DFT calculation on spin-orbit couplings: Insufficient memory in ecp int2


Clicked A Few Times
Hello everybody!
I'm running my first computations with Nwchem.
As said in the post title, I'm intersted in calculating second-order spin-orbit couplings.

this is my input:

start hocl_sodft
echo
Memory total 23000 mb
charge 0
geometry noautoz noautosym units angstrom
zmatrix
H
O 1 ho
Cl 2 ocl 1 a1
constants
ho 0.963282335298047
ocl 1.693339610308074
a1 102.908236727500196
end
end
basis "ao basis" cartesian print
...
END
ECP
...
END
SO
...
END
dft
mult 1
xc b3lyp
odft
grid fine
convergence energy 1.000000E-08
convergence density 1.000000E-06
convergence gradient 1E-06
iterations 100
end
task dft
task sodft

For the water molecule, everything worked fine: however, when I tried with heavier molecules (with halogens), I received the following error:

NWChem DFT Module
                                -----------------


 Caching 1-el integrals
Insufficient memory in ecp_int2 99

I think there should be an error in the memory space definition, beacuse it's not possible that a dft energy calculation (with simple basis set) dosn't run with 23000Mb.... :-)

Thank you in advance for your help!

Best,
Daniela

Forum Vet
Daniela
Could you post the complete input file?
It is likely that you have hit a NWChem bug and, in order to debug it, we need the full input file.

Clicked A Few Times
Yes, of course!!


start hobr-sodft
echo
Memory total 23000 mb
charge 0
geometry noautoz noautosym units angstrom
zmatrix
H
O 1 ho
Br 2 obr 1 a1
constants
ho 0.962794853275482
obr 1.819610101790230
a1 103.452259890883681
end
end
basis "ao basis" cartesian print
  1. BASIS SET: (4s)
H S
    13.3600000              1.0000000
H S
     2.0130000              1.0000000
H S
     0.4538000              1.0000000
H S
     0.1233000              1.0000000
  1. BASIS SET: (4s,4p)
O S
     8.6570000              1.0000000        
O S
     0.8692000              1.0000000
O S
     0.3994000              1.0000000
O S
     0.1978000              1.0000000      
O P
    13.3400000              1.0000000
O P
     3.0160000              1.0000000         
O P
     0.8489000              1.0000000
O P
     0.2371000              1.0000000    
  1. BASIS SET: (3s,3p,4d)
Br S
     1.3800000              1.0000000        
Br S
     0.5223000              1.0000000
Br S
     0.1727000              1.0000000
Br P
     2.5370000              1.0000000
Br P
     0.4497000              1.0000000
Br P
     0.1396000              1.0000000
Br D
    62.4100000              1.0000000
Br D
    17.2700000              1.0000000
Br D
     5.5830000              1.0000000
Br D
     1.7320000              1.0000000
END
ECP
O nelec 2
O ul
2 10.02859998 -0.79842000
2 34.19799995 -5.76684701
1 100.00389957 -1.48645601
O S
2 2.24790001 11.21630394
2 2.40490001 -16.34447694
1 4.37400001 1.04294400
0 2.18920001 2.19389099
Br nelec 18
Br ul
2 2.54780006 -3.20746708
2 9.90429974 -17.81955910
2 32.94390106 -41.36774826
2 98.63279724 -102.00952911
1 287.78079224 -13.96452522
Br S
2 1.81379998 41.54266739
2 2.07699990 -130.86531067
2 2.65140009 229.11135864
2 3.39140010 -139.15881348
1 5.07940006 34.69166565
0 22.00860023 3.13891602
Br P
2 2.34380007 -68.31208801
2 2.78999996 236.41137695
2 3.77500010 -285.20968628
2 5.22399998 192.27485657
1 15.22420025 13.39648247
0 14.20930004 5.33779001
END
SO
Br ul
2 2.54780006 -0.039374
2 9.90429974 0.262533
2 32.94390106 -0.562529
2 98.63279724 2.001467
1 287.78079224 0.751481
Br P
2 2.34380007 -2.114400
2 2.78999996 8.375434
2 3.77500010 -12.007087
2 5.22399998 5.418710
1 15.22420025 -0.570016
0 14.20930004 0.023705
O ul
2 10.028600 0.007783
2 34.198000 -0.006240
1 100.003900 0.036949
END
dft
mult 1
xc b3lyp
odft
grid fine
convergence energy 1.000000E-08
convergence density 1.000000E-06
convergence gradient 1E-06
iterations 100
end
task dft
task sodft

thank you!

Dani

Forum Vet
Daniela
Here is how to the fix the bug you have uncovered by recompiling your NWChem binary


cd $NWCHEM_TOP/src
wget http://nwchemgit.github.io/images/Ecpmem.patch.gz
gzip -d Ecpmem.patch
cd NWints/api
make
cd ../..
make link

Clicked A Few Times
Quote:Edoapra Apr 11th 6:10 pm
Daniela
Here is how to the fix the bug you have uncovered by recompiling your NWChem binary


cd $NWCHEM_TOP/src
wget http://nwchemgit.github.io/images/Ecpmem.patch.gz
gzip -d Ecpmem.patch
cd NWints/api
make
cd ../..
make link




Hello,
thanks for helping me.

However it didn't completely work. The input is always the same, but now the error is:

                                NWChem DFT Module
-----------------


 Caching 1-el integrals
------------------------------------------------------------------------
Insufficient memory in ecp_local2 99


Thank you in advance, again!

Best,
Daniela

Forum Regular
Hi Daniela,

It seems that in the presence of both ECPs and SO ECPs the memory estimator would always pick the last value of the estimated memory rather than the maximum. The patch below addresses this:

 cd $NWCHEM_TOP/src
wget http://nwchemgit.github.io/images/Ecpmem2.patch.gz
gunzip -d Ecpmem2.patch
patch -p0 < Ecpmem2.patch
cd NWints/api
make
cd ../..
make link

For me this fixed the problem with the HOBr molecule you posted above. Please give this a try and let me know if any problems persist.

Huub

Clicked A Few Times
Quote:Huub Apr 24th 4:09 pm
Hi Daniela,

It seems that in the presence of both ECPs and SO ECPs the memory estimator would always pick the last value of the estimated memory rather than the maximum. The patch below addresses this:

 cd $NWCHEM_TOP/src
wget http://nwchemgit.github.io/images/Ecpmem2.patch.gz
gunzip -d Ecpmem2.patch
patch -p0 < Ecpmem2.patch
cd NWints/api
make
cd ../..
make link

For me this fixed the problem with the HOBr molecule you posted above. Please give this a try and let me know if any problems persist.

Huub



Hi,
It worked!!! Thank you very much!
Now I can compute all the spin-orbit couplings I want!

thanks again :-)

best,
Daniela


Forum >> NWChem's corner >> Running NWChem