Solving ode with IMSL routines

[leo_vortex]

Dear experts,

I use IMSL for solving system of ode with the following routines: ivprk, ivpag, ivmrk. And I'm quite satisfied with them,

it's easy to use them, you don't need to deal with the code,

everythingwas done for you before.

But I realized that there are limits for the number of equations.

And my problem requires a huge number of equations.

The limit for ivrprk it's about 4.2*10e7, for ivpag and ivmrkit's 2.3*10e7.

It says: "Terminal Error 8 from ivpag.The required storage cannot be allocated. The work space requirements is based on NEQ=..."

Is it possible to overcome this problem? Should I increase

the amount of memory? Now I have 512 mb only.

You may say try something like NAG or Numerical recipes,

or try to code by yourself,

but can you tell will I get any benefits? Yes it's more

flexible to deal with the code than with compiled routines,

but it costs money.

Has anybody compared performance of IMSL with NAG or Numerical Recipes for my particular problem and in general?

Also Ihave a different kind ofproblem with CVF & IMSL.

I run the program with the ode routines,

but with a lot intermediate computations (discretization of

tripple integrals, solving algebraic equations),

butbegining with a certain number of

equations, after I ran the program,it begins to workwiththe hard drive intensivelyand nothing happens for a long time and I have to close theprogram from the task manager and after this the system works very slowly and I have to restart it.

I hope it's not too much for one time ;-)

I appreciate any advice and help.

Message Edited by Leo_vortex on 10-23-2004 03:48 PM

[g_f_thomas]

If your system of odes is linear, see if this helps:

www.maths.uq.edu.au/expokit/

Pay attention to the Krylov subspace methods for pruning the number of coupled linear odes you have to deal with.

Ciao,
Gerry T.