[SciPy-User] multiple ODE problem with scipy.integrate.ode.ode
Pauli Virtanen
pav at iki.fi
Mon May 2 16:08:08 EDT 2011
On Mon, 02 May 2011 09:47:06 -0400, Tony Yu wrote:
[clip]
> There's some weird sharing between the two calls to integrate (e.g. with
> a single loop, moving r2.integrate before r.integrate results in r2
> integrating smoothly, but r not). I don't know what's going on under the
> hood of ode.integrate, so that's all the help I can give.
The underlying Fortran code uses SAVE statements, and is therefore not
reentrant. The author of the Python wrappers apparently forgot about that.
Damnation.
So in short, it just doesn't work; the underlying integration engine does
not support this kind of usage. Short of rewriting considerable parts of
ancient Fortran 77 code, the only thing the wrappers should do is to
raise errors.
The ODE integration stuff in Scipy has been waiting for a proper cleanup
anyway for a long time.
--
Pauli Virtanen
More information about the SciPy-User
mailing list