Page 1 of 2

RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Mon Apr 08, 2024 1:45 pm
by lisg_sari
Dear All,

Can I bring up this issue again with respect to converging linear response calculations using the LEPSILON tag ?
I saw the old post for a while ago; is there something one can do with the newer version of VASP 6.4.2?

The errors below show a somewhat complicated system with transition metal oxide surface slab and frequency dependent macroscopic dielectric tensor was calculated using PBE+U.
The geometry, vibrational frequency calculations were fine; linear response without local field was also good.
When Linear response to external field was calculated, the energies jump up and down.

Could you please give some comments on the following questions to solve this issue?

1) is there an option to try a different optimizer ?
2) is it possible to play around for AMIX and BMIX to help converge? ( I am using: AMIX=0.4, BMIX=1.0)
btw, I tried both IALGO = 58 and 38.
3) the components at X, Y axis look not that bad, but the component in Z axis looks horrible, is there a way specifically to get it better ?
with Z-axis improvement, I am assuming qualitative results may be obtained.

Thanks a lot.


===============================================================================
Linear response to external field (no local field effect), progress :
Direction: 3
N E dE d eps ncg rms rms(c)
RMM: 1 -0.160069062237E+02 -0.16007E+02 -0.77091E+00 5502 0.203E+01
RMM: 2 -0.366082663562E+02 -0.20601E+02 0.45920E+01 4656 0.783E+00
RMM: 3 -0.420237261512E+02 -0.54155E+01 0.89858E+00 5207 0.300E+00
RMM: 4 -0.432981611292E+02 -0.12744E+01 0.24799E-01 6041 0.119E+00
RMM: 5 -0.434707240623E+02 -0.17256E+00 -0.26931E-01 6635 0.399E-01
RMM: 6 -0.434801493410E+02 -0.94253E-02 -0.26132E-02 6973 0.106E-01
RMM: 7 -0.434811811397E+02 -0.10318E-02 -0.42259E-03 6670 0.428E-02
RMM: 8 -0.434813929274E+02 -0.21179E-03 -0.92494E-04 6549 0.190E-02
RMM: 9 -0.434814363821E+02 -0.43455E-04 -0.19688E-04 6148 0.711E-03
change of polarisation eV/A/(eV/A) component 3 : 1.560 -2.862 86.963
dielectric tensor component 3 : 0.066 -0.121 4.665
Linear response to external field, progress :
Direction: 1
N E dE d eps ncg rms rms(c)
RMM: 1 -0.143474435784E+02 -0.14347E+02 -0.56946E+00 5597 0.190E+01
RMM: 2 -0.488183146020E+02 -0.34471E+02 0.61114E+01 4612 0.691E+00 0.375E+01
RMM: 3 -0.518670200485E+02 -0.30487E+01 -0.42160E+01 5463 0.829E+00 0.349E+01
RMM: 4 -0.471772930838E+02 0.46897E+01 -0.68562E+00 5024 0.346E+00 0.124E+01
RMM: 5 -0.458783190625E+02 0.12990E+01 -0.21910E+00 5963 0.199E+00 0.783E+00
RMM: 6 -0.450060061545E+02 0.87231E+00 -0.14770E+00 5848 0.168E+00 0.485E+00
RMM: 7 -0.444965028833E+02 0.50950E+00 -0.16394E+00 5744 0.139E+00 0.350E+00
RMM: 8 -0.443319645257E+02 0.16454E+00 -0.18482E+00 5932 0.791E-01 0.148E+00
RMM: 9 -0.442448834439E+02 0.87081E-01 -0.16807E+00 6941 0.329E-01 0.927E-01
RMM: 10 -0.441767110430E+02 0.68172E-01 -0.17960E+00 7188 0.155E-01 0.512E-01
RMM: 11 -0.440817261277E+02 0.94985E-01 -0.17630E+00 7008 0.141E-01 0.302E-01
RMM: 12 -0.440651549227E+02 0.16571E-01 -0.19031E+00 7282 0.112E-01 0.223E-01
RMM: 13 -0.440374431632E+02 0.27712E-01 -0.17473E+00 7588 0.893E-02 0.155E-01
RMM: 14 -0.440401460685E+02 -0.27029E-02 -0.18773E+00 7807 0.797E-02 0.135E-01
...........
RMM: 33 -0.440297494135E+02 0.99641E-02 -0.17689E+00 10632 0.514E-02 0.296E-03
RMM: 34 -0.440406352076E+02 -0.10886E-01 -0.18837E+00 10794 0.514E-02 0.317E-03
RMM: 35 -0.440333956727E+02 0.72395E-02 -0.18109E+00 10734 0.513E-02 0.272E-03
RMM: 36 -0.440403239160E+02 -0.69282E-02 -0.18745E+00 10821 0.514E-02 0.286E-03
RMM: 37 -0.440377098242E+02 0.26141E-02 -0.18462E+00 10746 0.513E-02 0.252E-03
RMM: 38 -0.440371548334E+02 0.55499E-03 -0.18432E+00 10797 0.513E-02 0.273E-03
change of polarisation eV/A/(eV/A) component 1 : 91.078 6.711 -0.727
dielectric tensor component 1 : 4.838 0.283 -0.031
Linear response to external field, progress :
Direction: 2
N E dE d eps ncg rms rms(c)
RMM: 1 -0.182917925034E+02 -0.18292E+02 0.98298E+00 5613 0.212E+01
RMM: 2 -0.676139043837E+02 -0.49322E+02 0.93511E+01 4655 0.849E+00 0.417E+01
RMM: 3 -0.823815393960E+02 -0.14768E+02 -0.88521E+00 5247 0.790E+00 0.315E+01
RMM: 4 -0.818694747612E+02 0.51206E+00 -0.32683E+00 5248 0.367E+00 0.185E+01
RMM: 5 -0.813384465338E+02 0.53103E+00 -0.22896E+00 5507 0.238E+00 0.115E+01
RMM: 6 -0.804983823201E+02 0.84006E+00 -0.26498E+00 5901 0.192E+00 0.614E+00
RMM: 7 -0.798402657216E+02 0.65812E+00 -0.86803E-01 5718 0.148E+00 0.393E+00
RMM: 8 -0.798119831835E+02 0.28283E-01 -0.53408E-01 5837 0.902E-01 0.218E+00
RMM: 9 -0.798453068596E+02 -0.33324E-01 -0.34620E-01 6161 0.431E-01 0.119E+00
RMM: 10 -0.799799243994E+02 -0.13462E+00 -0.31182E-01 6252 0.291E-01 0.791E-01
RMM: 11 -0.800443414466E+02 -0.64417E-01 -0.27461E-01 6269 0.213E-01 0.548E-01
RMM: 12 -0.800901476136E+02 -0.45806E-01 -0.25459E-01 6765 0.151E-01 0.439E-01
RMM: 13 -0.801415670167E+02 -0.51419E-01 -0.23936E-01 6805 0.145E-01 0.390E-01
.......
RMM: 45 -0.805100714634E+02 -0.45644E-02 -0.27155E-01 10407 0.259E-02 0.116E-03
RMM: 46 -0.805050760096E+02 0.49955E-02 -0.22148E-01 10419 0.259E-02 0.135E-03
RMM: 47 -0.805103486267E+02 -0.52726E-02 -0.27422E-01 10408 0.259E-02 0.115E-03
RMM: 48 -0.805049892395E+02 0.53594E-02 -0.22255E-01 10426 0.259E-02 0.131E-03
RMM: 49 -0.805102983943E+02 -0.53092E-02 -0.27313E-01 10416 0.259E-02 0.114E-03
RMM: 50 -0.805046719252E+02 0.56265E-02 -0.21892E-01 10426 0.259E-02 0.127E-03
RMM: 51 -0.805106613614E+02 -0.59894E-02 -0.27732E-01 10421 0.259E-02 0.114E-03
RMM: 52 -0.805045703438E+02 0.60910E-02 -0.21750E-01 10434 0.259E-02 0.124E-03
change of polarisation eV/A/(eV/A) component 2 : 6.634 159.109 -0.288
dielectric tensor component 2 : 0.280 7.705 -0.012
Linear response to external field, progress :
Direction: 3
N E dE d eps ncg rms rms(c)
RMM: 1 -0.160069062237E+02 -0.16007E+02 -0.77091E+00 5502 0.203E+01
RMM: 2 -0.366082663562E+02 -0.20601E+02 0.45920E+01 4656 0.783E+00 0.301E+01
RMM: 3 -0.292114466448E+02 0.73968E+01 -0.36756E+01 5735 0.733E+00 0.289E+01
RMM: 4 -0.179472887827E+02 0.11264E+02 -0.13737E+00 5383 0.407E+00 0.118E+01
RMM: 5 -0.168662547537E+02 0.10810E+01 0.16555E+00 5730 0.239E+00 0.687E+00
RMM: 6 -0.151716549515E+02 0.16946E+01 0.43200E+00 5998 0.198E+00 0.476E+00
RMM: 7 -0.145512506637E+02 0.62040E+00 0.69433E+00 6010 0.136E+00 0.335E+00
RMM: 8 -0.143115153670E+02 0.23974E+00 0.63621E+00 6554 0.884E-01 0.212E+00
RMM: 9 -0.140914874515E+02 0.22003E+00 0.82364E+00 7161 0.478E-01 0.136E+00
.........
RMM: 19 -0.139778535466E+02 0.14448E+00 0.82557E+00 9783 0.141E-01 0.336E-02
RMM: 20 -0.141297371793E+02 -0.15188E+00 0.67410E+00 10040 0.138E-01 0.257E-02
RMM: 21 -0.139905958056E+02 0.13914E+00 0.82179E+00 10050 0.137E-01 0.219E-02
RMM: 22 -0.141333447537E+02 -0.14275E+00 0.67892E+00 10172 0.136E-01 0.179E-02
RMM: 23 -0.139985388622E+02 0.13481E+00 0.81611E+00 10172 0.136E-01 0.166E-02
RMM: 24 -0.141332183484E+02 -0.13468E+00 0.68360E+00 10389 0.136E-01 0.145E-02
RMM: 25 -0.139985199466E+02 0.13470E+00 0.81468E+00 10340 0.136E-01 0.144E-02
RMM: 26 -0.141320217772E+02 -0.13350E+00 0.68157E+00 10297 0.136E-01 0.131E-02
RMM: 27 -0.139988908248E+02 0.13313E+00 0.82299E+00 10426 0.136E-01 0.134E-02
change of polarisation eV/A/(eV/A) component 3 : -0.720 -0.279 25.302
dielectric tensor component 3 : -0.030 -0.012 2.066

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Tue Apr 09, 2024 5:19 am
by marie-therese.huebsch
Dear lisg_sari,

Thank you for bringing up this issue!
I am happy to investigate what could be done to improve the convergence. Could you please provide a minimal working example and add the corresponding input and main output files so we can reproduce the issue? If you have the input and output for the linear response without local-field effects, please also add those.

Cheers,
Marie-Therese

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Tue Apr 09, 2024 5:01 pm
by lisg_sari
Dear Marie-Therese,

Thanks a lot for your help. Please check out the attached for a trial.
Actually I was trying to calculate Raman Intensities which requires dielectric tensor for each vibrational mode. Also WAVECAR was read in each step.
The attached single job converged to the same state and produced the same errors, although the tensor values are slightly different from my previous calculations.
Sorry about that the system is slightly big. If you have specific suggestions, I also can start to try.

Thanks again.

Best,

SG

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Wed Apr 10, 2024 1:11 pm
by marie-therese.huebsch
Hi Sari,

Thanks for providing the example.
I notice that the vacuum is quite small between the slabs, and you are not using a dipole correction. In terms of convergence, it would probably help to increase the vacuum, switch on LDIPOL and other tags related to the dipole correction.

I don't have experience computing Raman intensities, so I wonder how would you do it for a slab? It seems the convergence issues occur perpendicular to the surface. It is probably worth checking the assumptions.

Let me know what you think!
Marie-Therese

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Wed Apr 10, 2024 1:55 pm
by lisg_sari
Dear Marie-Therese,

Thanks a lot for your suggestions.
Actually I tried larger distance for vacuum as well as turning on LDIPOL = 3;
However, no improvement was reached. I was thinking the dipole corrections may be helpful, however, no success so far.
Any specific suggestions for LDIPOL settings?

For Raman intensities, I followed the script shown here:
https://github.com/raman-sc/VASP

Previous post: https://www.vasp.at/forum/viewtopic.php?t=17588
mentioned charge sloshing issue, could you please provide suggestions if you have ?

Thanks a lot.

Best,

SG

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Wed Apr 10, 2024 9:33 pm
by lisg_sari
by the way, With "IDIPOL=3 LDIPOL=.TRUE. " added in the input file, I did see the improvement.
Is there something one can do further help converge ? Thanks a lot.

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Sun Apr 14, 2024 2:22 pm
by lisg_sari
by the way, can I ask how the parameters for "DIPOL = 0.5 0.5 0.5" is set up?
I found this setting is sensitive to the calculations; I found the using geometry centers and mass centers make huge difference.
Thanks.

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Mon Apr 15, 2024 2:36 pm
by michael_wolloch
Hi lisg_sari,

great that your system is converging now with the dipole corrections turned on! DIPOL should be set to the center of mass for your slab. If your calculation is very sensitive to the exact position of DIPOL, your vacuum region is likely too small, and the charge density between the replicas of your slab does not go entirely to zero. Since you only have around 10 Angstrom of vacuum in your system, this seems plausible to me. It should also help with convergence to increase the vacuum region, as @marie-therese.huebsch already suggested.

Let me know if you have further questions,
Michael Wolloch

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Tue Apr 16, 2024 4:33 am
by lisg_sari
Dear Michael ,

Thanks a lot for your suggestions.
I increased the vacuum distance to 40 Ang and turned on dipole corrections with DIPOL = 0.5 0.5 0.492 (the mass center in the Z axis ).
Initial calculations with "AMIX=0.4 BMIX=1" still did not converge in Z direction (positive dE);
I played a little bit by changing AMix and BMIX, the convergence improved actually, converging and then diverged.
Please check out the attached and the calculations are shown in the vasp_raman.out file.

Feel like this job can be converged in some way; I Just has little knowledge with respect to the setting of AMIX and BMIX for this system.
Could you please provide some further suggestions ?

Really appreciate your help.

Best,

SG

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Tue Apr 16, 2024 4:15 pm
by michael_wolloch
Dear SG,

I apologize, I misunderstood this message:
by the way, With "IDIPOL=3 LDIPOL=.TRUE. " added in the input file, I did see the improvement.
and thought you already had converged the calculation.

So you already tried slow linear mixing as it is detailed on the wiki for magnetic calculations, but maybe AMIX=0.01 is a bit too small. I would have set it to 0.2 or 0.1 initially. Also, do not forget to set BMIX_MAG = 0.0001 as well.

We did change the linear response routines a bit a couple of months ago, so it would be a good idea to recompile the newest version, 6.4.3, and try again. Look at the changelog for some details.

I would also advise you to change your input according to the warnings you get. That means:

1) Change the nr. of MPI-ranks to the nr. of GPUs you are using (just for performance reasons, this will not change the convergence)

2) Set your MAGMOM flags.

3) Set LREAL = .FALSE.

This is a complex system and a complex calculation. It might also help to trim all things not strictly necessary and try again with a minimal setup. E.g. ISPIN = 1, no GGA+U, no dipole corrections, no ADDGRID. That might not give you the correct result, but it might converge, and when you then slowly add more tags, you will see what causes problems.

I am still thinking about this and consulting with some colleagues, I will get back to you soon with more information, but I thought these steps are already worth pursuing and might already get you a bit closer to a solution.

All the best, Michael

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Tue Apr 16, 2024 7:06 pm
by lisg_sari
Dear Michael, Thank you so much for your suggestions. It is a challenging system for sure. -SG

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Fri Apr 19, 2024 10:23 am
by michael_wolloch
Dear SG,

I had some discussions and did some tests (although not on your system directly due to size) and can report some insights:

1) dipole corrections do work correctly with the density functional perturbation theory routines used with LEPSILON = .TRUE.

2) Mixing parameters enter the RMM minimization of the linear Sternheimer equations in the same way as the "normal" electronic minimization routines since the same routines are called internally. So you can use those mixing flags exactly as you would use to optimize the SCF cycle.

3) LDA+U also works as expected with LEPSILON = .TRUE. I was able to reproduce the experimental dielectric constant for NiO when using a large U.

4) You have some unoccupied 4f states in your POTCAR reference configurations. This is not surprising. However, those states my play a role in your DFPT calculations, and it could be beneficial to set LMAXMIX = 6, to allow the mixer to work with those states as well.

Please let me know what you have tried and if you can converge your calculations.

All the best, Michael

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Sun Apr 21, 2024 6:39 am
by lisg_sari
Dear Michael,

Thanks a lot for your help for testing a system with transition metal oxide system.
I am trying with LMAXMIX = 6, combined with dipole corrections etc.

(1) My main issue for my surface slab calculations occurs on the third direction ( z-axis) and the other two look fine, although dE = E-1 or E-2, not ideal E-4 or less.
dE on the Z-axi sometimes are as large as 0.2 eV; I attached an example with a bad convergence on the z-axis. [I also tested a slab with adsorbents on both sides and dipole moments are near to zero on Z-axis with 40 ang vacuum, still challenging to converge]

(2) I am calculating dielectric tensors for several different vibrational modes to get Raman Intensity. The linear response converge slightly better for some modes; others are bad.
Not sure why.

(3) Mixing parameters do change the calculations. For example different AMIX setting give different values, especially for the linear response energies on Z-axis.

(4) Have you tried a slab ? Some settings may be critical to converge a slab. Not sure how big the vacuum distance is enough ?

I will update if I find out something new.

Thanks again.

SG

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Mon Apr 22, 2024 9:12 am
by michael_wolloch
Dear SG,

I have tried a slab, but not a polar one so far. I had no issues with convergence.

I am not sure what settings you used for the attached output, but there are still a few warnings that you can get rid of. Please try a minimal setup, as discussed previously first:
It might also help to trim all things not strictly necessary and try again with a minimal setup. E.g. ISPIN = 1, no GGA+U, no dipole corrections, no ADDGRID. That might not give you the correct result, but it might converge, and when you then slowly add more tags, you will see what causes problems.
Then add LMAXMIX = 6 and see if it improves something.

Please organize your input and output files for each test, and provide them if you have further questions. Also, please do not forget to reply if you get it to work and what the solution was, so I can close the thread and other users can profit from your diligent work.

Cheers, Michael

Re: RMM Convergence Issue with LEPSILON=.TRUE.

Posted: Fri Apr 26, 2024 4:13 pm
by lisg_sari
Dear Michael,

With the latest version of 6.4.3, I was able to get the calculations converged with energy convergence criteria of 'EDIFF = 1.0E-6' together with dipole corrections 'IDIPOL=3 LDIPOL=.TRUE. DIPOL = 0.5 0.5 0.513' , and 'LMAXMIX = 6'. I kept the mixing 'AMIX = 0.4 BMIX = 1', by the way. Even though " ISPIN = 1 or no GGA+U" did not help for test in this case.
The linear response results are really nice and thanks a lot for your help.
-----------------------
Linear response to external field, progress :
Direction: 1
N E dE d eps ncg rms rms(c)
RMM: 1 -0.138459142356E+02 -0.13846E+02 0.11414E+00 5588 0.189E+01
RMM: 2 -0.466349827461E+02 -0.32789E+02 0.58226E+01 4475 0.681E+00 0.359E+01
RMM: 3 -0.480609861244E+02 -0.14260E+01 -0.37212E+01 5439 0.778E+00 0.287E+01
RMM: 4 -0.444532375750E+02 0.36077E+01 -0.63499E+00 5006 0.361E+00 0.109E+01
RMM: 5 -0.436398905044E+02 0.81335E+00 -0.18841E+00 5726 0.198E+00 0.665E+00
---
RMM: 49 -0.428069984082E+02 -0.19008E-05 0.19652E-05 3966 0.987E-05 0.370E-05
RMM: 50 -0.428070013613E+02 -0.29531E-05 0.62955E-06 3298 0.784E-05 0.269E-05
RMM: 51 -0.428070019561E+02 -0.59478E-06 0.15181E-05 2885 0.728E-05 0.205E-05
RMM: 52 -0.428070020480E+02 -0.91900E-07 0.18581E-05 2936 0.749E-05 0.172E-05
RMM: 53 -0.428070021429E+02 -0.94899E-07 0.22945E-05 2827 0.730E-05 0.136E-05
RMM: 54 -0.428070046545E+02 -0.25116E-05 -0.14014E-06 2827 0.730E-05 0.109E-05
RMM: 55 -0.428070033288E+02 0.13257E-05 0.14952E-05 2739 0.718E-05 0.839E-06
change of polarisation eV/A/(eV/A) component 1 : 87.159 5.298 -0.038
dielectric tensor component 1 : 4.673 0.223 -0.002
Linear response to external field, progress :
Direction: 2
N E dE d eps ncg rms rms(c)
RMM: 1 -0.172084334715E+02 -0.17208E+02 0.30509E+01 5592 0.212E+01
RMM: 2 -0.646594820158E+02 -0.47451E+02 0.10101E+02 4528 0.844E+00 0.399E+01
RMM: 3 -0.790312241619E+02 -0.14372E+02 -0.25428E+01 5137 0.702E+00 0.266E+01
RMM: 4 -0.781699310464E+02 0.86129E+00 -0.59454E+00 5118 0.390E+00 0.181E+01
RMM: 5 -0.782370474485E+02 -0.67116E-01 -0.22313E+00 5421 0.270E+00 0.957E+00
RMM: 6 -0.782012817276E+02 0.35766E-01 -0.11870E+00 5830 0.200E+00 0.597E+00
RMM: 7 -0.780177098167E+02 0.18357E+00 -0.52907E-01 5652 0.154E+00 0.419E+00
RMM: 8 -0.779227288375E+02 0.94981E-01 -0.20422E-01 5743 0.882E-01 0.206E+00
RMM: 9 -0.780144790872E+02 -0.91750E-01 -0.52954E-02 5876 0.445E-01 0.137E+00
RMM: 10 -0.781921302744E+02 -0.17765E+00 -0.31042E-02 5828 0.322E-01 0.880E-01
----
RMM: 26 -0.789289285115E+02 -0.38512E-02 -0.77636E-05 6654 0.217E-02 0.185E-02
RMM: 27 -0.789302116693E+02 -0.12832E-02 -0.89376E-06 6485 0.135E-02 0.132E-02
RMM: 28 -0.789303845398E+02 -0.17287E-03 -0.34451E-06 6515 0.839E-03 0.103E-02
RMM: 29 -0.789307817596E+02 -0.39722E-03 -0.25204E-05 6577 0.557E-03 0.836E-03
RMM: 30 -0.789315332135E+02 -0.75145E-03 -0.22769E-05 6678 0.570E-03 0.688E-03
RMM: 31 -0.789318223840E+02 -0.28917E-03 -0.30380E-05 6698 0.414E-03 0.562E-03
RMM: 32 -0.789318223185E+02 0.65446E-07 0.68601E-06 6615 0.350E-03 0.450E-03
change of polarisation eV/A/(eV/A) component 2 : 5.215 154.656 -0.202
dielectric tensor component 2 : 0.220 7.517 -0.009
Linear response to external field, progress :
Direction: 3
N E dE d eps ncg rms rms(c)
RMM: 1 -0.156416123407E+02 -0.15642E+02 -0.15682E+00 5478 0.204E+01
RMM: 2 -0.405879699007E+02 -0.24946E+02 0.52820E+01 4515 0.780E+00 0.386E+01
RMM: 3 -0.240238017746E+02 0.16564E+02 -0.50340E+01 5641 0.805E+00 0.366E+01
RMM: 4 -0.158059815539E+02 0.82178E+01 -0.15721E+01 5181 0.459E+00 0.119E+01
RMM: 5 -0.132417645572E+02 0.25642E+01 -0.56159E+00 5792 0.248E+00 0.624E+00
RMM: 6 -0.114077012627E+02 0.18341E+01 -0.15864E+00 5830 0.200E+00 0.582E+00
----
RMM: 44 -0.102527548465E+02 -0.11750E-05 0.56702E-06 3907 0.103E-04 0.795E-05
RMM: 45 -0.102527538447E+02 0.10018E-05 0.68163E-06 4010 0.106E-04 0.608E-05
RMM: 46 -0.102527526655E+02 0.11791E-05 0.21093E-05 3171 0.836E-05 0.447E-05
RMM: 47 -0.102527506832E+02 0.19824E-05 0.37283E-05 3219 0.856E-05 0.334E-05
RMM: 48 -0.102527548203E+02 -0.41371E-05 -0.43976E-07 2954 0.829E-05 0.257E-05
RMM: 49 -0.102527547133E+02 0.10703E-06 0.22774E-06 2719 0.775E-05 0.171E-05
change of polarisation eV/A/(eV/A) component 3 : -0.137 -0.218 21.577
dielectric tensor component 3 : -0.006 -0.009 1.909
--------------------------------------