discreteevent

Discrete-event describes systems in which state changes occur at discrete points in time as a result of events, rather than continuously. The term is used for both discrete-event systems (DES) in control theory and operations research, and discrete-event simulation (DES), a modeling technique. In a discrete-event model, time advances by jumping from one event to the next, and the system state remains constant between events.

Key concepts include entities or tokens that move through a model, resources that can process entities, and queues that buffer demand. Each event has a time stamp and an effect on the state, such as an arrival, a service completion, or a departure. An event calendar or priority queue schedules future events; the simulation loop repeatedly removes the next event, updates the clock, processes the event's effects, and possibly generates new events.

Discrete-event models are common for systems where changes occur irregularly and involve handling competition for limited resources, such as customers in a bank, parts in a factory, packets in a network, or patients in a hospital. Randomness is often introduced through probability distributions for inter-arrival times, service times, or routing decisions.

Applications include manufacturing operations, logistics and supply chains, telecommunications and data networks, healthcare, and service systems. Outputs typically focus on performance metrics such as throughput, utilization, queue lengths, waiting times, and cycle times. Modelers often validate and calibrate models against real data and use experiments or optimization to evaluate policy alternatives.

Challenges include ensuring model fidelity, dealing with initialization bias and warm-up periods, and choosing appropriate variance reduction and statistical techniques to interpret results.