This thread is locked.Only browsing is available.
 Top Page > Browsing
 One more NEGF issue Date: 2014/07/15 23:10 Name: Artem   Dear Taisuke,Please take a look at the following (selected lines of) output of the first example fromhttp://www.openmx-square.org/openmx_man3.7/node107.htmlThe number of threads in each node for OpenMP parallelization is 4.************************************************************************************************************** Welcome to OpenMX Ver. 3.7.8 Copyright (C), 2002-2013, T. Ozaki OpenMX comes with ABSOLUTELY NO WARRANTY. This is free software, and you are welcome to redistribute it under the constitution of the GNU-GPL.**************************************************************************************************************...******************* MD= 1 SCF=35 ******************* Solving Poisson's equation... Hamiltonian matrix for VNA+dVH+Vxc... Solving the eigenvalue problem... KGrids2: 0.00000 KGrids3: 0.00000 time=2.423550 1 C MulP 2.0000 2.0000 sum 4.0000 2 C MulP 2.0002 2.0002 sum 4.0004 3 C MulP 2.0030 2.0030 sum 4.0060 4 C MulP 1.9970 1.9970 sum 3.9940 5 C MulP 2.0000 2.0000 sum 4.0001 6 C MulP 2.0000 2.0000 sum 4.0001 7 C MulP 2.0000 2.0000 sum 4.0000 8 C MulP 2.0000 2.0000 sum 4.0000 9 C MulP 2.0000 2.0000 sum 4.0000 10 C MulP 2.0000 2.0000 sum 4.0000 11 C MulP 2.0000 2.0000 sum 4.0000 12 C MulP 2.0000 2.0000 sum 4.0000 13 C MulP 2.0000 2.0000 sum 4.0000 14 C MulP 2.0000 2.0000 sum 4.0000 15 C MulP 2.0000 2.0000 sum 4.0000 16 C MulP 2.0000 2.0000 sum 4.0000 17 C MulP 2.0000 2.0000 sum 4.0000 18 C MulP 2.0000 2.0000 sum 4.0000 19 C MulP 2.0000 2.0000 sum 4.0001 20 C MulP 2.0000 2.0000 sum 4.0001 Sum of MulP: up = 48.00057 down = 48.00057 total= 96.00113 ideal(neutral)= 96.00000 Total Spin Moment (muB) = 0.000000000000 Mixing_weight= 0.020000000000 Uele = 0.000000000000 dUele = 1.000000000000 NormRD = 0.000000066231 Criterion = 0.000000100000...The system looks ideal to me (please correct me if I am wrong). The transmission calculation yields values close to integers: 0.987 for example at E=-10eV.This error is definitely large. It can be due to smearing 0.005 eV. But I think that there is also an error due to a wrong (not-an-ideal in this case) charge density calculated. I am not an expert in NEGF but I can clearly see that there is something wrong with connections of the leads to the scattering region. The Mulliken populations of 3rd and 4th atoms differ by some small amount.In my project I encountered this problem for non-transparent structures with MoS2 as leads. For relatively small scattering region I have 1 Mo MulP 6.95 6.95 sum 13.89 diff 0.00 (180.00 -5.10) Ml 0.00 (180.00 -4.19) Ml+s 0.00 (180.00 -5.10) 2 S MulP 3.02 3.02 sum 6.04 diff 0.00 ( 80.27 -0.03) Ml 0.00 (172.72 180.01) Ml+s 0.00 ( 84.08 -0.03) 3 S MulP 3.02 3.02 sum 6.04 diff 0.00 ( 80.27 179.97) Ml 0.00 (172.72 0.01) Ml+s 0.00 ( 84.08 179.97) 4 Mo MulP 6.77 6.77 sum 13.54 diff 0.00 (180.00 -76.60) Ml 0.00 ( 0.00 258.39) Ml+s 0.00 ( 0.00 -77.04) 5 S MulP 2.85 2.85 sum 5.70 diff 0.00 (108.07 179.76) Ml 0.00 ( 70.37 180.00) Ml+s 0.00 (104.46 179.78) 6 S MulP 2.85 2.85 sum 5.70 diff 0.00 (108.07 -0.24) Ml 0.00 ( 70.37 0.00) Ml+s 0.00 (104.46 -0.22)---- Left-to-center connection ---- 7 Mo MulP 7.33 7.33 sum 14.66 diff 0.00 ( 0.00 26.09) Ml 0.00 ( 0.00 69.00) Ml+s 0.00 ( 0.00 27.24) 8 S MulP 3.09 3.09 sum 6.18 diff 0.00 ( 88.43 -0.06) Ml 0.00 (140.78 180.02) Ml+s 0.00 ( 88.53 -0.06) 9 S MulP 3.09 3.09 sum 6.18 diff 0.00 ( 88.43 179.94) Ml 0.00 (140.78 0.02) Ml+s 0.00 ( 88.53 179.94) 10 Mo MulP 6.97 6.97 sum 13.93 diff 0.00 ( 0.01 252.17) Ml 0.00 (180.00 184.19) Ml+s 0.00 ( 0.01 251.83) 11 S MulP 3.02 3.02 sum 6.04 diff 0.00 ( 89.94 179.90) Ml 0.00 (155.13 179.94) Ml+s 0.00 ( 89.94 179.90) 12 S MulP 3.02 3.02 sum 6.04 diff 0.00 ( 89.94 -0.10) Ml 0.00 (155.13 -0.05) Ml+s 0.00 ( 89.94 -0.10) 13 Mo MulP 6.95 6.95 sum 13.91 diff 0.00 ( 0.23 23.15) Ml 0.00 (180.00 185.71) Ml+s 0.00 (179.77 23.54) 14 S MulP 3.03 3.03 sum 6.06 diff 0.00 ( 90.00 180.01) Ml 0.00 ( 84.26 180.00) Ml+s 0.00 ( 90.00 180.01) 15 S MulP 3.03 3.03 sum 6.06 diff 0.00 ( 90.00 0.01) Ml 0.00 ( 84.25 0.00) Ml+s 0.00 ( 90.00 0.01) 16 Mo MulP 6.93 6.93 sum 13.87 diff 0.00 ( 4.67 21.68) Ml 0.00 (180.00 181.79) Ml+s 0.00 (179.88 22.23) 17 S MulP 3.04 3.04 sum 6.07 diff 0.00 ( 89.35 154.13) Ml 0.00 ( 90.80 -0.00) Ml+s 0.00 ( 89.39 149.13) 18 S MulP 3.04 3.04 sum 6.07 diff 0.00 ( 89.35 -25.98) Ml 0.00 ( 90.80 180.00) Ml+s 0.00 ( 89.39 -31.00)Specifically Mo have 14.66 and 13.54 instead of ideal 13.9 electrons. Note that the rest of Mo populations are quite close to ideal ones.I increased the scattering region but the problem did not go away: 1 Mo MulP 6.95 6.95 sum 13.90 diff 0.00 (180.00 -5.85) Ml 0.00 (180.00 -6.60) Ml+s 0.00 (180.00 -5.85) 2 S MulP 3.02 3.02 sum 6.04 diff 0.00 ( 80.25 -0.03) Ml 0.00 (172.63 180.00) Ml+s 0.00 ( 84.10 -0.03) 3 S MulP 3.02 3.02 sum 6.04 diff 0.00 ( 80.25 179.97) Ml 0.00 (172.63 0.00) Ml+s 0.00 ( 84.10 179.97) 4 Mo MulP 6.81 6.81 sum 13.63 diff 0.00 (180.00 -85.44) Ml 0.00 ( 0.00 267.98) Ml+s 0.00 ( 0.01 -85.58) 5 S MulP 2.89 2.89 sum 5.79 diff 0.00 (140.56 179.17) Ml 0.00 ( 69.85 180.00) Ml+s 0.00 (127.29 179.41) 6 S MulP 2.89 2.89 sum 5.79 diff 0.00 (140.57 -0.83) Ml 0.00 ( 69.85 0.00) Ml+s 0.00 (127.29 -0.59)---- Left-to-center connection ---- 7 Mo MulP 7.24 7.24 sum 14.48 diff 0.00 ( 0.00 54.55) Ml 0.00 ( 0.00 55.00) Ml+s 0.00 ( 0.00 54.55) 8 S MulP 3.08 3.08 sum 6.15 diff 0.00 ( 50.49 -1.86) Ml 0.00 (146.82 180.02) Ml+s 0.00 ( 51.61 -1.95) 9 S MulP 3.08 3.08 sum 6.15 diff 0.00 ( 50.50 178.14) Ml 0.00 (146.82 0.02) Ml+s 0.00 ( 51.61 178.06) 10 Mo MulP 6.95 6.95 sum 13.91 diff 0.00 ( 0.00 -52.82) Ml 0.00 (180.00 196.86) Ml+s 0.00 ( 0.00 -53.48) 11 S MulP 3.03 3.03 sum 6.05 diff 0.00 ( 89.87 179.84) Ml 0.00 (160.15 179.95) Ml+s 0.00 ( 89.88 179.84) 12 S MulP 3.03 3.03 sum 6.05 diff 0.00 ( 89.87 -0.16) Ml 0.00 (160.15 -0.05) Ml+s 0.00 ( 89.88 -0.16) 13 Mo MulP 6.95 6.95 sum 13.91 diff 0.00 ( 0.07 -13.48) Ml 0.00 (180.00 266.24) Ml+s 0.00 ( 0.08 -13.50) 14 S MulP 3.02 3.02 sum 6.04 diff 0.00 ( 89.98 -0.14) Ml 0.00 (103.73 -0.01) Ml+s 0.00 ( 89.98 -0.14) 15 S MulP 3.02 3.02 sum 6.04 diff 0.00 ( 89.98 179.86) Ml 0.00 (103.72 180.00) Ml+s 0.00 ( 89.98 179.86) 16 Mo MulP 6.97 6.97 sum 13.93 diff 0.00 ( 0.97 254.39) Ml 0.00 (180.00 1.08) Ml+s 0.00 ( 3.79 254.41) 17 S MulP 3.02 3.02 sum 6.03 diff 0.00 ( 90.00 179.89) Ml 0.00 ( 68.85 0.02) Ml+s 0.00 ( 90.00 179.89) 18 S MulP 3.02 3.02 sum 6.03 diff 0.00 ( 90.00 -0.11) Ml 0.00 ( 68.85 180.02) Ml+s 0.00 ( 90.00 -0.11)My scattering region contains 22 + 6 + 6 = 34 atoms (3 nm along transport direction) in the first case and 34 + 6 + 6 = 46 (4 nm along transport direction) in the second case. The atoms are sorted along the transport direction so that the first atom is quite far away from the feature. The tragedy here is that transmission differs by 4 times (1.3e-3 vs 5e-3 for example) in these 2 calculations while the feature geometry is exactly the same. Both calculations converged to 1e-9.Any suggestions?Faithfully yours,Artem
 Page: [1]