Phase Kickback
Phase Kickback
Phase kickbackPhase Kickback: Phase kickback is a phenomenon in which an eigenvalue phase moves from the target qubit to the control qubit of a controlled gate; it is a key mechanism in Deutsch–Jozsa… is a phenomenon in which an eigenvalue phase moves from the target qubitQubit: The qubit is the unit of quantum information: a two-level system whose state is a unit vector α0⟩+β1⟩ in ℂ^2. to the control qubit of a controlled gate; it is a key mechanism in Deutsch–Jozsa and phase estimation.
If the target is in an eigenvector $|u\rangle$ of $U$ with eigenvalue $e^{i\phi}$, the controlled gate gives $$CU\,(\alpha|0\rangle+\beta|1\rangle)|u\rangle=(\alpha|0\rangle+\beta e^{i\phi}|1\rangle)|u\rangle.$$ The target has not changed, and the phase has "kicked back" onto the control qubit as a relative phase. This contradicts the classical intuition that the control is unaffected.
The simplest example: CNOT with the target in $|-\rangle$, an eigenvector of $X$ with eigenvalue $-1$. So $\mathrm{CNOT}\,|{+}\rangle|{-}\rangle=|{-}\rangle|{-}\rangle$: the control has moved from $|+\rangle$ to $|-\rangle$. In the phase oracle $U_f|x\rangle|-\rangle=(-1)^{f(x)}|x\rangle|-\rangle$, computing $f$ turns into a phase on the input register, which is what the Deutsch, Deutsch–Jozsa, Bernstein–Vazirani and Grover algorithms exploit.
In quantum phase estimationQuantum Phase Estimation: Quantum phase estimation estimates the phase φ in Uu⟩=e^2π iφu⟩ to t bits of precision using t counting qubits, controlled U^2^j gates and the inverse Fourier transform., $CU^{2^k}$ is applied so that the phase $2^k\phi$ kicks back onto qubit $k$, and then the inverse Fourier transform extracts $\phi$. The Hadamard testHadamard Test: The Hadamard test estimates Re⟨ψUψ⟩ (or the imaginary part) with an ancilla qubit and a controlled U gate: P(0)=(1+Re⟨ψUψ⟩)/(2). relies on the same mechanism to estimate $\mathrm{Re}\langle\psi|U|\psi\rangle$.
The lesson is that the notions of "control" and "target" depend on the basis: in the $X$ basis, CNOT swaps the roles of control and target, since $(H\otimes H)\,\mathrm{CNOT}_{0\to1}\,(H\otimes H)=\mathrm{CNOT}_{1\to0}$.
Mathematical formulation
Worked example
With $q_0$ as the control in $|+\rangle$ and $q_1$ as the target in $|1\rangle$, apply $CP(\lambda)$ (the controlled version is symmetric). $|1\rangle$ is an eigenvector of $P(\lambda)$ with eigenvalue $e^{i\lambda}$, so the control becomes $\tfrac{1}{\sqrt2}(|0\rangle+e^{i\lambda}|1\rangle)$. With $\lambda=\pi$ it becomes $|-\rangle$, and measuring it after $H$ gives 1 with certainty.
Try it yourself: Deutsch–Jozsa Algorithm
Test your understanding
The quiz needs JavaScript.
Related topics
References
- Michael A. Nielsen; Isaac L. Chuang (2010). Quantum Computation and Quantum Information (10th Anniversary Edition). Cambridge University Press doi:10.1017/CBO9780511976667The standard reference for quantum computing; the Crossref record gives 2012 for the Cambridge Core online edition.
- John Watrous / IBM Quantum (2025). Fundamentals of Quantum Algorithms (course). IBM Quantum Learning quantum.cloud.ibm.com
