CODEWITHSUNDEEP
pythonscipy
Durd3nT
Durd3nT

Reputation: 177

SciPy: status of solve_ivp during integration

I am running a long ODE-integration in Python using scipy.integrate.solve_ivp. Is it possible to access the status of the integration or check at which integration step the routine is, while it is running? My integration is taking longer than expected and I would like to know whether the integrator is stuck at some step or whether the individual steps just take really long.

For future tasks; if I split the integration with solve_ivp into sub-intervals to print status messages in between, could this mess with the step-size adaptivity of certain solvers?

Thanks for any feedback!

Upvotes: 1

Views: 1252

Answers (1)

Laurent90
Laurent90

Reputation: 293

There was a GitHub pull request to add a verbose option to solve_ivp, but this has not yet been done. You can either implement it yourself by modifying scipy's solve_ivp function (should be easy), or just print the time t that is given by the solver to your ODE function. That's what I do. If your system is not too small, then you don't loose much time because of the printing.

Splitting the integration the way you suggest can work, however if you split every few time steps, you will lose time as the solver restarts each tile. The impact is related with implicit algorithms, as they compute the Jacobian of your system anew at each start.

Upvotes: 4

Related Questions