Compute initial concentrations from initial OCV

[NicolaCourtier]

Feature description

Currently, a FittingProblem can take an initial SOC value but it would be better to compute the initial concentrations from the open-circuit voltage (OCV) at the start of a measurement.

We should replace the init_soc option with the functionality to pass an init_ocv. When init_ocv is specified, the model should recalculate the initial concentrations before running the simulation.

Motivation

Typically it is the initial OCV and not the initial SOC that can be obtained from voltage measurements.

Possible implementation

No response

Additional context

No response

[MarkBlyth]

Is it possible to have the choice of specifying either an initial SOC or an initial OCV? Ie. adding an initial_ocv option, without removing the initial_soc kwarg, so that users can choose one or the other. (This might be what you're already planning, but just wanted to check and make sure!)

For fitting empirical models, it's more helpful to specify the initial SOC than the initial OCV, as the standard approach is to work with known SOCs, taken from a combination of coulometry and capacity-measuring experiments. Removing the option for specifying an initial_soc would therefore add an extra burden when working with these sorts of whole-cell data.

[NicolaCourtier]

Short answer: yes!

For PyBaMM models, the initial SOC (for ECMs) or initial concentrations (for EChems) form part of the parameter set.
Therefore it will remain possible to define an initial state via the parameter_set of the model. In this case, an initial value does not need to be passed to the problem class.

For a FittingProblem where the initial SOC or concentrations are unknown, then these values can be determined from the initial OCV. In #425, this estimation can be performed automatically on each iteration of the optimisation by passing the init_ocv value to FittingProblem.

For a DesignProblem with an EChem model, the initial concentrations must be recomputed on each iteration for consistent evaluation of the design cost. For an ECM, the initial SOC is just part of the parameter set.

I hope that makes sense, I'll request a review for you to check when it's ready.