[MITgcm-support] FLT package

[Andreas Klocker]

Hi Sergey,

You have not found any comprehensive documentation because there is 
none...so far...I've been using the flt package over the last two years 
and fixed a few things which I needed for my work but haven't written 
any documentation - i should probably do that...unless anyone else is 
keen ;)?

re comparison with other codes: I've compared trajectories with 
different advection codes written by different people in Matlab. The 
results were very similar (but that was only done in 2D). I've 
calculated trajectories in 3D and results look very reasonable...but I 
have to  look into it in more detail to be 100% happy with it.

I must admit I've never had a segmentation fault error using the flt 
package...

cheers,
Andreas


On 13/06/12 17:26, fancer fancer wrote:
> Dear mitgcm developers,
>
> I am trying to run our 3D internal waves numerical experiments with 
> flt package using. It runs well and even results look quite 
> reasonable. But still I have few questions:
>
> I have not found any comprehensive documentation about the package, 
> source code is only one. According to the source code the kernel of 
> package is 2nd and 4d order Runge-Kutta algorithm to solve ODE and 
> linear interpolation to find velocity in the middle of neighbouring 
> array indexes. But I have not got the purpose of additional 
> functionality, such as up and down movement of particles. Is there any 
> comprehensive document, which explains theoretical base of flt package?
> I found README.flt, which looks obsolete. Isn't it? For example, kpart 
> has to be initialized by vertical position in coordinate units, but 
> README says that it is actual level. Is it the same or not?
>
> In addition I have not found any publications, which concern numerical 
> experiments using MITgcm and pkg/flt. Are there such articles? Does 
> somebody compare results, which are got from the flt package, with 
> trajectories from another packages, for example, TRACMASS?
>
> And finally I have found the segmentation fault error in case of using 
> FLT package, which occurs if the main model is near to blow up, when 
> velocity is extremely high.
> The error occurs in the FLT_TRILINEAR() subroutine during actual uu 
> calculating. In this case horizontal indexes, which is passed to the 
> function, become extremely big. I guess, that such behavior can be 
> related with package limitation. But still segfault error is not well 
> way of program working, is it? May be it is reasonable to add some 
> check of array indexes in the FLT_REUNGE4() subroutine at least in the 
> 2nd, 3d and 3th steps?
>
> Yours Faithfully,
> Sergey. V. Semin
>
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20120613/26ae5049/attachment.htm>