How do the times in t_eval or in my experiment relate to the solver steps?

[rtimms]

When I solve a model how do the times I request in t_eval (or the times set by the period of my experiment) relate to the timesteps taken by the solver?

[rtimms]

Internally the solvers use adaptive timestepping to meet the tolerances (atol and rtol) passed to the solver. The times in t_eval are the times at which you would like the solver to return a solution. The solver takes the steps it needs to in order to meet the error tolerances, but then gives you back a solution at the times you request (which may be different from the steps the solver actually took).

There is a nice description of this in some usage notes on the SUNDIALS website. You can think of the times in t_eval as the times in tout in the "normal" mode described in the linked usage notes.

[ljt999]

Hi Dr. Robert Timms,

I have reviewed the casadi_solver and solution files, but I was unable to find the specifics of how PyBaMM performs interpolation to obtain values at the t_eval points. Could you possibly pinpoint the section of the code where this process takes place? I greatly appreciate your assistance and time in advance.

Best regards,
Yunlu