[MITgcm-support] Re: mdsio slice routines
Patrick Heimbach
heimbach at MIT.EDU
Wed May 28 14:32:35 EDT 2008
Hi Suneet,
that will be a bit hard for Matt since the Southern Ocean
doesn't really have a Southern boundary.
One could always artificially create a setup that does,
but here is where you should help out:
I had suggested earlier that you provide us
a "toy" setup that mimics your setup, includes most features
that you want to have, has input fields etc.
You would have to test and make sure that it works fine
in a forward (not adjoint) calculation
(Tom, who I reckon you are in touch with, will know what I mean).
If this works, we could take a look at what works or needs changing,
based on exactly what you have and what you want.
-Patrick
On May 28, 2008, at 12:44 PM, Suneet Dwivedi wrote:
> Hi Matthew,
> May I request you to kindly check whether the optimization run for
> southern boundary works for you with your model setup using the mdsio
> slice routines (which are working for northern boundary), whenever you
> find some time. I just want to confirm whether the problem is due to
> some discrepancy in the slice routines....
>
> I am really sorry for bothering you again...but even after seriously
> looking at the lsopt and mdsio routines, I could not find the error
> responsible for this problem.
> Anticipating your help,
> Best regards,
> Suneet
>
> On Tue, May 27, 2008 at 3:02 PM, Suneet Dwivedi
> <suneetdwivedi at gmail.com> wrote:
>> Hi Matt,
>> Thanks for your reply. The error message of the optimization run
>> (only
>> with southern or western boundary) is
>> ---------------------------------------------------------------------
>> -----------------------------------
>> forrtl: severe (67): input statement requires too much data, unit 20,
>> file /tmp/mitgcm_so_suneet/JHU_ctrl _MIT_CE_000.opt0000
>> ---------------------------------------------------------------------
>> -----------------------------------
>>
>> The interesting points to notice are that:
>>
>> (i) when I don't use mdsio slice routines and copy obcs control
>> vector
>> files from different nodes manually on the master node, my
>> optimization run (optim.x) works fine for all the boundaries without
>> any error, whereas, on using those mdsio routines I get abovesaid
>> error message. Now, the problem here is that I can not manually copy
>> obcs control vector files every time while running assimilation
>> experiment in which model is run several times in a loop.
>>
>> (ii) On using mdsio slice routines for northern and eastern boundary,
>> my optimization run (optim.x) again works fine but not with southern
>> and western boundary.
>>
>> Therefore, personally, I think there is some bug in the mdsio
>> routines
>> because the present problem arises as soon as I try to use them for
>> southern/western boundaries?
>>
>> Though I tried hard but I am not able to figure out the likely bug or
>> inconsistency either in linesearch or mdsio slice routines.
>>
>> Please help me sort out this problem.
>> Regards,
>> Suneet
>>
>> On Tue, May 27, 2008 at 2:01 PM, Matthew Mazloff
>> <mmazloff at mit.edu> wrote:
>>> Hi Suneet,
>>>
>>> I have never checked. I only use northern boundary. I do not
>>> think there
>>> is a problem with your model set-up because you said the model
>>> runs. Its
>>> the linesearch, optim.x, that doesn't work, correct? So there is
>>> likely a
>>> bug or inconsistency in there somewhere...what error does the
>>> optimization
>>> give you?
>>>
>>> -Matt
>>>
>>>
>>> On May 27, 2008, at 1:36 PM, Suneet Dwivedi wrote:
>>>
>>>> Hi Matt,
>>>>
>>>> May I request you to please let me know whether your mdsio slice
>>>> routines work for you (for your model setup) for southern or
>>>> western
>>>> boundary? I am not sure whether my current problem with
>>>> optimization
>>>> is associated with mdsio slice routines or there is any other
>>>> problem
>>>> with my model setup.
>>>>
>>>> Suneet
>>>>
>>>> On Fri, May 23, 2008 at 4:06 PM, Suneet Dwivedi
>>>> <suneetdwivedi at gmail.com>
>>>> wrote:
>>>>>
>>>>> Hi Matt,
>>>>> Just to let you know that when I don't use those mdsio slice
>>>>> routines
>>>>> and copy obcs control vector files from different nodes
>>>>> manually on
>>>>> the master node, my optimization run (optim.x) works fine for
>>>>> all the
>>>>> boundaries. But, this is not a practical solution for running
>>>>> assimiation experiment. So, I am sure that there is some
>>>>> discrepancy
>>>>> between call routines for southern and western boundaries.
>>>>> Cheers,
>>>>> Suneet
>>>>>
>>>>> On Fri, May 23, 2008 at 2:28 PM, Suneet Dwivedi
>>>>> <suneetdwivedi at gmail.com>
>>>>> wrote:
>>>>>>
>>>>>> yes everytime when i make any modification in code (including
>>>>>> change
>>>>>> in ECCO_CPPOPTIONS.h) , I always recompile optim.x.
>>>>>> Thanks,
>>>>>> Suneet
>>>>>>
>>>>>> On Fri, May 23, 2008 at 2:08 PM, Matthew Mazloff
>>>>>> <mmazloff at mit.edu>
>>>>>> wrote:
>>>>>>>
>>>>>>> just checking, but did you recompile optim.x when you added
>>>>>>> these
>>>>>>> controls
>>>>>>> (i.e. changed ECCO_CPPOPTIONS.h)?
>>>>>>>
>>>>>>>
>>>>>>> On May 23, 2008, at 1:49 PM, Suneet Dwivedi wrote:
>>>>>>>
>>>>>>>> Hi Matt,
>>>>>>>> Thanks for your reply. My do not 'work' means that the
>>>>>>>> assimilation/optimization run fails when I try to allow
>>>>>>>> southern/western open boundary control. So the model runs
>>>>>>>> absolutely
>>>>>>>> fine (as before) but optimization (optim.x) fails with south/
>>>>>>>> west OB
>>>>>>>> control.
>>>>>>>> Cheers,
>>>>>>>> Suneet
>>>>>>>>
>>>>>>>> On Fri, May 23, 2008 at 12:16 PM, Matthew Mazloff
>>>>>>>> <mmazloff at mit.edu>
>>>>>>>> wrote:
>>>>>>>>>
>>>>>>>>> Hi Suneet,
>>>>>>>>>
>>>>>>>>> Northern and southern boundaries should both use same I/O
>>>>>>>>> routines.....and
>>>>>>>>> eastern and western should both use same I/O routines. If the
>>>>>>>>> northern
>>>>>>>>> boundary works but the southern doesnt than than there must
>>>>>>>>> be a
>>>>>>>>> discrepancy
>>>>>>>>> between call routines....that should not be the case....I
>>>>>>>>> had a quick
>>>>>>>>> look
>>>>>>>>> for this discrepancy but couldnt find it....not sure whats
>>>>>>>>> going on.
>>>>>>>>> What
>>>>>>>>> do you mean the routines do not 'work' for the south and west
>>>>>>>>> boundaries?
>>>>>>>>>
>>>>>>>>> -Matt
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> On May 22, 2008, at 5:18 PM, Suneet Dwivedi wrote:
>>>>>>>>>
>>>>>>>>>> Hi Matt,
>>>>>>>>>> Thanks for your reply. But, what is the difference between
>>>>>>>>>> northern
>>>>>>>>>> and southern boundary and similarly between eastern and
>>>>>>>>>> western
>>>>>>>>>> boundary?
>>>>>>>>>> As per your advice, I modified MDSREADFIELDYZ and
>>>>>>>>>> MDSWRITEFIELDYZ.
>>>>>>>>>> Now
>>>>>>>>>> the routines work for eastern boundary in addition to the
>>>>>>>>>> northern
>>>>>>>>>> boundary but still not for south and west.
>>>>>>>>>> Waiting for your response,
>>>>>>>>>> Cheers,
>>>>>>>>>> Suneet
>>>>>>>>>>
>>>>>>>>>> On Thu, May 22, 2008 at 3:04 PM, Matthew Mazloff
>>>>>>>>>> <mmazloff at mit.edu>
>>>>>>>>>> wrote:
>>>>>>>>>>>
>>>>>>>>>>> Hi Suneet,
>>>>>>>>>>>
>>>>>>>>>>> The northern (southern) boundary is xz. I think you just
>>>>>>>>>>> need to
>>>>>>>>>>> apply
>>>>>>>>>>> these changes to the eastern (western) boundary = yz. so
>>>>>>>>>>> for
>>>>>>>>>>> example
>>>>>>>>>>> just
>>>>>>>>>>> change (appropriately) MDSREADFIELDYZ to look like
>>>>>>>>>>> MDSREADFIELDXZ.
>>>>>>>>>>>
>>>>>>>>>>> -Matt
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> On May 22, 2008, at 2:39 PM, Suneet Dwivedi wrote:
>>>>>>>>>>>
>>>>>>>>>>>> Hi Patrick and Matt,
>>>>>>>>>>>> I wish to modify mdsio slice routines (mdsio_gl_slice.F,
>>>>>>>>>>>> mdsio_slice.F, mdsio_slice_loc.F, mdsio_readvector.F,
>>>>>>>>>>>> mdsio_writevector.F) sent to me by Matt so as to make
>>>>>>>>>>>> them work
>>>>>>>>>>>> for
>>>>>>>>>>>> other boundaries in addition to the northern boundary.
>>>>>>>>>>>> Matt told
>>>>>>>>>>>> me
>>>>>>>>>>>> that these routines are written only for the northern
>>>>>>>>>>>> boundary. I
>>>>>>>>>>>> am
>>>>>>>>>>>> not sure which variables in the routines denote the
>>>>>>>>>>>> northern
>>>>>>>>>>>> boundary.
>>>>>>>>>>>> This information is necessary for me to extend the
>>>>>>>>>>>> routines for
>>>>>>>>>>>> any
>>>>>>>>>>>> other boundary. May I request you to please tell me the
>>>>>>>>>>>> same.
>>>>>>>>>>>> Cheers,
>>>>>>>>>>>> Suneet
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> MITgcm-support mailing list
>>>>>>>>> MITgcm-support at mitgcm.org
>>>>>>>>> http://mitgcm.org/mailman/listinfo/mitgcm-support
>>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> MITgcm-support mailing list
>>>>>>>> MITgcm-support at mitgcm.org
>>>>>>>> http://mitgcm.org/mailman/listinfo/mitgcm-support
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> MITgcm-support mailing list
>>>>>>> MITgcm-support at mitgcm.org
>>>>>>> http://mitgcm.org/mailman/listinfo/mitgcm-support
>>>>>>>
>>>>>>
>>>>>
>>>
>>>
>>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support
---
Patrick Heimbach | heimbach at mit.edu | http://www.mit.edu/~heimbach
MIT | EAPS 54-1518 | 77 Massachusetts Ave | Cambridge MA 02139 USA
FON +1-617-253-5259 | FAX +1-617-253-4464 | SKYPE patrick.heimbach
More information about the MITgcm-support
mailing list