aoforce crashes on hessian calculation

[patrikrydberg]

aoforce just stops after some time with the message "error in hessian step". Some help to fix this would be appreciated

the last few lines before stopping are:

 Iteration IRREP Converged      Max. Euclidean
                 roots          residual norm

    1       a        0        2.181776522050201D-01
 
    2       a        0        2.839880319146313D-02
 
    3       a        0        3.204038207749773D-03
 
    4       a       66        3.333393416994136D-04
 
error in hessian step


the run is a DFT BP86 with RI with the following keywords to calculate freqencies

$dipgrad  file=hessian
$hessian  file=hessian
$vibrational normal modes  file=hessian
$maxcor 600

[uwe]

Hi,

I assume that you are running the em64t-unknown-linux-gnu binary of aoforce, right? There seems to be a problem with the Intel Compiler on 64 Bit machines, since even an unoptimized executable crashes at that point.

Please use aoforce from $TURBODIR/bin/x86_64-unknown-linux-gnu, i.e. the AMD64 binary version. That should run on Intel EM64T machines, and since we are using different compilers for this platform, aoforce should run without problems (just a bit slower).

We are seeking smaller cases where this bug occurs. Unfortunately, it seems to be stable if the number of basis functions in C1 is smaller than 600, so debugging takes ages...

Could you please send your input (coord and control file should be sufficient, unless you have used a non-Turbomole basis set - in that case please also include the basis file) to the Turbomole support (turbomole q cosmologic , de)

Regards,

Uwe

[patrikrydberg]

I'm using the version of aoforce in bin/i686-pc-linux-gnu/aoforce on a 64 bit machine I noticed right now, could this be the problem? I'm starting a job on the amd64 version right now. However this will take at least 20 hours so I will get back when I get some results from that