Exercise 8.2¶
Consider a quantum measurement \(\{M_m\}\) with outcome labeled by \(m\), where \(M_m\) is not a unitary operator. For a pure state \(|\phi\rangle\), the probability of measuring outcome \(m\) is \(p(m) = \langle \phi|M^{\dagger}_mM_m|\phi\rangle\). Here, the expectation value of operator \(M^{\dagger}_m M_m\) can be written as
where \(\rho_{\phi} = |\phi\rangle\langle \phi|\). The post-measurement state \(|\phi'\rangle\) is given by
which can be also written in the form of density matrix,
Probability of measurement¶
Suppose density matrix \(\rho\) describes an ensemble of pure states \(\{q(j), |\psi_j\rangle\}\),
where \(\rho_j = |\psi_j\rangle\psi_j|\). The probability of outcome \(m\) after measuring the ensemble is given by
This is the product of probability of “getting pure state \(|\psi_j\rangle\)”, or \(q_j\), and the probability of “measuring \(m\) for \(|\psi_j\rangle\)”, or \(p(m|j)\). Here, \(p(m|j)\) is obtained from eq. (1). One can also rewrite eq. (5) as
where \(\mathcal{E}(\rho) = M_m\rho M^{\dagger}_m\).
Post-measurement state¶
After measurement, the new ensemble is described by a new probability distribution and post-measurement states. From eq. (3), the normalized post-measurement state \(\rho'_j\) for each \(|\psi_j\rangle\) is given by
The probability of getting \(\rho'_j\) is \(q(j)p(m|j)\). However, \(\{q(j)p(m|j)\}\) does not give a valid probability distribution since \(\sum_j q(j)p(m|j) = P(m) \) is NOT EQUAL TO ONE in general. As a result, we re-normalize and get a new probability distribution for post-measurement ensemble as \(\{q(j)p(m|j) / P(m)\}\). In this context, the density matrix that describes post-measurement ensemble is given by
which is equivalent with \(\mathcal{E}_m (\rho) / {\rm tr}\left(\mathcal{E}(\rho) \right)\).