agent-based modeling

**Agent-Based Modeling (ABM)** for semiconductor manufacturing is a **bottom-up simulation paradigm where individual entities (agents) follow local rules** — with system-level behavior emerging from the interactions between thousands of agents representing wafers, tools, operators, and controllers. **ABM vs. Traditional Simulation** - **Bottom-Up**: Define rules for individual agents — system behavior emerges (vs. top-down equations). - **Heterogeneity**: Each agent can have unique properties (different recipes, priorities, tool states). - **Adaptation**: Agents can learn and adapt their behavior based on experience. - **Spatial**: Agents can be embedded in physical space (fab layout, AMHS tracks). **Why It Matters** - **Complex Interactions**: Captures tool-lot-operator interactions that analytical models cannot represent. - **Decentralized Decision Making**: Models real fab operations where decisions are made locally, not centrally. - **Disruption Modeling**: Naturally handles disruptions (tool failures, hot lots) through agent-level responses. **ABM** is **the microscopic view of fab dynamics** — simulating every individual entity's behavior to understand how complex factory patterns emerge.