[MITgcm-support] ptracer start at advanced time

[Patrick Heimbach]

Hi Suneet,

you are not asking the relevant question.
What you've asked can be easily accomplished with a
dump & restart at the relevant time when actual tracers
start to become available.

What you want to ask is, how to make it work *and*
running the adjoint back in time beyond tracer start time.

That can probably be done, "easiest" is to
insert ptracers initialization into forward model
at the time you want it, and then remove these
portions in the AD-generated code, but it's a bit fiddly.
Another alternative is to use "divided adjoint" (DIVA)
and trick the outermost checkpoint levels.
That's also fiddly.

Either way requires in-detailed look at your problem
(former way seems "easier" if you know where in the
adjoint code to intervene).

-p.

On Mar 17, 2010, at 10:39 PM, Suneet Dwivedi wrote:

> Hi Everybody,
>
> I want to start ptracer at an advanced timestep than the model  
> initial time. I tried to put "PTRACER_niter0 > niter0" for this  
> purpose, but it's of no use and returned me zeros everywhere.
>
>
> Is there any simple work around of this problem? I wish to use  
> niter0=0 and PTRACERS_Iter0=1000, for example to start the tracers  
> at 1000 time step rather than at initial time 0.
>
>
> Please suggest me way to do this.
>
> Thanks,
>
> Suneet

---
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


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20100317/a40b3491/attachment.htm>