A property of a stable, causal linear time‑invariant system whose inverse is also causal and stable; equivalently, all zeros of the system’s transfer function lie inside the stability region (inside the unit circle for discrete‑time z‑domain, in the left half‑plane for continuous‑time s‑domain). Such systems attain the smallest possible phase (or group delay) consistent with a given magnitude response, with any other system sharing that magnitude differing by an all‑pass factor.

Principle

Within the class of causal, stable LTI systems that share the same magnitude response, the minimum‑phase member minimizes phase and group delay; algebraically, this corresponds to zero locations inside the stability region so that no unstable or non‑causal inverse is required.

Demonstration

Illustrative scenario: designing a digital equalizer to match a target magnitude response—selecting a minimum‑phase filter with the target magnitude yields the smallest achievable phase distortion and the most compact causal inverse; converting a non‑minimum‑phase filter to minimum phase requires altering zero locations and hence cannot preserve both magnitude and causal phase unless an all‑pass compensator is introduced.

Misapplication

Assuming that any system can be converted to minimum phase without changing its magnitude response or without introducing noncausal/instable elements is incorrect; the semantic error is conflating magnitude preservation with phase neutrality and ignoring the necessary all‑pass factor when zeros lie outside the stability region.

Consequence

When the minimum‑phase assumption holds, phase can be recovered from magnitude (up to at most additive constants) and causal inversion is feasible; when it does not hold, inversion may be unstable or noncausal and phase cannot be deduced from magnitude alone without specifying additional all‑pass structure.

Reversal

If the system is noncausal, unstable, or multichannel with constraints on phase (e.g., prescribed linear phase), the minimum‑phase principle either does not apply or must be modified; in multichannel or structured designs, a single‑channel minimum‑phase transformation may conflict with other constraints.

Boundary

Applies to single‑input single‑output, stable, causal LTI systems where pole/zero locations are the relevant parametrization; excludes systems that are nonlinear, time‑varying, noninvertible, or those for which phase constraints (e.g., exact linear phase) are required instead of minimal phase.

Semantic Tension

Minimum phase (phase minimization and causal invertibility) ↔ Linear‑phase or symmetric‑impulse‑response designs (which trade phase minimality for phase linearity): designers must choose between minimal group delay and other desirable phase characteristics.

Synthesis

Minimum‑phase property ties algebraic zero locations to operational consequences: if zeros lie inside the stability region the system is causally invertible and has the least phase for its magnitude; without that property, magnitude does not determine phase and inversion may be impossible without instability or noncausality.