Segfault calculating frequencies


Clicked A Few Times
I've successfully run an NWChem DFT geometry optimization, and would now like to calculate the vibrational frequencies, but I get a segmentation fault during the frequency calculation. My input file is at the end of this message. When I run the calculation, I get
 Parallel integral file used   46901 records with       0 large values


 HESSIAN: the one electron contributions are done in       6.5s

MA_verify_allocator_stuff: starting scan ...
stack block 'hess scratch', handle 48, address 0x2aaadc26b4e8:
        current right signature 3943625999 != proper right signature 1431655765
stack block


followed by a lot of binary digits. I can't tell what's going on, whether NWChem is running out of memory, and I have to put an explicit MEMORY directive in the input file, whether I'm running out of memory on the nodes, and I have to run on ones with more memory per node, or whether something else is going wrong. Can someone help me sort this out?

Here's my input file.

start tmp.a1.freq

title "a1"
charge 0

geometry noautoz nocenter noautosym
 Si2        -0.05806682    -0.93779950     1.34064200
 Si2        -0.03445032     1.27405854     1.91638552
 Si2         3.99744209    -1.23220438     1.75703796
 Si2         3.95513098     0.97433893     1.36297561
 Si2        -3.98096312     0.99205950     1.49953738
 Si2        -4.02947076    -1.36024123     2.00855570
 Si          2.02773129    -1.70194383     0.51961444
 Si          1.87316484     1.70664443     0.53395843
 Si          1.98153436     0.00854821    -1.17707859
 Si         -2.14326158    -1.71440159     0.54317727
 Si         -1.92155562     1.68712045     0.49134515
 Si         -2.01205412    -0.01319687    -1.17897465
 Si         -5.89641519    -1.79982777     0.51586771
 Si         -5.97129709     1.77242849     0.42007603
 Si         -6.04046924    -0.05317540    -1.09920790
 Si          5.93480373    -1.79234046     0.48024169
 Si          5.97139635     1.78644238     0.43938930
 Si          6.03796059    -0.02226486    -1.10782795
 Si         -4.05243358    -0.00342492    -2.42071023
 Si         -0.01307394     0.00115986    -2.50558389
 Si          4.03686808    -0.00205712    -2.41547886
 H           1.98364000     3.11880000     0.04817000
 H           1.98364000    -3.11880000     0.04817000
 H          -1.98364000    -3.11880000     0.04817000
 H          -1.98364000     3.11880000     0.04817000
 H          -5.83159000     3.14518000    -0.15149000
 H          -7.17306000     1.79451000     1.33163000
 H          -7.17306000    -1.79451000     1.33163000
 H          -5.83159000    -3.14518000    -0.15149000
 H           5.83159000    -3.14518000    -0.15149000
 H           7.17306000    -1.79451000     1.33163000
 H           7.17306000     1.79451000     1.33163000
 H           5.83159000     3.14518000    -0.15149000
 H          -7.21415000     0.00000000    -2.04898000
 H          -4.08404000     1.19802000    -3.30985000
 H          -4.08404000    -1.19802000    -3.30985000
 H           0.00000000     1.19890000    -3.39832000
 H           0.00000000    -1.19890000    -3.39832000
 H           4.08404000     1.19802000    -3.30985000
 H           4.08404000    -1.19802000    -3.30985000
 H           7.21415000     0.00000000    -2.04898000
 P2         -3.91684710     1.92203730     3.65888921
 H2         -2.78633679     1.51005747     4.38988189
 H2         -4.97150790     1.54232998     4.51276243
 H2         -3.88610055     3.32189383     3.85231833
end

set geometry:actlist 1:21 42:45

basis
  * library 6-31G**
end

driver
  maxiter 100
end

dft
  xc b3lyp
  mult 1
  maxiter 100
end

task dft freq

Clicked A Few Times
Quote:Rpmuller Dec 15th 2:14 pm
I've successfully run an NWChem DFT geometry optimization, and would now like to calculate the vibrational frequencies, but I get a segmentation fault during the frequency calculation. My input file is at the end of this message. When I run the calculation, I get...followed by a lot of binary digits. I can't tell what's going on, whether NWChem is running out of memory, and I have to put an explicit MEMORY directive in the input file, whether I'm running out of memory on the nodes, and I have to run on ones with more memory per node, or whether something else is going wrong. Can someone help me sort this out?

Here's my input file.
...


I think the problem is the line that freezes the geometries of some of the atoms:
set geometry:actlist 1:21 42:45


Removing that line makes the calculation run. Sorry for the excitement; I should have spotted that before I posted.


Forum >> NWChem's corner >> Running NWChem