Quantum Measurement
Quantum measurement returns a random outcome with probability given by the Born rule and updates the state to the normalized projection associated with that outcome.
The measurement postulate, in its projective form, states the following: a measurement is described by a set of orthogonal projectors $\{P_m\}$ with $\sum_m P_m=I$. If the state before the measurement is $|\psi\rangle$, the probability of outcome $m$ is $$p(m)=\langle\psi|P_m|\psi\rangle,$$ and the state after obtaining $m$ is $P_m|\psi\rangle/\sqrt{p(m)}$. In the computational basis $P_x=|x\rangle\langle x|$, which gives the familiar Born rule $p(x)=|\langle x|\psi\rangle|^2$.
Three important consequences. First, measurement is irreversible: phase and amplitude information is lost, and it cannot be represented by a unitary matrix on the system alone. Second, a measurement gives a single sample; to estimate probabilities we repeat the preparation and measurement over a number of shots, and the statistical error is of order $1/\sqrt{N}$. Third, measuring part of an entangled system changes the conditional state of the other part, but it does not change the unconditional distribution of its outcomes, so it transmits no signal.
On hardware and in Qiskit, measurement is usually done in the computational basis, i.e. the $Z$ basis. To measure in another basis we rotate the state before measuring: for the $X$ basis we apply $H$ and then measure, and for the $Y$ basis we apply $S^\dagger$, then $H$, then measure. More generally, measuring a Hermitian observable $O=\sum_\lambda\lambda P_\lambda$ gives the eigenvalue $\lambda$ with probability $\langle\psi|P_\lambda|\psi\rangle$, and the expectation value is $\langle O\rangle=\langle\psi|O|\psi\rangle$.
Mid-circuit measurement followed by classical conditions, as in teleportation, makes each shot a different trajectory; the simulator handles this by simulating each shot separately. There are also generalised measurements (POVMs) that can describe imperfect measurement, and readout errors on real devices sometimes make the recorded outcome differ from the actual one.
It is worth distinguishing measurement from "looking" at the state vector in the simulator: displaying the amplitudes changes nothing because it is a computation on a classical computer, while an actual measurement on a quantum device changes the state and returns only one outcome per shot.
Mathematical formula
Worked example
Measure only qubit $q_0$ of the Bell state $|\Phi^+\rangle=\tfrac{1}{\sqrt2}(|00\rangle+|11\rangle)$.
The projector for outcome 0 on $q_0$ is $P_0=I\otimes|0\rangle\langle0|$ (qubit $q_0$ is the right-hand factor in the order $q_1q_0$). $P_0|\Phi^+\rangle=\tfrac{1}{\sqrt2}|00\rangle$, so $p(0)=\tfrac12$ and the state after measurement is $|00\rangle$. Likewise the outcome 1 has probability $\tfrac12$ and the state collapses to $|11\rangle$. After that, measuring $q_1$ gives the same outcome with certainty.
An example of the basis: $|+\rangle$ in the $Z$ basis gives 0 or 1 with probability $\tfrac12$; after applying $H$ (i.e. measuring in the $X$ basis) it gives 0 with probability 1.
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). Basics of Quantum Information (course). IBM Quantum Learning quantum.cloud.ibm.com
- Microsoft (2025). Pauli Measurements in Quantum Computing. Azure Quantum documentation learn.microsoft.com
