Regularization for deterministic first-stage

odow / SDDP.jl

A JuMP extension for Stochastic Dual Dynamic Programming

https://sddp.dev

Other

304 stars 62 forks source link

Regularization for deterministic first-stage #623

Closed odow closed 1 year ago

odow commented 1 year ago

Discussed with someone that capacity expansion models with a deterministic first stage would benefit from some regularization of the outgoing state variables in the first-stage.

odow commented 1 year ago

Very co-indicentally, There's also https://pubsonline.informs.org/doi/10.1287/ijoc.2021.0255 that was published a couple of days ago.

They used SDDP.jl so it should be easy-ish to integrate. cc @cfcarnabiitkgp

odow commented 1 year ago

@jarandh has a model that will need this so I may work on it. A couple of notes while it's fresh:

L1 or L2 norm?
This is needed only on the forward pass. Can we use a new AbstractForwardPass for this?
All states, or selected outgoing states?