Representing systems computationally.
WHAT A MODEL IS
A simplified representation of a system, built to answer a question.
WHAT SIMPLIFICATION MEANS
Every model omits things. The question determines what may be omitted.
WHAT TYPES EXIST
Continuous, described by equations of change Discrete event, where state changes at points in time Agent-based, where behaviour emerges from individuals Statistical, sampling from distributions
WHAT DISCRETE EVENT SIMULATION SUITS
Queues, logistics, manufacturing and service systems.
WHAT AGENT-BASED MODELLING SUITS
Systems where individual behaviour produces collective patterns.
WHAT MONTE CARLO METHODS DO
Estimate outcomes by sampling many random scenarios.
WHAT THEY SUIT
Problems where uncertainty dominates and analysis is intractable.
WHAT VALIDATION MEANS
Establishing that the model represents reality adequately for the question.
WHAT VERIFICATION MEANS
Establishing that it was implemented correctly.
WHY THAT DISTINCTION MATTERS
A correctly implemented wrong model is confidently wrong.
WHAT TO DO ALWAYS
State assumptions explicitly Test sensitivity to them Report uncertainty alongside results
WHAT TO NEVER DO
Present simulation output as prediction without stating its limits.
