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

RE: GSL solver (was: Re: [ProgSoc] New struct in C++)



> -----Original Message-----
> From: owner-progsoc@xxxxxxxxxxxxxxxxxx [mailto:owner-
> progsoc@xxxxxxxxxxxxxxxxxx] On Behalf Of Matthew Beauregard
> Sent: Tuesday, 26 October 2004 6:19 PM
> To: progsoc@xxxxxxxxxxxxxxxxxx
> Subject: GSL solver (was: Re: [ProgSoc] New struct in C++)
> 
> 
> Have you (or anyone) worked with the GSL DE solver before?  I have two

Yikes! No can't say that I have... I had to find out what the Runge-Kutta
method was, but now I at least understand what you are doing to some degree.

> systems of equations, one providing inputs to the other one, and I
> need to link them up.  Both systems work fine by themselves, but I
> haven't yet worked out how to combine them.  I thought to wrap one of
> them into a function that returns the values for some time t, which
> the "master" solver func could then call, but that is tricky because t
> might be a time that the "slave" solver has already advanced past.
> Any ideas?
> 

You might be able to force the step size to a discrete value, and keep them
in sync. But it sounds like this will affect the error, since it uses a
control function to keep the error below a certain amount, and adjusts the
step size accordingly.

Do you have to run the systems in parallel, or can you do them one after the
other? (i.e. must they be solved together, simultaneously). I have some
Differential Equation notes/books here, but having an idea of the form of
the equations in the two systems might help in working out what you need.

Nigel


-
You are subscribed to the progsoc mailing list. To unsubscribe, send a
message containing "unsubscribe" to progsoc-request@xxxxxxxxxxxxxxxxxxx
If you are having trouble, ask owner-progsoc@xxxxxxxxxxxxxxxxxx for help.