[sci.math] Need info on stiff ODE solvers

mbrown@hubcap.UUCP (Mike Brown) (10/07/87)

I am looking for information (and source if it is available) for
programs/routines which can be used to solve stiff ODEs.  I need to
solve a set of equations which have rates in the range 1.e-7 to 1.e5.

I am especially interested in hearing opinions about routines such as
DGEAR and EPISODE.

Thanks in advance.
-- 
Mike Brown				Department of Computer Science
UUCP: ...gatech!hubcap!mbrown		Clemson University
Internet: mbrown@hubcap.clemson.edu	Clemson SC 29634-1906
					(803)656-2838

evan@ndcheg.UUCP (Evan Bauman) (10/07/87)

In article <545@hubcap.UUCP>, mbrown@hubcap.UUCP (Mike Brown) writes:
> I am looking for information (and source if it is available) for
> programs/routines which can be used to solve stiff ODEs.  I need to
> solve a set of equations which have rates in the range 1.e-7 to 1.e5.
> 
> I am especially interested in hearing opinions about routines such as
> DGEAR and EPISODE.
> 

I'm a heavy user of DGEAR in my research.  It will use either 12th order
Adams predictor-corrector methods or Gear's method for stiff equations.
So far, it's been very reliable and moderately fast.  It's certainly
faster than Runge-Kutta and other similar methods.

DGEAR is part of the IMSL package so it's not cheap.  Check with your
organization's computing center with the hopes that they will foot
the bill.

	Evan Bauman
	University of Notre Dame
	..!iuvax!ndmath!ndcheg!evcae

ags@j.cc.purdue.edu (Dave Seaman) (10/08/87)

In article <292@ndcheg.UUCP> evan@ndcheg.UUCP (Evan Bauman) writes:
>DGEAR is part of the IMSL package so it's not cheap.  Check with your
>organization's computing center with the hopes that they will foot
>the bill.

IMSL has released a new version of their library which is split into three
parts:  MATH/LIBRARY, STAT/LIBRARY and SFUN/LIBRARY.  The subroutine DGEAR
has been replaced by a new routine IVPAG, which is part of MATH/LIBRARY.
IVPAG can use the Adams-Moulton or Gear method.  I have not actually used
this routine, so I can't tell you how it compares with DGEAR.
-- 
Dave Seaman	  					
ags@j.cc.purdue.edu