[SciPy-user] PyDSTool and SciPy (integrators)

Rob Clewley rob.clewley@gmail....
Mon Apr 21 15:59:10 CDT 2008


Anne,

I am very interested in getting spline support to a point where it
could be used as a more accurate way of representing solutions to
ODEs. I will certainly use it in my Trajectory class if you give it a
similar interface to that already used by interp1d.

>  If I want to evaluate the solution at multiple points,
>  it is usually because I want an approximation to the true solution
>  function, but such an approximation should normally have points
>  distributed in a way that takes into account the behaviour of the
>  function - more where it behaves in a complicated fashion, and fewer
>  where it is simple.

In case this can help in your problem, our current interface to the
H&W codes allows you to specify time points at which the integrator is
guaranteed to step on to as it goes along.

If someone is willing to help improve the low level interfaces to the
H&W codes by exposing internal information like the polynomial
coefficients for the dense solution then I would include that and
other trajectory-related data in our ODE support.

-Rob


More information about the SciPy-user mailing list