Phonon frequencies mismatch between EPW and DFPT

[nadeemnatt]

Hi
Phonon frequencies in my calculation have a lot of mismatch between DFPT and EPW. I put iverbosity=1 and from comparison of printed frequencies in epw output I found that EPW is not reading dynamical matrices correctly. I did try using both simple and crystal ASR types but results are same. From matdyn.x frequencies are same as in output of PHONON for the given q. I have tried using lifc=.true. but it gives same results. I am giving phonon frequencies from DFPT and EPW's verbose output for one q point for comparison.

from matdyn.x

-0.486758 -0.632688 -0.296344
-341.7646 -309.2714 -277.0700 -235.8558 -220.5651 -196.3433
-164.4339 -127.3330 -90.9485 -83.2322 -38.2135 -33.2744
43.8301 71.0865 75.4791 86.3332 111.4275 113.1080
130.8742 136.3269 153.7312 157.5570 165.3727 171.4914
182.6157 187.2937 199.8036 201.9878 212.9406 215.6147
226.3429 233.6455 237.6778 248.6412 255.1553 273.4223
278.2931 295.8813 300.1513 304.7217 332.3497 333.3433
339.6165 348.8189 370.8470 376.6381 382.3377 385.5286
400.4463 409.5673 417.8579 442.0152 455.3104 461.1314
461.9972 480.1973 493.9785 495.7011 496.5539 510.4307
522.7998 523.8000 526.6186 529.1809 568.6496 575.0094
580.8300 581.8086 601.6297 611.6363 632.9535 638.4611
642.4900 648.8750 657.7290 658.6110 666.5768 683.7854
683.9998 699.0302 709.2716 712.9863 716.0196 722.0612
722.2860 731.5531 735.9692 736.7075 817.5901 817.8540
828.1595 838.1828 844.2015 850.1441 854.1922 857.5675
859.6078 862.2960 869.8027 880.9200 899.4811 900.3555
909.5267 921.1068 926.2198 932.1964 938.1325 946.7661
951.7155 973.8699 974.0794 979.6136 981.6835 989.4059
997.2703 998.2545 1005.6869 1008.3790 1018.3381 1023.8811
1028.8585 1033.9799 1038.5457 1040.7248 1050.6562 1054.7426
1060.3866 1061.3946 1076.5980 1081.9682 1111.8423 1129.9421
1140.9009 1149.2672 1150.1235 1161.6517 1161.9153 1178.0782
1183.5132 1198.5697 1215.4877 1223.8173 1273.0684 1274.3997
1285.9824 1295.6391 1301.8121 1319.2628 1328.5799 1329.1205
1364.6281 1365.7953 1367.7240 1368.7996 1378.9629 1387.0742
1388.9848 1389.2359 1415.3628 1416.0173 1417.8675 1417.9539
1456.7401 1461.1713 1461.4364 1468.5987 1473.5287 1483.6324
1491.7007 1494.9917 1504.4426 1509.3989 1510.7742 1512.6997
1516.6228 1519.7881 1532.2189 1543.0259 1554.3765 1560.5059
1613.0508 1621.4194 3008.5581 3012.0994 3017.9729 3025.2269
3033.6785 3034.2004 3041.4641 3041.9695 3050.9396 3057.4994
3060.3172 3066.6294 3071.6915 3085.6112 3087.5836 3093.7664
3122.1164 3125.2096 3125.3787 3149.0139 3178.7606 3178.9512
3188.7618 3194.8342 3196.9323 3204.1535 3220.4304 3229.4383


From EPW

===================================================================
irreducible q point # 8
===================================================================

Symmetries of small group of q: 2
in addition sym. q -> -q+G:

Number of q in the star = 1
List of q in the star:
1 -0.486757527 -0.632688436 -0.296344465
Dyn mat calculated from ifcs
Frequencies of the matrix for the current q in the star
-361.25464 -355.57088 -277.65082 -236.64971 -211.97702 -204.54160
-168.60669 -147.75796 -135.54132 -116.94212 -99.62299 -95.24023
-77.52861 -57.65510 35.64194 59.13003 78.73797 94.80298
114.84391 122.78474 126.69274 146.87634 159.22379 171.33843
175.58199 186.90353 198.52850 205.76101 208.70629 227.34263
232.40802 238.33665 246.21294 263.22079 266.32779 276.21854
284.93310 289.17632 302.42671 307.10400 318.82218 328.53856
347.01739 350.64000 354.50547 369.79466 386.92731 398.15100
403.73738 411.40584 414.62306 445.49414 454.27418 468.41810
478.83417 485.26475 493.18354 495.31609 508.01954 515.66384
520.15216 526.52899 535.35667 540.02861 572.12558 573.64297
580.88845 587.37617 604.39163 608.65435 629.22610 638.60273
641.50318 644.78529 651.16714 653.90372 672.74400 683.30352
688.94145 698.62464 709.49372 716.32406 718.26686 723.22691
725.22957 733.42553 738.86330 752.67349 814.84594 821.54836
827.12279 831.72505 842.95354 847.28765 856.63143 858.57572
863.30981 866.22231 867.29147 879.80048 889.17818 896.53126
909.89801 917.18116 923.95652 930.17705 934.98244 944.55442
951.75858 955.45941 964.86050 968.35317 979.02045 988.42855
992.69092 1001.39659 1007.72484 1009.81966 1012.48260 1020.54101
1033.15660 1036.65880 1042.76523 1050.84933 1054.49531 1058.75758
1072.39774 1074.03332 1086.24489 1096.83428 1115.33959 1117.51487
1131.14001 1145.05718 1152.87370 1160.37088 1168.72399 1175.28769
1183.67609 1192.02195 1214.53956 1233.65319 1258.12114 1264.81558
1275.11055 1302.64111 1305.62569 1309.50468 1323.28497 1326.36818
1353.44169 1359.65147 1366.15857 1369.01127 1377.81015 1378.58299
1384.95521 1390.43268 1410.78402 1416.50111 1418.68980 1425.76396
1457.53729 1461.91913 1464.04600 1466.88835 1469.78428 1477.04586
1488.03203 1488.85951 1503.60961 1507.53365 1509.69669 1513.61682
1517.63254 1520.47288 1538.76355 1540.09586 1556.58768 1559.02143
1614.83557 1617.75682 2994.50713 3009.63689 3018.14418 3023.02900
3029.34318 3034.12167 3039.59199 3043.27447 3049.05775 3054.05299
3058.27276 3060.75082 3062.84162 3072.23586 3096.60102 3101.56071
3114.85138 3122.36329 3130.95644 3138.93773 3157.68977 3168.39397
3170.73025 3175.67579 3195.30859 3199.17409 3216.46472 3224.98262


How can I make EPW to read my DFPT results properly?

I have another question about electron-phonon matrix elements for given m,n,q,k on coarse grid Bloch grid. I read that they are given in prefix.epb (binary files) files but I am too naive to read binary files. Can you suggest me some tool to covert these files to .text data files?

Regards
Nadeem

[sponce]

Hello,

To use the crystal sum rule you have to put:

Code: Select all

  lifc        = .true.
  asr_typ     = 'crystal'

Not just lifc.

For you other question, in EPW/src/elphon_shuffle_wrap.f90, you can change the epbwrite from unformatted to formatted. Should be around line 704.

Best,
Samuel

[nadeemnatt]

Thank you Samuel for the prompt reply.
I did use

Code: Select all

asr_typ='crystal'

both in q2r.x and epw but getting same results.
Are Imaginary modes in DFPT results the reason? Actually removing imaginary modes in doped aromatic hydrocarbons (I am working on) is quiet hard. Would you please suggest me some way to overcome this issue?

Regards
Nadeem

[nadeemnatt]

I also did for Al and it works well with very small differences that I guess come from numerical errors.

[sponce]

Hello,

It is possible that your system does have soft phonon (phase instabilities). A way to solve this is to used anharmonic phonons (you need to used specialized code for this).

The theory for electron-phonon is not well defined when you have soft phonon.
One possible option (done by default in EPW) is to simply neglect soft(negative freq) phonons. This is an approximation and might not be a good one.

Best,
Samuel