Discontinuity in the phonon branches

[apandit2021]

Dear Dr. Tadano,

I am getting unusual fluctuations at/along some wave-vectors boundaries for several phonon branches as in the attached image below. I checked by changing some of the input parameters as well. But the pattern is still the same. Do you have any suggestions on this issue?

Best,
Abhiyan

[ttadano]

Hi. Thank you for the report. Could you upload a higher resolution figure? I can't see the "fluctuation" you mentioned. Also, did you get a similar problem in the harmonic phonon calculation?

[apandit2021]

Dear Dr. Tadano,

I think the image attached below should be clear enough. By the "fluctuation", I mean if you look some of the phonon branches closely at X- and Y k-points, mainly for the lower optical branches, there is sudden rise/fall at the intersection, which is not smooth. I didn't have this problem in the harmonic phonon calculations.
For your view, following is the input file I used for the scp calculation (I also checked by increasing the KMESH_*, but the problem is still there):
&general
PREFIX = 551_scph
MODE = SCPH
FCSXML = *551_quartic_Opt.xml
NKD = 2
KD = * *
NONANALYTIC = 2
BORNINFO = *.born
TMIN = 0; TMAX = 700
DT = 50
/

&scph
SELF_OFFDIAG = 0
MAXITER = 4000
MIXALPHA = 0.01
KMESH_INTERPOLATE = 5 5 1
KMESH_SCPH = 10 10 2
TOL_SCPH = 1.0e-8
/

&cell
50.00 # factor in Bohr unit
0.1753993041 0.0000000000 0.0000000000
0.0000000000 0.1730004475 0.0000000000
0.0000000000 0.0000000000 0.8353772789
/

&kpoint
1 # KPMODE = 1: line mode
G 0.0 0.0 0.0 X 0.5 0.0 0.0 51
X 0.5 0.0 0.0 S 0.5 0.5 0.0 51
S 0.5 0.5 0.0 Y 0.0 0.5 0.0 51
Y 0.0 0.5 0.0 G 0.0 0.0 0.0 51
/

#&kpoint

2

30 30 2

#/

&analysis
PRINTMSD = 1
/

[ttadano]

This is the expected behavior of the mixed-space approach NONANALYTIC=2, which gives discontinuous dispersion curves at q points that are not commensurate with the supercell size. In your case, you used a 5x5x1 supercell, where the Brillouin zone boundaries such as X and Y are not commensurate. This problem should not happen when you used a 2m x 2m x 1 supercell (m: integer).

To solve this problem, please run an SCPH calculation with NONANALYTIC=2 first and then switch to NONANALYTIC=3 in the restart mode (RESTART_SCPH=1). The Ewald method (NONANALYTIC=3) is expected to give continuous curves. Note: This two-step procedure is necessary for the current implementation which does not support NONANALYTIC=3 in the SCPH iteration.

BTW, I think setting KMESH_SCPH = 10 10 1 gives the same result as KMESH_SCPH = 10 10 2 with lower computational costs.

[apandit2021]

Dear Dr. Tadano,

Thank you for the explanation. Accordingly, I first performed the SCPH run and then used the restart mode calculation with the input tags as below. I already have PREFIX.scph_dymat generated, where I used ALAMODE-V-1.2.0. However, I am getting error with the input settings. Could you please look and let me know where I need the correction?

&general
PREFIX = **
MODE = RESTART
FCSXML =551_quartic_Opt.xml
NKD = 2
KD = * *
NONANALYTIC = 3
BORNINFO = *.born
TMIN = 0; TMAX = 700
DT = 50
/

&scph
RESTART_SCPH = 1
SELF_OFFDIAG = 0
MAXITER = 4000
MIXALPHA = 0.01
KMESH_INTERPOLATE = 5 5 1
KMESH_SCPH = 10 10 1
TOL_SCPH = 1.0e-8
/

&cell
50.00 # factor in Bohr unit
0.1753993041 0.0000000000 0.0000000000
0.0000000000 0.1730004475 0.0000000000
0.0000000000 0.0000000000 0.8353772789
/

&kpoint
1 # KPMODE = 1: line mode
G 0.0 0.0 0.0 X 0.5 0.0 0.0 51
X 0.5 0.0 0.0 S 0.5 0.5 0.0 51
S 0.5 0.5 0.0 Y 0.0 0.5 0.0 51
Y 0.0 0.5 0.0 G 0.0 0.0 0.0 51
/

#&kpoint

2

30 30 2

#/

&analysis
PRINTMSD = 1
/

Best,
Abhiyan

[ttadano]

Please use MODE = SCPH.

[apandit2021]

Dear Dr. Tadano,

I used MODE = SCPH also before. However, it gives the error as below:

..................
..................
Diagonalizing dynamical matrices for all k points ...
ERROR in add_longrange_matrix MESSAGE: Hermiticity of Dynamical matrix is broken.

Any help/suggestion is highly appreciated.

Best,
Abhiyan

[ttadano]

Please try the GitHub version (develop branch). I have recently updated the related source code.

[apandit2021]

Dear Dr. Tadano,

Did you mean the ALAMODE-V-1.2.0? I tried with this version as well, but the same error is there as well.

Best,
Abhiyan

[ttadano]

I made some corrections after the release of 1.2.0, but they are still not packaged as a new release. For the moment, please git pull the changes in the develop branch and see if the issue is fixed.

[apandit2021]

Dear Dr. Tadano,

The issue is resolved now. Thank you very much.

Best,
Abhiyan