Dynamic & RL Guide

Dynamic & RL Guide

A live demo of dynamic LCO with stochastic Poisson arrivals and rolling-horizon re-optimization over a 14-day horizon.

What is the Dynamic LCO & RL Demo?

This page simulates bookings arriving as a stochastic Poisson process over a rolling 14-day horizon (H=14). Each day the demo re-optimizes the lexicographic tiers, while RL-tunable intra-tier parameters (overbooking robustness, housekeeping penalty, energy weight) shape the outcomes. It illustrates how the static LCO model extends to a dynamic, online setting where floors Z*ₖ are preserved at every step through safety shielding.

A vertical flow showing the five stages that repeat each day:

  1. Booking Arrivals (Poisson)
  2. Tier L1 Feasibility Gate
  3. State Update
  4. Nightly LCO Re-optimization
  5. RL Parameter Update

While running, the active stage highlights in teal to show the pipeline progressing day by day.

Typical workflow

  1. Set the three intra-tier parameter sliders to your starting values.
  2. Click Start Simulation to begin the 14-day rollout.
  3. Watch the flow diagram advance and the live metrics update.
  4. Adjust sliders mid-run to see how each parameter shifts the operational metrics.
  5. Use Reset to restart with new parameters.
Dynamic & RL • Stochastic arrivals with rolling-horizon re-optimization • Valamontes, 2026