[MITgcm-support] MIT with OpenAD

Wed Jul 4 08:50:05 EDT 2012

       Dear Utke,

   Thank you very much for your reply. At a current moment I read all 
manual on openAD, studied the suggested examples and now I am trying to 
compile MITgcm under openAD. Unfortunately I do not have much success 
with it and would be happy if you can give me some guidance how to do 
it. I did the following steps: make adDepend, make adAll. As I  followed 
the Makefile, it had created ad_input_code_sf.f and now tries to convert 
it to ad_input_code_sf.B. However, it turned out that ad_input_code_sf.f 
contains a lot of syntax errors (IHMO, produced by cb2m scripts). So I 
have a feeling that I am doing something wrong.  I would be very 
grateful, if you can provide me with any guidance/manual/howto on MITgcm 
with OpenAD.

   Just in case, I attached my makefile to this letter.

   With best regards,

   Andrey.

On 2012-07-03 22:17, Jean Utke wrote:
> Hi,
>
> going back through my inbox - I am not sure why but I think I somehow
> mentally merged
> you and Abhisek  into one person - did you see the info I had sent  -
> do you need more info?
>
> Jean
>
>
> On 06/22/2012 10:03 AM, andrey.vlasenko wrote:
>>
>>   Dear Partick and Jean,
>>
>>   I probably confused you with my request in last e-mail. Currently 
>> I have already installed and working OpenAD. As Abhisek, I need some 
>> pointers on how to connect OpenAD to *MITgcm*. I would be very 
>> grateful to you, if you send it to me.
>>
>>   With best regards,
>>
>>   Andrey.
>>
>>
>>
>> On 2012-06-22 03:40, Abhisek Chakraborty wrote:
>>> Respected Dr. Patrick Heimbach & Dr. Jean,
>>>
>>> I requested Thomas Kaminski (one of the TAF developres at FastOpt) 
>>> for
>>> an access to TAF, however he informed me about paying 10,000 Euro 
>>> for
>>> the non-commercial use of TAF, he didn't mention about trial 
>>> license
>>> or something like that. Now at present, my organization cann't pay
>>> such a huge money.
>>>
>>> So, I think, I should opt for OpenAD. I have downloaded OpenAd and
>>> working on building it.
>>>
>>> I need guidance from you and Jean to link MITgcm with openAd
>>> (preferably some step by step guidelines).
>>>
>>> Thanking in advance,
>>> Abhisek
>>>
>>> On 6/22/12, Patrick Heimbach <heimbach at mit.edu> wrote:
>>>> Oops,
>>>> a little typo (Freudian?) in my previous email:
>>>> I meant to say:
>>>> "Jean may send you some specific pointers on how to connect OpenAD 
>>>> to
>>>> *MITgcm*, if need be."
>>>>
>>>> -Patrick
>>>>
>>>> On Jun 21, 2012, at 2:48 PM, Patrick Heimbach wrote:
>>>>
>>>>>
>>>>> Abhisek,
>>>>>
>>>>> MITgcm works with essentially two different choices of AD tools:
>>>>> 1. TAF (and to limited extent its predecessor TAMC)
>>>>> 2. OpenAD
>>>>>
>>>>> Our default tool used is TAF, a commercial tool which grew
>>>>> out of the non-commercial predecessor TAMC.
>>>>> For you to be able to use TAF (via a script called "staf")
>>>>> you need to obtain a license from Fastopt which develops TAF.
>>>>> You should send a request to info at fastopt.de
>>>>> They may be able to give you a certain (hopefully significant)
>>>>> period of free trial license for you to explore the utility of 
>>>>> the tool.
>>>>>
>>>>> Regarding the alternative open-source, OpenAD,
>>>>> I've sent some initial pointers in below email on how to
>>>>> obtain and implement the OpenAD tool locally.
>>>>> Jean may send you some specific pointers on how to connect
>>>>> OpenAD to TAF, if need be.
>>>>>
>>>>> -Patrick
>>>>>
>>>>> On Jun 21, 2012, at 12:54 PM, Jean Utke wrote:
>>>>>
>>>>>> Hi Abhisek,
>>>>>>
>>>>>> I will get back to you with some info before  COB today.
>>>>>>
>>>>>> jean
>>>>>>
>>>>>> On 06/21/2012 11:50 AM, Abhisek Chakraborty wrote:
>>>>>>> Respected Dr. Patrick Heimbach,
>>>>>>>
>>>>>>> Generation of AD code for OpenAD example depends on the 
>>>>>>> adjoint_f95
>>>>>>> ("-adof=...adjoint_f95") which uses "tamc"&  "staf" executables
>>>>>>> (compilers). Now my query is whether TAMC/TAF are freely 
>>>>>>> available? if
>>>>>>> so, please post the link to download these packages.
>>>>>>>
>>>>>>> Thanking in advance,
>>>>>>> Abhisek
>>>>>>>
>>>>>>> On 6/21/12, Patrick Heimbach<heimbach at mit.edu> wrote:
>>>>>>>>> Andrey,
>>>>>>>>>
>>>>>>>>> below website is a good starting point (it's not 
>>>>>>>>> password-protected):
>>>>>>>>> http://www.mcs.anl.gov/OpenAD/
>>>>>>>>>
>>>>>>>>> It contains extensive instructions on how to download and 
>>>>>>>>> install
>>>>>>>>> OpenAD,
>>>>>>>>> along with a suite of test examples.
>>>>>>>>>
>>>>>>>>> Relevant MITgcm verification experiment is
>>>>>>>>> (if you've downloaded the MITgcm repository):
>>>>>>>>> verification/OpenAD/
>>>>>>>>>
>>>>>>>>> Please give that a try.
>>>>>>>>>
>>>>>>>>> -Patrick
>>>>>>
>>>>>>
>>>>>> -- Jean Utke
>>>>>> Argonne National Lab. / MCS (240-1156)
>>>>>> 9700 S Cass Ave., Argonne, IL, 60439, USA
>>>>>> phone +1 630 252 4552 / mobile  +1 630 363 5753
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> 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 54-1420 | 77 Massachusetts Ave | Cambridge MA 02139 
>>>>> USA
>>>>> FON +1-617-253-5259 | FAX +1-617-253-4464 | SKYPE 
>>>>> patrick.heimbach
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> 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 54-1420 | 77 Massachusetts Ave | Cambridge MA 02139 USA
>>>> FON +1-617-253-5259 | FAX +1-617-253-4464 | SKYPE patrick.heimbach
>>>>
>>>>
>>>>
>>>
>>> _______________________________________________
>>> 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
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Makefile
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20120704/0b71f948/attachment-0001.el>