[MITgcm-support] end-of-file during read file divided.ctrl

[Dani Jones]

Hi Mario,

Thank you for telling the 2*N trick, but I guess the issue is not here,
> because the cost function can be calculated.


The cost function is calculated during the forward part of the run. So,
back when I used to run into this issue, the cost function would indeed be
non-zero. To say it another way, you can have a non-zero cost function and
still run into this issue.


> Now I have all the expected outputs, costfinal and ADJ* files, the
> costfinal and sensitivity seems right. (Precisely my cost was zero because
> the readBinaryPrec in data does not match)
>

Great!

So the model finished, but weirdly  this this "end of file during read"
> error still exists,   after the  the forward call and then
> 1st divided.ctrl is created, this err message appears
>

I take it you're using the divided adjoint (DIVA) approach? I'll have to
send my apologies, as I don't know much about that. I'm guessing you've
checked that the end-of-file characters are consistent in all the namelist
files?

Thanks,
Dani




On Wed, Mar 22, 2023 at 8:33 PM mario wrk <wrkmario at gmail.com> wrote:

> HI Dani,
> Thank you for telling the 2*N trick, but I guess the issue is not here,
> because the cost function can be calculated. Now I have all the expected
> outputs, costfinal and ADJ* files, the costfinal and sensitivity seems
> right. (Precisely my cost was zero because the readBinaryPrec in data does
> not match)
>
> So the model finished, but weirdly  this this "end of file during read"
> error still exists,   after the  the forward call and then 1st
> divided.ctrl is created, this err message appears
> it points to some compiler issue libifcoremt.so.5, executable and srun,
> but then the second and last adjoint call ended normally and I do have all
> results
> *forrtl: severe (24): end-of-file during read, unit 76, file
> /MITgcm/verification/lab_sea/run/divided.ctrl*
> Image              PC                Routine            Line        Source
> libifcoremt.so.5   00001555553E6E79  for__io_return        Unknown  Unknown
> libifcoremt.so.5   000015555542B06E  for_read_seq_lis      Unknown  Unknown
> mitgcmuv_ad        00000000007309E0  Unknown               Unknown  Unknown
> srun: error: l30211: tasks
> 834,838,840,855,859,864-865,877,879,882-883,892,895,898,902-906,909-910,912-914,917:
> Exited with exit code 24
>
> still checking...
>
> many thanks
> Mario
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20230323/676892e0/attachment.html>