A time‑domain mathematical representation of a dynamical system using a finite vector of state variables x(t) and first‑order differential (continuous) or difference (discrete) equations ẋ = Ax + Bu, y = Cx + Du (or their nonlinear generalizations), relating internal state evolution to inputs and outputs for analysis, simulation and control design.

Principle

Specifying a minimal‑dimension state vector and matrices (A,B,C,D) provides a complete linear time‑domain description of finite‑dimensional LTI dynamics observable at the outputs and controllable by the inputs; the same framework generalizes to nonlinear or time‑varying systems by replacing matrices with state‑dependent mappings.

Demonstration

Illustrative scenario → A control engineer models a DC motor's dynamics with states [rotor speed; armature current] (situation). They write linearized state equations around the operating point, derive A and B matrices (recognition), design a state‑feedback controller that places closed‑loop poles to meet settling‑time requirements (action), and then simulate time responses showing the motor reaches target speed within the predicted time (consequence).

Misapplication

Assuming state variables and system matrices are unique without specifying the state basis or coordinate transformation: two equivalent state‑space models related by a similarity transform represent the same input‑output behavior but have different A,B,C,D entries. The semantic error is conflating parameter values with physical observables instead of focusing on invariant properties (transfer function, eigenvalues).

Consequence

State‑space models enable time‑domain simulation, observer and controller design, multi‑input multi‑output (MIMO) analysis and rigorous stability tests; misidentifying states, linearizing inapplicably, or using low‑order approximations without validation can produce controllers that fail in practice or incorrect predictions of transient performance.

Reversal

For systems inherently infinite‑dimensional (distributed parameter systems) or when only steady‑state frequency response is needed, state‑space either requires approximation (model reduction, spatial discretization) or is unnecessary; for strongly nonlinear behaviors the linear state‑space form must be replaced by a nonlinear state‑space model or other methods appropriate to the regime.

Boundary

Clearly within: a finite‑order LTI mechanical or electrical plant where state variables (positions, currents) are identifiable and a linear approximation applies. Boundary case: a weakly time‑varying or lightly nonlinear system where linearization yields useful design guidance but demands validation. Clearly outside: an exact description of an electromagnetic field distribution in continuous space (an infinite‑dimensional PDE) without reduction.

Semantic Tension

Model order and interpretability ↔ Computational tractability: higher‑order state models can capture more physics and improve fidelity but increase design complexity and computational cost; practitioners balance minimality against adequacy for control or simulation tasks.

Synthesis

State‑space modelling unifies dynamical representation for simulation and control by exposing internal states that govern transient behavior; the representation is powerful for design but requires explicit choices about state definitions, linearization validity and model order that determine whether predictions transfer to the real system.