[MITgcm-support] Divided adjoint

[C Spencer Jones]

Hi everyone,
I have been trying to run the divided adjoint. When I run the non-diva
code, the model runs successfully. For the divided adjoint, the model runs
forward fine and then blows up a few hundred timesteps into running
backwards. The first few sensitivities look sensible. Can anyone suggest
what I might be doing wrong?

My setup is based on the verification experiment tutorial_global_oce_optim,
but has been modified significantly. I am attaching data, packages.conf,
AUTODIFF_OPTIONS.h and ECCO_CPPOPTIONS.h.
Thanks very much,
Spencer
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20170501/74d56271/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: packages.conf
Type: application/octet-stream
Size: 192 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20170501/74d56271/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: AUTODIFF_OPTIONS.h
Type: text/x-chdr
Size: 2083 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20170501/74d56271/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ECCO_CPPOPTIONS.h
Type: text/x-chdr
Size: 2974 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20170501/74d56271/attachment-0001.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: data
Type: application/octet-stream
Size: 2729 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20170501/74d56271/attachment-0001.obj>