Oscar Dowson
Oscar Dowson
This needs some local changes to SDDP.jl (to add the plotting callback), so you won't be able to run the code. But, the blue line is us, the orange dashed...
Yeah. I think I encountered this before. The dashboard is very much experimental, which is why it's off by default.
* Stationary MDPs - A single box with a squiggly line - There is only one decision rule - But we don't quite generalize the decision-dependent transition probabilities. My conjecture...
A few points here: > I think that the best way to deal with steady-state is simply to start the next iteration using the ending state values from the current...
> which is quite extreme, if we are thinking about modelling infinite horizon. Do you have a way of looking at a model and checking if it meets the necessary...
> Do you have a proof of convergence for this? We didn't include the algorithm in the paper, so we never proved. But: assume the state space is closed and...
You could probably also write a plugin that swapped out different sampling schemes at different iterations. It's not too much work. Here's an example of a trivial plug-in: https://github.com/odow/SDDP.jl/blob/092eeb32171f4ec82d57efa6e0e5650f303ca144/src/plugins/sampling_schemes.jl#L393-L430
You could also write a wrapper around the Historical one to allow you to return the terminated_on_cycle
Totally un-tested, but maybe something like this: ```Julia mutable struct SamplingSchemeSwapper{F}
Did you make any progress on this?