Dear all,
I'm having divergence problem on DFT calculation on multi-node:
If I launch on single-node, DFT calculation converge for small (56 atoms) and bigger molecules (164 atoms). On the other hand, the same input file on multi-node calculation fail for bigger molecule.
The input file is:
start pchg_trimpolys4
title "Point Charge trimpolys4"
permanent_dir /home/scurcio/qm/prova/tmp
scratch_dir /home/scurcio/qm/scratch
memory stack 2000 heap 1000 global 5000 mb
echo
geometry units angstroms noautoz nocenter
symmetry c1
O -8.80880 3.49361 -12.08015
C -9.40638 2.73080 -13.03864
...............................
H -20.00256 6.88882 -44.71623
H -19.33917 4.54448 -45.16761
end
basis "ao basis"
* library "6-31G*"
end
dft
mult 1
xc b3lyp
iterations 900
convergence energy 1e-6
convergence density 2e-5
sym off
adapt off
noio
direct
end
task dft energy
esp
recalculate
range 0.5
spacing 0.05
radius 1 0.12
radius 6 0.15
radius 8 0.14
radius 16 0.175
probe 0.01
end
task esp
And I obtain this divergence on multiple node launch:
2-e energy = 17248.727228
HOMO = -0.222877
LUMO = -0.222233
Time after variat. SCF: 32.5
Time prior to 1st pass: 32.5
Grid_pts file = /home/scurcio/qm/scratch/pchg_trimpolys4.gridpts.00
Record size in doubles = 12289 No. of grid_pts per rec = 3070
Max. records in memory = 22 Max. recs in file = 726430
Grid integrated density: 697.997079342626
Requested integration accuracy: 0.10E-05
Memory utilization after 1st SCF pass:
Heap Space remaining (MW): 130.68 130679246
Stack Space remaining (MW): 262.11 262108148
convergence iter energy DeltaE RMS-Dens Diis-err time
---------------- ----- ----------------- --------- --------- --------- ------
d= 0,ls=0.0,diis 1 -5217.9290028279 -2.07D+04 2.17D-02 2.01D+02 66.2
Grid integrated density: 698.002276988876
Requested integration accuracy: 0.10E-05
d= 0,ls=0.5,diis 2 -5166.9595547187 5.10D+01 1.09D-01 1.90D+02 96.4
Grid integrated density: 698.031084147238
Requested integration accuracy: 0.10E-05
d= 0,ls=0.5,diis 3 -4036.2259749324 1.13D+03 1.07D+00 3.51D+03 127.0
Grid integrated density: 699.954251616508
Requested integration accuracy: 0.10E-05
d= 0,ls=0.5,diis 4 -4245.4281228506 -2.09D+02 1.11D+00 1.08D+04 158.7
Grid integrated density: 701.539046732465
Requested integration accuracy: 0.10E-05
d= 0,ls=0.5,diis 5 -4277.4683733196 -3.20D+01 2.86D-01 5.69D+03 188.1
Grid integrated density: 714.976146461869
Requested integration accuracy: 0.10E-05
d= 0,ls=0.5,diis 6 -4292.8378592634 -1.54D+01 4.57D-01 1.01D+04 220.0
Grid integrated density: 719.711880549781
Requested integration accuracy: 0.10E-05
d= 0,ls=0.5,diis 7 -4362.6109741423 -6.98D+01 1.33D+00 8.62D+03 251.6
Grid integrated density: 718.761258279434
Requested integration accuracy: 0.10E-05
d= 0,ls=0.5,diis 8 -4379.7764637137 -1.72D+01 1.25D+00 9.70D+04 283.6
Grid integrated density: 719.687780893165
Requested integration accuracy: 0.10E-05
d= 0,ls=0.5,diis 9 -4443.0102377642 -6.32D+01 4.55D-01 1.01D+04 313.5
Grid integrated density: 720.652136329789
Requested integration accuracy: 0.10E-05
d= 0,ls=0.5,diis 10 -4501.0330406169 -5.80D+01 6.10D-01 9.60D+03 344.5
Grid integrated density: 722.463252373448
Requested integration accuracy: 0.10E-05
d= 0,ls=0.5,diis 11 -4578.4446665504 -7.74D+01 1.16D+00 8.11D+03 376.2
Can you help me to solve this problem?
Kind regards,
Frank
|