Re: NaNs in error term in BiCGstab error 15 years into simulation

Hi Leif,

To perturb the dump we have the perturbation script installed under $UMDIR. You just need to load python, e.g.

module load cray-python

DUMP=du415a.da20150101_00

$UMDIR/scripts/perturb_theta.py $DUMP.orig --output $DUMP

I had a look at your dump files and (0,388) is in there “THETAVD AFTER TIMESTEP”

You can use mule as a quick way to check that the dumps are different:

mule-cumf $DUMP.orig $DUMP

You should then at least see the model follow a different evolution. If it is still failing you can try and restart from an earlier (perturbed) dump and/or reduce the timestep as you had tried.

Let me know how you get on with this.

Annette