[MITgcm-support] adjoint run

Miroslaw Andrejczuk Andrejczuk at atm.ox.ac.uk
Mon Mar 11 05:25:07 EDT 2013


Dear Patrick,

Thanks a lot.

Best Regards,
Mirek
________________________________________
From: Patrick Heimbach [heimbach at MIT.EDU]
Sent: Monday, March 11, 2013 12:46 AM
To: mitgcm-support at mitgcm.org
Subject: Re: [MITgcm-support] adjoint run

Hi Mirek,

yes, there's a template for this in place that you should be able to adapt without too much pain:
At each timestep, the model calls a generic subroutine cost_tile
(see pkg/cost/ )
This routine
1. calls individual cost routines, depending on your choice (CPP option),
2. evaluates this cost function term as long as
      IF ( myTime .GT. (EndTime - lastinterval) ) THEN

I.e. the cost function is evaluated over the last "lastinterval" seconds.
Default is 2592000 seconds (= 30 days).
You can easily change this in data.cost by setting "lastinterval"
to something different (e.g., 31104000 = 3600*24*360 = one 360-day year).

Also, in cost_tile.F add a call to whichever CO2-related cost routine you have,
and make sure to sum them up in cost_final.F (also in pkg/cost/ ).
A good generic example is the cost function used for the shelfice package
(grep for ALLOW_COST_SHELFICE).

Cheers
-Patrick

On Mar 8, 2013, at 8:03 AM, Miroslaw Andrejczuk <Andrejczuk at atm.ox.ac.uk> wrote:

>
> Dear Support,
>
> I was wondering if you could help me to set up an adjoint simulation. I need to calculate the sensitivity for the integral over the whole globe of pCO2. However, I want to keep this function (that is the value of the integral) constant for the duration of the integration (~10 years).
>
> I would appreciate if you could either help mi to set up such simulation or provide me with the contact to person who could do that.
>
> Best Regards,
> Mirek Andrejczuk
>
> _______________________________________________
> 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




More information about the MITgcm-support mailing list