RIM

[sunxl]

Dear all，

When I use the RIM method, the following parameters appear in the input file.

RandQpts=1000000 # [RIM] Number of random q-points in the BZ
RandGvec= 100 RL # [RIM] Coulomb interaction RS components
CUTGeo= "slab z" # [CUT] Coulomb Cutoff geometry: box/cylinder/sphere/ws/slab X/Y/Z/XY..
% CUTBox
0.000000 | 0.000000 | 0.000000 | # [CUT] [au] Box sides
%

How do I know if the above Settings are reasonable? Is it determined by whether the RIM in the output file is equal to the Real RIM?
What are the criteria to ensure convergence？

[04.02] RIM integrals
=====================

Gamma point sphere radius : 0.014163 [a.u.]
Points outside the sphere : 800143
[Int_sBZ(q=0) 1/q^2]*(Vol_sBZ)^(-1/3) =: 7.084143
should be <: 7.795600
[WR./all_Bz//ndb.RIM]-----------------------------------------------------------
Brillouin Zone Q/K grids (IBZ/BZ) : 164 324 164 324
Coulomb cutoff potential : none
Coulombian RL components : 101
Coulombian diagonal components : yes
RIM random points : 1000000
RIM RL volume [a.u.] : 0.128822
Real RL volume [a.u.] : 0.128530
Eps^-1 reference component : 0
Eps^-1 components : 0.000000 0.000000 0.000000
RIM anysotropy factor : 0.000000
- S/N 009635 ---------------------------------------------- v.05.02.01 r.22792 -

Summary of Coulomb integrals for non-metallic bands |Q|[au] RIM/Bare

Q [1] 0.100000E-4 0.908736 * Q [11] 0.060319 0.841699
Q [2] 0.060319 0.841971 * Q [28] 0.060319 0.841847
Q [27] 0.104475 0.960277 * Q [12] 0.104475 0.960483
Q [46] 0.104475 0.960364 * Q [29] 0.120638 0.979271
Q [45] 0.120638 0.979216 * Q [3] 0.120638 0.979230
Q [26] 0.159589 1.000904 * Q [63] 0.159589 1.001072
Q [13] 0.159589 1.001111 * Q [64] 0.159589 1.001093
Q [44] 0.159589 1.000998 * Q [30] 0.159589 1.001128
Q [4] 0.180956 1.005550 * Q [47] 0.180956 1.005623
Q [62] 0.180956 1.005556 * Q [43] 0.208950 1.007883
Q [31] 0.208950 1.008024 * Q [81] 0.208950 1.008004
Q [48] 0.217483 1.008311 * Q [61] 0.217483 1.008214
Q [82] 0.217483 1.008299 * Q [14] 0.217483 1.008275
Q [25] 0.217483 1.008107 * Q [80] 0.217483 1.008267
Q [5] 0.241275 1.008425 * Q [65] 0.241275 1.008496
Q [79] 0.241275 1.008436 * Q [60] 0.262924 1.008129
Q [49] 0.262924 1.008234 * Q [32] 0.262924 1.008219
Q [42] 0.262924 1.008092 * Q [99] 0.262924 1.008225
Q [98] 0.262924 1.008213 * Q [66] 0.276415 1.007979
Q [78] 0.276415 1.007903 * Q [15] 0.276416 1.007937
Q [24] 0.276416 1.007801 * Q [97] 0.276416 1.007940
Q [100] 0.276416 1.007975 * Q [6] 0.301594 1.007357
Q [96] 0.301594 1.007369 * Q [83] 0.301594 1.007422
Q [116] 0.313426 1.007139 * Q [59] 0.313426 1.007043
Q [50] 0.313426 1.007142 * Q [117] 0.319177 1.007019
Q [115] 0.319177 1.007000 * Q [41] 0.319177 1.006888
Q [33] 0.319177 1.007000 * Q [77] 0.319177 1.006940
Q [67] 0.319177 1.007022 * Q [23] 0.335841 1.006507
Q [118] 0.335841 1.006663 * Q [114] 0.335841 1.006629
Q [16] 0.335841 1.006622 * Q [84] 0.335841 1.006664
Q [95] 0.335841 1.006601 * Q [7] 0.361913 1.006089
Q [101] 0.361913 1.006148 * Q [113] 0.361913 1.006102
Q [76] 0.366905 1.005967 * Q [68] 0.366905 1.006048
Q [58] 0.366905 1.005947 * Q [51] 0.366905 1.006039
Q [134] 0.366905 1.006047 * Q [133] 0.366905 1.006040
Q [34] 0.376691 1.005855 * Q [40] 0.376691 1.005757
Q [135] 0.376691 1.005880 * Q [132] 0.376691 1.005859
Q [94] 0.376691 1.005814 * Q [85] 0.376691 1.005881
Q [112] 0.395537 1.005524 * Q [102] 0.395537 1.005578
Q [136] 0.395537 1.005577 * Q [131] 0.395537 1.005545
Q [22] 0.395537 1.005438 * Q [17] 0.395537 1.005536
Q [75] 0.417901 1.005168 * Q [69] 0.417901 1.005244
Q [151] 0.417901 1.005246 * Q [57] 0.422232 1.005096
Q [119] 0.422232 1.005200 * Q [52] 0.422232 1.005179
Q [93] 0.422232 1.005126 * Q [8] 0.422232 1.005147
Q [152] 0.422232 1.005194 * Q [86] 0.422232 1.005193
Q [150] 0.422232 1.005183 * Q [130] 0.422232 1.005159
Q [111] 0.434965 1.004980 * Q [103] 0.434965 1.005037
Q [153] 0.434965 1.005037 * Q [149] 0.434965 1.005016
Q [39] 0.434965 1.004923 * Q [35] 0.434965 1.005009
Q [129] 0.455397 1.004760 * Q [120] 0.455397 1.004806
Q [154] 0.455397 1.004807 * Q [21] 0.455397 1.004682
Q [148] 0.455397 1.004776 * Q [18] 0.455397 1.004767
Q [159] 0.471105 1.004446 * Q [160] 0.471105 1.004452
Q [87] 0.471105 1.004637 * Q [74] 0.471105 1.004560
Q [70] 0.471105 1.004631 * Q [92] 0.471105 1.004572
Q [110] 0.478766 1.004511 * Q [104] 0.478766 1.004569
Q [161] 0.478766 1.004389 * Q [158] 0.478766 1.004373
Q [56] 0.478766 1.004476 * Q [53] 0.478766 1.004551
Q [147] 0.482550 1.004503 * Q [9] 0.482550 1.004491
Q [137] 0.482550 1.004539 * Q [162] 0.493731 1.004270
Q [157] 0.493731 1.004244 * Q [36] 0.493731 1.004410
Q [121] 0.493731 1.004439 * Q [128] 0.493731 1.004389
Q [38] 0.493731 1.004334 * Q [20] 0.515364 1.004154
Q [19] 0.515364 1.004230 * Q [163] 0.515364 1.004112
Q [156] 0.515364 1.004079 * Q [146] 0.515364 1.004226
Q [138] 0.515364 1.004267 * Q [88] 0.522376 1.004204
Q [142] 0.522376 1.004037 * Q [91] 0.522376 1.004142
Q [109] 0.525847 1.004127 * Q [105] 0.525847 1.004183
Q [143] 0.525847 1.004018 * Q [71] 0.525847 1.004174
Q [141] 0.525847 1.004009 * Q [73] 0.525847 1.004107
Q [55] 0.536125 1.004027 * Q [54] 0.536125 1.004096
Q [122] 0.536125 1.004116 * Q [127] 0.536125 1.004066
Q [144] 0.536125 1.003957 * Q [140] 0.536125 1.003940
Q [155] 0.542869 1.003897 * Q [164] 0.542869 1.004042
Q [10] 0.542869 1.003958 * Q [139] 0.552831 1.004010
Q [37] 0.552831 1.003983 * Q [145] 0.552831 1.003967
Q [106] 0.575405 1.003872 * Q [89] 0.575405 1.003868
Q [125] 0.575405 1.003721 * Q [124] 0.575405 1.003718
Q [90] 0.575405 1.003809 * Q [108] 0.575405 1.003818
Q [123] 0.581694 1.003842 * Q [126] 0.581694 1.003792
Q [72] 0.581694 1.003829 * Q [107] 0.626851 1.003623
Non-periodic chartesian directions : none
Optical renormalization : 49.90624 [au]
Polarizability dimension : length

Best，
sunxl

[Daniele Varsano]

Dear Sunxl,

note that accoridin to your input RIM is applied only to the bare potential and not the screened potential.
If you want to use the W_rim algorithms as in Guanadlini et al (https://www.nature.com/articles/s41524-023-00989-7) you should activate it e.g.:

Code: Select all

yambo -rw -gw0 p

In this way, the rim_w runlevel is activated and the RandGvecW variable will be present in the input.

About the values of the variable, you can safely add more random q points (e.g. 3 millions). This is nearly costless and will improve the stochastic evaluation of the integration. About the RandGvec and RandGvecW, usually few tenths are enough, the safer way to check if they are converged is to look at the quaisparticle energies.

Best,

Daniele

[sunxl]

Dear Daniele，

Thank you very much for your prompt reply.

1、I don't understand. Can you explain it in detail? Are you saying that it is wrong to use -r to add cut off when calculating gw band? Should I use -rw instead of -r? When -rw is used, if the random q point is 3 million, what is the appropriate amount of RandGvec and RandGvecW? Is the way to test convergence to see the change in gw gap after gw band is calculated?

2、After calculating gw band, I also need to calculate GW-BSE first order spectrum and GW-SHG-BSE. In the tutorial, I saw that the input file was generated using -r, does it need to be changed to -rw?

Thank you in advance for your help!

Best，
sunxl

[Daniele Varsano]

Dear sunxl,

Are you saying that it is wrong to use -r to add cut off when calculating gw band?

No, I was just wondering if you wanted to use the integration technique for W described in the paper I indicated to you. If not, it is correct what you did.

When -rw is used, if the random q point is 3 million, what is the appropriate amount of RandGvec and RandGvecW

RandGvec and Random q points are independent, so you can put a large number of random q point to be sure that integrals are calculated in accurate way. Then, the Gvec should be checked looking at the GW gap. In any case, a few tenths are usually enough.

2) As before, it depends if you want to evaluate the W integral stocastichally, otherwise the -r is snough.

Best,

Daniele