Multi-stage Optimization

Coming soon

Multi-stage optimization is planned but not yet implemented. This page describes the concept for context.

What is multi-stage optimization?

In stochastic optimization, you make decisions considering multiple scenarios, but all decisions are made at once (two-stage: decide now, then observe uncertainty).

Multi-stage optimization extends this to a sequence of decision points over time. At each stage, you observe new information and can adjust your decisions accordingly. This models real-world operations more closely -- for example, you might bid in the day-ahead market today, then adjust in the intraday market tomorrow once you have better forecasts.

How it relates to stage_fixed

The stage_fixed parameter on EnergyMarket is a first step toward multi-stage behavior. Markets with stage_fixed=True represent earlier decision stages where you commit before uncertainty is resolved. Markets without it represent later stages where you can react to the realized scenario.

When full multi-stage support arrives, you'll be able to define an explicit sequence of stages, each with its own information set and decision variables.

Stay tuned

Follow the GitHub repository for updates on this feature.