octave-maintainers
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 1D PDE solver for Octave


From: Bill Greene
Subject: Re: 1D PDE solver for Octave
Date: Wed, 6 Jul 2016 12:30:05 -0400

I believe that octave/ode15s will have the same performance problems as daspkĀ 
when implementing a MOL PDE solver that requires FD calculation of the jacobian.
Either Sundials will have to support that for sparse jacobians or it would have to
be implemented specially for octave/ode15s.

On Wed, Jul 6, 2016 at 7:50 AM, Carlo De Falco <address@hidden> wrote:

On 6 Jul 2016, at 13:42, Bill Greene <address@hidden> wrote:

> lsode wouldn't handle the DAE that results when some of the PDE are elliptic.
>
> I believe that a daspk implementation would be much slower than the current one
> because it would use a full-matrix jacobian whereas Sundials ida is exploiting
> the narrow bandwidth of the matrix.

If you had a complete implementation based on daspk done, switching to ode15s
at a later time would be trivial.

If you cannot wait for ode15s to be finished, you could help with its development.

c.





reply via email to

[Prev in Thread] Current Thread [Next in Thread]