Abstract
A measure for the compatibility of state assignments of independent observers of the same quantum system is proposed, and the possibility of arriving at a unique state assignment by pooling of information is investigated.
Introduction
In quantum mechanics, a density matrix $\rho$ represents an observer’s knowledge or belief about a system. When two observers, Alice and Bob, possess different information about the same system, they naturally assign it different states, $\rho_A$ and $\rho_B$. Traditionally, compatibility was viewed as a binary property: states are compatible if the intersection of their supports is non-empty. However, this definition fails to quantify the degree of agreement or provide a systematic method for information pooling. This paper addresses these gaps by proposing a quantitative measure of compatibility and exploring rules for reaching a consensus state $\rho_{AB}$.
Quantifying Compatibility
The authors adopt a compatibility measure $K$ originally suggested by Alexei Kitaev. This measure identifies the "largest" positive matrix $R$ that is bounded by both observers' density matrices ($\rho_A - R \ge 0$ and $\rho_B - R \ge 0$). The compatibility $K(\rho_A, \rho_B)$ is defined as the maximum trace of such a matrix $R$.
Physically, $K$ represents the maximum probability that a joint measurement would yield the specific outcomes that led Alice and Bob to their respective state assignments. Mathematically, it is upper-bounded by the trace distance: $K(\rho_A, \rho_B) \le 1 - D(\rho_A, \rho_B)$, linking it to standard metrics of state distinguishability.
Information Pooling and SDP
A key challenge in quantum information is that a density matrix does not record the full history of how information was acquired. Consequently, $\rho_A$ and $\rho_B$ alone may not uniquely determine a pooled state $\rho_{AB}$ without making assumptions about the underlying measurement process. The authors demonstrate that a natural pooling rule emerges when assuming information is gathered via positive measurement operators: $\rho_{AB} = R / \text{Tr}(R)$, where $R$ is the optimizer for $K$.
To handle multiple observers, the compatibility measure is formulated as a Semidefinite Program (SDP). This allows for efficient calculation even as the number of observers increases. The authors also establish a hierarchy among different compatibility criteria, proving that $K_{PP}(\mathcal{P}) \ge K_{BFM}(\mathcal{P}) \ge K_{ES}(\mathcal{P})$, where "Equal Support" (ES) is the most restrictive and "Post-Peierls" (PP) is the most permissive.
Results for Qubit Systems
The proposed measure is applied to single-qubit systems to illustrate how compatibility varies with the relative orientation of measurement bases. For instance, when comparing states derived from measurements along different axes on the Bloch sphere, the compatibility score $K_{A,B}$ fluctuates as a function of the angle $\Theta$ between these axes.
As shown in the figures above, the compatibility measure provides a continuous scale for assessing how well different state assignments align. The variation in $K_{A,B}$ across different ranges of $\Theta$ highlights how the degree of compatibility depends on the specific measurement configurations and the resulting state assignments, moving beyond simple binary classifications.
Discussion and Future Work
While this framework provides a robust quantitative approach to compatibility, it currently assumes that all observers share the same initial "prior" (typically the maximally mixed state). Future research is needed to address consensus-building when observers start with different biases. Additionally, while the pooling rule for positive measurement operators is well-defined, alternative information-gathering processes may necessitate different rules. Expanding the SDP formulation to include other notions of compatibility, such as "Weak" compatibility, remains an open mathematical challenge.
Summary updated