Why Gates Are Unitary

qc/math

A valid quantum operation must preserve the norm of the state vector — probabilities have to sum to 1 both before and after, so for every valid input state . Key insight: this requirement is exactly equivalent to unitarity. Since must equal for all , it forces — the definition of a unitary matrix. Norm-preservation isn’t a separate property gates happen to have; it’s the reason the unitary requirement exists in the first place.

Why this rules some matrices out

Not every matrix is a valid gate. A plain scaling matrix breaks norm preservation immediately. More importantly: a Hamiltonian itself is generally not unitary (see Hamiltonians and Encoding for Quantum Circuits) — which is exactly why you can’t apply directly as a gate, only , which is unitary by construction.

Unitarity means reversibility

Every unitary has an inverse, , so every quantum gate (aside from measurement, which is not unitary — see Measurement and Collapse) is reversible. This is a sharp contrast with classical logic: an AND gate has no inverse (you can’t recover both inputs from the output), but every gate in this vault’s gate set can always be undone by applying its adjoint.

Self-Check

  • Could you derive starting from the requirement that preserves a state’s norm?
  • Why isn’t a Hamiltonian itself a valid gate, even though it’s a perfectly good Hermitian matrix?
  • Why does unitarity imply every quantum gate is reversible, and what’s the classical contrast?
  • What would it take to deliberately construct a matrix that is not unitary, and how would you verify computationally (e.g. via an is_unitary()-style check) that it fails ?