Search and Belief-Space Planning › Key Parameters
Michael BrenndoerferJuly 9, 20261 min read
World Models HandbookAll chapters Part 7 of 12Planning and Agency
Search and Belief-Space Planning › Key ParametersOne section · Sections in this chapter8 / 13 ↓
Sections in this chapter13
Reading tools Definitions & inline links
Choose your expertise level to adjust how many terms are explained. Beginners see more tooltips, experts see fewer to maintain reading flow. Hover over underlined terms for instant definitions.
Article links
Make inline references clickable
The key parameters for the hidden-state maze experiment are:
- initial belief: The probability distribution over cells after the first observation. Its support and weights determine which latent starts the planner must account for.
- n_sims: Simulations per POMCP planning call. This controls how thoroughly the fixed current belief is searched; the real belief narrows only after an action and observation update.
- max_depth: Maximum simulated horizon from the current history node. It bounds cost per simulation and should be long enough to expose the decision-relevant consequences being compared.
- belief filter: The exact Bayes update used between planning calls. A biased or approximate filter propagates error directly into the next plan.