The treatment of the DFN model to achienve such high computational efficiency in PyBaMM

[xwpken]

Hello, I am surprised by the high computational efficiency of PyBaMM. For the DFN model, solving for 1000 time steps may only take 3 seconds！

I am curious about the treatment of the DFN model, as we all know it is composed of several PDEs and algebraic equations. I noticed that pybamm relies on CasADi to solve the DAE system. Is it to first discretize the PDEs in DFN model in space to obtain the time-dependent ODEs and algebraic equations, and then call the integrator generated by CasADi to solve it?

Thank you!