Convergence of LR-CCSD calculation

From NWChem

Viewed 1821 times, With a total of 2 Posts
Jump to: navigation, search

Just Got Here
Threads 2
Posts 4
Dear All

I am doing linear response CCSD calculation and i want to calculate dynamical polarizabilities at imaginary frequencies for set of molecules for some molecules it works fine but for some it is not converging at lower frequencies.
Below is the input followed by "trimmed error output" for which it is not working
Please let me know possible solution for this problem or if there is any mistake in the input.
-
echo
start HF
title "HF"
memory stack 1000 mb heap 1000 mb global 1000 mb noverify
permanent_dir .
scratch_dir /ptmp/vigo/scratch
charge 0
geometry
symmetry c1
H 0.034818 0.000000 0.000000
F 0.965182 0.000000 0.000000
end
BASIS "ao basis" spherical
F S
74530.0000000 0.0000950 -0.0000220
11170.0000000 0.0007380 -0.0001720
2543.0000000 0.0038580 -0.0008910
721.0000000 0.0159260 -0.0037480
235.9000000 0.0542890 -0.0128620
85.6000000 0.1495130 -0.0380610
33.5500000 0.3082520 -0.0862390
13.9300000 0.3948530 -0.1558650
5.9150000 0.2110310 -0.1109140
F S
1.8430000 1.0000000
F S
0.7124000 1.0000000
F S
0.2637000 1.0000000
F S
0.0859400 1.0000000
F P
80.3900000 0.0063470
18.6300000 0.0442040
5.6940000 0.1685140
F P
1.9530000 1.0000000
F P
0.6702000 1.0000000
F P
0.2166000 1.0000000
F P
0.0656800 1.0000000
F D
5.0140000 1.0000000
F D
1.7250000 1.0000000
F D
0.5860000 1.0000000
F D
0.2070000 1.0000000
F F
3.5620000 1.0000000
F F
1.1480000 1.0000000
F F
0.4600000 1.0000000
F G
2.3760000 1.0000000
F G
0.9240000 1.0000000

H S
82.6400000 0.0020060
12.4100000 0.0153430
2.8240000 0.0755790
H S
0.7977000 1.0000000
H S
0.2581000 1.0000000
H S
0.0898900 1.0000000
H S
0.0236300 1.0000000
H P
2.2920000 1.0000000
H P
0.8380000 1.0000000
H P
0.2920000 1.0000000
H P
0.0848000 1.0000000
H D
2.0620000 1.0000000
H D
0.6620000 1.0000000
H D
0.1900000 1.0000000
H F
1.3970000 1.0000000
H F
0.3600000 1.0000000
END

set int:acc_std 1d-20
scf
rhf
singlet
semidirect memsize 100000000 filesize 1000000
thresh 1e-10
tol2e 1e-16
maxiter 1000
print "final evals"
end
tce
freeze 0
ccsd
thresh 1e-10
maxiter 10000
io ga
2eorb
2emet 13
tilesize 32
attilesize 32
diis 5
diis2 5
diis3 5
lshift 0.1 # CC-T
lshiftl 0.1 # CC-Lambda
lshift2 0.4 0.2 # CC-LR
lshift3 0.5 0.2 # CC-IR
end
set tce:lineresp F
set tce:imagresp T
set tce:ifreqauto F
set tce:ifreq 0.0392901 0.118358 0.198912 0.282029 0.368919
task tce energy


MICROCYCLE DIIS UPDATE: 9990 5
9991          0.0098053515183     3.7     3.9
9992 0.2005379883579 3.7 3.9
9993 0.0720092869102 3.6 3.9
9994 0.1233912092436 3.6 3.9
9995 0.0612804272654 3.6 3.8
MICROCYCLE DIIS UPDATE: 9995 5
9996 0.0098053515183 3.6 3.8
9997 0.2005379883579 3.7 3.9
9998 0.0720092869102 3.6 3.9
9999 0.1233912092436 3.7 3.9
**** 0.0612804272654 3.6 3.8
MICROCYCLE DIIS UPDATE: 10000 5
------------------------------------------------------------------------
tce_energy: maxiter exceeded 10001
------------------------------------------------------------------------
current input line :
144: task tce energy
------------------------------------------------------------------------
This type of error is most commonly
associatated with calculations not reaching convergence criteria
------------------------------------------------------------------------
For more information see the NWChem manual at
http://www.nwchem-sw.org/index.php/NWChem_Documentation
For further details see manual section:
0:0:tce_energy: maxiter exceeded:: 10001
(rank:0 hostname:aims326 pid:22508):ARMCI DASSERT fail. armci.c:ARMCI_Error():260 cond:0
tce_energy: maxiter exceeded    10001
------------------------------------------------------------------------
------------------------------------------------------------------------


Regards
Vivekanand Gobre

  • Guest -
    Send PM
Hi,
Would you please try the same calculations with other basis sets. I would suggest the
Sadlej's pVTZ basis set. Usually, the convergence of the LR-CC equations for imaginary frequencies is much harder to achieve and may be strongly dependent on the quality of the basis set employed. The Sadlej basis set was sepcifically designed to deal with the molecular properties.
Could you also send me a copy of your output file (karol.kowalski@pnnl.gov).
Best,
Karol

Just Got Here
Threads 2
Posts 4
Hi

Thanks

I will try now with Sadlej's basis set and see what i get !


Forum >> NWChem's corner >> Running NWChem



Who's here now Members 0 Guests 0 Bots/Crawler 1


AWC's: 2.5.10 MediaWiki - Stand Alone Forum Extension
Forum theme style by: AWC