[SciPy-User] odeint: evolving ODE system with non-penetration constraints causes issues?

[Brian Merchant]

Hi all,

I am modelling a physical system using a system of ODEs, and am imposing a
non-penetration constraint between distinct objects.

In Python, I implement the non-penetration rules in the "right hand side
function", simply denoted `f` in the `odeint` documentation. The
integration problem becomes stiff, because there can be sharp change in the
forces applied based on the activation/de-activation of non-penetration
penalties.

There are two issues I would like feedback on:

* when I supply `ixpr = True` as an argument to `odeint`, I don't get any
messages printing out that `odeint` has changed methods once the problem is
stiff -- how can I be sure that `odeint` is switching to a stiff method?

* I often get "excess work done" messages if I don't manually specify a
high `mxstep` number -- if I do specify a high number, my program tends to
crash due to running out of memory. How can I prevent this?

Kind regards,
Brian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20150704/cea0592d/attachment.html>