Please describe the problem you are trying to solve.

Iterative mappings start time windows with a zero initial guess.
In case of read mappings, this initial guess is used to map all data samples in sequence.

So if Solver A writes data at 0, 0.5, 1 during a time window and Solver B uses an iterative read mapping, then there are 3 stamples to be mapped.
Stamples are mapped in sequence and during each map, the initial guess is updated.
The initial guess is zero in the first map of the first iteration or the result of the map of the last stample of the last iteration.

Does this make sense? How does this impact the convergence of the iterative mapping.

Additional context

Each iteration may use different timesteps, so it is not possible to save initial guesses per timestamp.

Example: