We consider the information flow on a system observable
X corresponding to a positive-operator-valued measure under a quantum measurement process
Y described by a completely positive instrument from the viewpoint of the relative entropy. We establish a sufficient condition for the relative-entropy conservation law which states that the average decrease in the relative entropy of the system observable
X equals the relative entropy of the measurement outcome of
Y, i.e., the information gain due to measurement. This sufficient condition is interpreted as an assumption of classicality in the sense that there exists a sufficient statistic in a joint successive measurement of
Y followed by
X such that the probability distribution of the statistic coincides with that of a single measurement of
X for the premeasurement state. We show that in the case when
X is a discrete projection-valued measure and
Y is discrete, the classicality condition is equivalent to the relative-entropy conservation for arbitrary states. The general theory on the relative-entropy conservation is applied to typical quantum measurement models, namely, quantum nondemolition measurement, destructive sharp measurements on two-level systems, a photon counting, a quantum counting, homodyne and heterodyne measurements. These examples except for the nondemolition and photon-counting measurements do not satisfy the known Shannon-entropy conservation law proposed by Ban [M. Ban, J. Phys. A: Math. Gen. 32, 1643 (1999)], implying that our approach based on the relative entropy is applicable to a wider class of quantum measurements.
When the von Neumann entropy (VNE) of a system increases due to measurements, certain information is lost, some of which may be recoverable. We define information retrievability (IR) and information loss (IL) as functions of the density matrix through VNE to illustrate the relationship between gain and loss. We demonstrate that when a pure, unbiased m-qubit state collapses into a maximally mixed state, it experiences the maximal loss of information and the highest gain in entropy, equivalent to the m-bit classical Shannon entropy. We analyze the VNE, IR, and IL of single qubits, entangled photon pairs in Bell tests, three-qubit systems in quantum teleportation, multiple-qubit systems of GHZ and W states, and two-qubit Werner mixed states, emphasizing their IL dependence on parameters such as polarization bias and qubit count. Data exchange between two observers in Bell tests can recover some of the lost quantum information and eliminate the associated quantum entropy, even years later. The need to recover knowledge explains why no spooky action occurs at a distance. We show that measuring the Bell, GHZ, and marginally entangled Werner states yields the same minimum entropy gain (ln2) and equal minimal information loss (50 percent).