Re: pdepe

[Carlo De Falco]

Hi,

Please keep the maintainers list in CC as others may wish
to comment on this topic too.

On 28 Apr 2016, at 07:55, Nikola Yuroukov <address@hidden> wrote:

> Hi Carlo,
> 
> Do you have any idea where the pdepe code should go?
> I don't see anything like ./octave/scripts/pde

If you plan to make an m-file-only implementation (which I do reccommend)
I would suggest you start working on this as an indipendent code, we
can then move it into Octave later.

If you have a publicly visible repository for your code, e.g. on 
bitbucket or github, it would help to see your code while you work 
on it, in order to give more detailed advice.

> Do we have a basic idea about the code architecture?

The algorithm described in that project is not the most modern
nor the most commonly used, the main reason for having that function
in Octave is to allow to "liberate" existing m-file code that depends 
on it.

For this reason, the main requirement is to make this function as
much as possible compatible to the Matlab version:

www.mathworks.com/help/matlab/ref/pdepe.html

The algorithm is described in detail in this paper:

[1] Skeel, R. D. and M. Berzins, "A Method for the Spatial Discretization of 
Parabolic 
Equations in One Space Variable," SIAM Journal on Scientific and Statistical 
Computing, 
Vol. 11, 1990, pp.1–32.

While working on this project, please consider that, while it is OK 
to read public documentation for Matlab functions, you should 
NEVER LOOK AT THEIR SOURCE CODE.


> Regards,
> Nikola Yurukov

Thanks for your interest,
c.