Computation speed: scipy.integrate.ode vs PYDStool

[Fabio Boschetti]

Hello, i am new to Python and to PYDStool, so i apologise in advance if this question is silly. i have tested solving a system of 6 odes with both scipy.integrate.ode vs PYDStool. i have parameterised scipy.integrate.ode as .set_integrator('vode',atol=1e12,rtol=1e9,method='adams',first_step=0.01,min_step=0.0,max_step=0.0,nsteps=1000)to compare to PYDStool (using Generator.Vode_ODEsystem). The results match very nicely, but PYDStool seems to be much slower. is this normal? am i simming something? maybe some parameters are still different between the two implementations?

[Rob Clewley]

Hi Fabio, that is normal. The VODE wrapper is esesentially a kludge to provide some minimal support for features unsupported in scipy.integrate, such as event detection and auxiliary variables
(see http://www.ni.gsu.edu/~rclewley/PyDSTool/Generators.html#VODE_integrator). PyDST is really meant to be used with the Dopri and Radau integrators, which will give you great speed boosts ove scipy. Just carefully follow the instructions to set them up on the wiki.

[Fabio Boschetti]

thank you for your prompt reply, very appreciated. i will follow the suggested links