[Top][All Lists]

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

Re: DAE daspk solver problem

From: Bill Greene
Subject: Re: DAE daspk solver problem
Date: Fri, 28 Oct 2016 14:00:01 -0400

I did not try this simple test in daspk but I did try it in ida (with C-api).
Even after carefully
insuring that I had consistent initial conditions, ida could not advance to the
first time step. So I am very doubtful that daspk will solve this example in its
index-3 form.

Frankly, I was quite surprised that radau5 was able to do so particularly since
it doesn't appear to have an option to specify an initial y'. 

On Fri, Oct 28, 2016 at 1:17 PM, SunilShah <address@hidden> wrote:
As I understand it, Radau5.f demands a structure of DAEs with Linearly
implicit form:

M Y' = F(Y,T)

DASPK admits allows a more general form F(Y,Y',T)=0.  There can be problems
that can not be transformed to the linearly implicit form.

For large scale problems, (> 1e5 variables), retaining sparsity is an
important issue, so DASPK structure can offer some advantages.  Even when
the equations can be symbolically transformed, symbolic manipulation may not
scale well (perhaps the problem with scaling in Modelica),  or may destroy
sparsity and therefore have inefficient discrete propagation.

Lastly, the fortran DASPK / DASKR can take advantage of implicit sparse
solvers.  Do octave implementations allow this?

View this message in context:
Sent from the Octave - General mailing list archive at

Help-octave mailing list

reply via email to

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