting gap = 0.000000 meV




I attach the input and output files for epw, pw and ph codes. Thanks for any idea how to resolve this.
Timon

Re: NaN sup. gap in Pade approximation

Posted: Thu Aug 17, 2023 2:23 pm
by hpaudya1
Hi Timon,

Looks like your Pade approximation did not reach convergence. I recommend decreasing the value of npade (https://docs.epw-code.org/doc/Inputs.html#npade) in your input file. I hope it helps.

Happy EPWing,
Hari