[Pw_forum] BFGS convergence
PRASENJIT GHOSH
prasenjit at jncasr.ac.in
Sun Aug 27 08:18:03 CEST 2006
Dear all,
I have 2 doubts:
1) specific to my cal.
I'm trying to relax 4layer MgO(100) surface with IML Rh atom & NO on it.
I'm keeping the bottom 2 layers fixed & relaxing rest of the layers along
all directions. I have set forc_conv_thr as 5.0D-4. When the BFGS
optimization is complete the total force printed out is 0.001184 Ry/au.,
which is higher than the threshold.
The force on individual atoms are as given below:
atom 1 type 4 force = 0.00000000 0.00019439 0.00035388
atom 2 type 2 force = 0.00000000 0.00003994 0.00030581
atom 3 type 3 force = 0.00004990 0.00006395 0.00026932
atom 4 type 3 force = -0.00004990 0.00006395 0.00026932
atom 5 type 3 force = 0.00000510 -0.00010716 0.00017645
atom 6 type 3 force = -0.00000510 -0.00010716 0.00017645
atom 7 type 3 force = -0.00015740 0.00021421 0.00025840
atom 8 type 3 force = 0.00015740 0.00021421 0.00025840
atom 9 type 1 force = 0.00000000 0.00017531 0.00012835
atom 10 type 1 force = 0.00000000 0.00012586 0.00009343
atom 11 type 2 force = -0.00003854 -0.00009570 -0.00010196
atom 12 type 2 force = 0.00003854 -0.00009570 -0.00010196
atom 13 type 1 force = 0.00000000 0.00002684 0.00020808
atom 14 type 1 force = 0.00000000 -0.00008463 0.00035348
atom 15 type 2 force = -0.00002782 -0.00001293 0.00001952
atom 16 type 2 force = 0.00002782 -0.00001293 0.00001952
atom 17 type 1 force = 0.00000000 0.00001629 0.00009910
atom 18 type 1 force = 0.00000000 -0.00003038 0.00014986
atom 19 type 2 force = 0.00000763 0.00008195 0.00007784
atom 20 type 2 force = -0.00000763 0.00008195 0.00007784
atom 21 type 2 force = 0.00000000 0.00003195 0.00010070
atom 22 type 2 force = 0.00000000 0.00003854 0.00016380
atom 23 type 1 force = -0.00010405 -0.00006846 0.00012527
atom 24 type 1 force = 0.00010405 -0.00006846 0.00012527
atom 25 type 2 force = 0.00000000 -0.00002769 0.00001703
atom 26 type 2 force = 0.00000000 -0.00007519 0.00008055
atom 27 type 1 force = -0.00004277 0.00010317 -0.00017790
atom 28 type 1 force = 0.00004277 0.00010317 -0.00017790
atom 29 type 2 force = 0.00000000 0.00001178 0.00015564
atom 30 type 2 force = 0.00000000 0.00009848 0.00005032
atom 31 type 1 force = -0.00009405 -0.00007765 0.00004686
atom 32 type 1 force = 0.00009405 -0.00007765 0.00004686
atom 33 type 1 force = 0.00000000 -0.00027438 -0.00258973
atom 34 type 1 force = 0.00000000 0.00085927 -0.00230357
atom 35 type 2 force = -0.00002762 0.00106477 0.00266372
atom 36 type 2 force = 0.00002762 0.00106477 0.00266372
atom 37 type 1 force = 0.00000000 -0.00127835 -0.00262410
atom 38 type 1 force = 0.00000000 -0.00187757 -0.00264360
atom 39 type 2 force = 0.00033189 -0.00088088 0.00323964
atom 40 type 2 force = -0.00033189 -0.00088088 0.00323964
atom 41 type 1 force = 0.00000000 0.00132099 -0.00267980
atom 42 type 1 force = 0.00000000 0.00088146 -0.00307944
atom 43 type 2 force = 0.00043540 -0.00046399 0.00208586
atom 44 type 2 force = -0.00043540 -0.00046399 0.00208586
atom 45 type 2 force = 0.00000000 -0.00001967 -0.01458796
atom 46 type 2 force = 0.00000000 0.00001272 -0.01458917
atom 47 type 1 force = -0.00001307 -0.00005348 0.01390984
atom 48 type 1 force = 0.00001307 -0.00005348 0.01390984
atom 49 type 2 force = 0.00000000 -0.00068276 -0.01433952
atom 50 type 2 force = 0.00000000 -0.00070884 -0.01425615
atom 51 type 1 force = 0.00001464 -0.00004340 0.01348247
atom 52 type 1 force = -0.00001464 -0.00004340 0.01348247
atom 53 type 2 force = 0.00000000 0.00075245 -0.01418097
atom 54 type 2 force = 0.00000000 0.00074367 -0.01430795
atom 55 type 1 force = 0.00003199 0.00014034 0.01388566
atom 56 type 1 force = -0.00003199 0.00014034 0.01388566
Total force = 0.001184 Total SCF correction = 0.000147
Atoms 33-56 are kept fixed. Hence the forces on them is high whereas the
forces on each of the individual components of the relaxed atoms are below
the threshold.
My question is when the code calculates the force (grad) & their
differencec between 2 consecutive BFGS iter. (grad_error) & is it only for
the atoms which are allowed to relax or for the full system?
If its the second case then can anybody suggest where I'm going wrong?
With regards,
Prasenjit.
PRASENJIT GHOSH,
Ph. D STUDENT,
THEORETICAL SCIENCES UNIT,
JAWAHARLAL NEHRU CENTRE,
JAKKUR P. O.,
BANGALORE - 560064,
INDIA.
PHONE:+91-80-22082835 (OFFICE)
+91 9880519401 (MOBILE)
More information about the Pw_forum
mailing list