For a unitary U and its eigenvector |ψ⟩, applying U multiplies the state by an eigenvalue λ. Because unitary eigenvalues have magnitude 1, write λ = e^(2πiθ), where 0 ≤ θ < 1. The real number θ is the eigenphase. Phase-estimation tasks receive U and |ψ⟩ as operations for application and state preparation, then seek θ rather than a matrix-derived eigenvalue. section 5.3
Z gate
Z has eigenvalue 1 on |0⟩ and −1 on |1⟩, giving eigenphases 0 and 0.5. Thus e^(2πi·0) = 1 and e^(2πi·0.5) = −1. section 5.3
Table 5.2 Unitaries, their eigenvectors, and corresponding eigenphases with different values of phase
| Unitary \(U\) | Eigenvector \(\ket{\phi}\) | Eigenvalue | Eigenphase (decimal) | Eigenphase (binary) |
| \(Z = \begin{pmatrix} 1 & 0 \\ 0 & -1 \end{pmatrix}\) | \(\ket{0}\) | \(1\) | \(0.0\) | \(0.0\) |
| \(\ket{1}\) | \(-1\) | \(0.5\) | \(0.1\) | |
| \(INC = \begin{pmatrix} 0 & 0 & 0 & 1 \\ 1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 1 & 0 \end{pmatrix}\) | \(\frac12(\ket{0}+\ket{1}+\ket{2}+\ket{3})\) | \(1\) | \(0.0\) | \(0.00\) |
| \(\frac12(\ket{0}-\ket{1}+\ket{2}-\ket{3})\) | \(-1\) | \(0.5\) | \(0.10\) | |
| \(\frac12(\ket{0}-i\ket{1}-\ket{2}+i\ket{3})\) | \(i\) | \(0.25\) | \(0.01\) | |
| \(\frac12(\ket{0}+i\ket{1}-\ket{2}-i\ket{3})\) | \(-i\) | \(0.75\) | \(0.11\) | |
| \(T = \begin{pmatrix} 1 & 0 \\ 0 & e^{i \pi /4} \end{pmatrix}\) | \(\ket{0}\) | \(1\) | \(0.0\) | \(0.000\) |
| \(\ket{1}\) | \(e^{i \pi /4}\) | \(0.125\) | \(0.001\) |
Compare Z, INC, and T eigenvectors with their unit-circle eigenvalues and decimal and binary eigenphase representations. section 5.3
Controlled-U exposes an eigenphase through phase kickback: with a control in |+⟩ and target |ψ⟩, the state becomes (|0⟩ + e^(2πiθ)|1⟩)⊗|ψ⟩/√2. The target remains unchanged and unentangled, while the control carries a measurable relative phase. section 5.3.1
Phase estimation recovers the encoded phase in several ways:
- Iterative estimation repeats a fixed controlled-U circuit, estimates outcome probabilities, and derives the eigenphase from measurement statistics. section 5.3.2
- Adaptive estimation learns binary digits from least to most significant, changing later circuits according to earlier results. section 5.3.3
- Quantum phase estimation uses one control qubit per precision bit, then applies the inverse quantum Fourier transform to convert phase encoding into measurable amplitude encoding. section 5.3.5
Implementations prepare the supplied eigenstate, apply controlled unitaries and decoding gates, measure control qubits, and return a classical eigenphase. Tests can use Z and INC eigenvectors, compare estimates against calculated phases, and repeat runs where required. section 5.3.6 section 5.3.7
