Abstract
This paper formalizes the "ability to distinguish" between quantum states as a quantifiable physical resource. By introducing the concept of "quantum boxes"—ordered pairs of quantum states—and defining "bits of asymmetric distinguishability" as a standard currency, we establish a rigorous Resource Theory of Asymmetric Distinguishability (RTAD). We demonstrate that in the asymptotic limit, this resource theory is perfectly reversible, with the quantum relative entropy serving as the fundamental rate of conversion. Furthermore, we characterize one-shot distillation and cost using min- and max-relative entropies, respectively, and provide efficient computational methods for these transformations.
Introduction
In classical and quantum statistics, the fundamental task of hypothesis testing involves distinguishing between a null hypothesis (state $\rho$) and an alternative hypothesis (state $\sigma$). While the statistical limits of this task have been studied for decades, the notion of distinguishability as a manipulable resource—akin to energy in thermodynamics or entanglement in quantum information—has only recently been formalized.
This work develops a systematic Resource Theory of Asymmetric Distinguishability. We focus specifically on the asymmetric case: where the priority is to minimize the probability of a Type II error (falsely accepting the null) while keeping the Type I error (falsely rejecting the null) below a fixed threshold. This framework allows us to treat the "ability to distinguish" as a currency that can be traded, distilled, and converted.
The Resource Theory Framework
The RTAD is built upon three primary pillars:
- Quantum Boxes: The basic unit of the theory is a "quantum box," denoted as $(\rho, \sigma)$, representing an ordered pair of quantum states on the same Hilbert space.
- Free Operations: We define the set of free operations as any quantum channel (a completely positive, trace-preserving map). A transformation from box $(\rho, \sigma)$ to box $(\tau, \omega)$ is possible if there exists a channel $\mathcal{N}$ such that $\mathcal{N}(\rho) = \tau$ and $\mathcal{N}(\sigma) = \omega$. If such a channel exists, the second box is considered "easier" to distinguish than the first.
- Resource Currency: To quantify distinguishability, we introduce the "bit of asymmetric distinguishability." This is represented by a specific reference box $(|0\rangle\langle 0|, \pi)$, where $\pi$ is the maximally mixed state.
One-Shot Characterizations
A central contribution of this work is the operational characterization of abstract entropy measures in the "one-shot" regime, where only a single copy of a quantum box is available.
- Distillable Distinguishability: The maximum number of bits of distinguishability that can be extracted from a single box $(\rho, \sigma)$ is exactly the min-relative entropy, $D_{\min}(\rho||\sigma)$.
- Distinguishability Cost: The minimum number of bits required to create a copy of $(\rho, \sigma)$ from the standard currency is the max-relative entropy, $D_{\max}(\rho||\sigma)$.
When we allow for a small approximation error $\epsilon$ (measured via trace distance), these quantities generalize to the smooth min- and max-relative entropies. This provides a concrete operational meaning to these variables, which are standard tools in quantum cryptography and hypothesis testing.
Asymptotic Reversibility and the Fundamental Rate
One of the most striking findings of this research is that in the limit of many independent and identically distributed (i.i.d.) copies of a state ($n \to \infty$), the gap between the cost of creation and the yield of distillation vanishes. This implies that the resource theory is asymptotically reversible.
In this limit, you do not lose any "distinguishability" when converting back and forth between different states. The ultimate exchange rate between any two pairs of states is governed by the quantum relative entropy, $D(\rho||\sigma)$. Specifically, the optimal rate $R$ at which one can convert copies of $(\rho, \sigma)$ into copies of $(\tau, \omega)$ is given by the ratio of their respective quantum relative entropies:
$$R = \frac{D(\rho||\sigma)}{D(\tau||\omega)}$$
Computational Efficiency
Beyond theoretical proofs, the authors demonstrate that the approximate box transformation problem is a Semi-Definite Program (SDP). This means that determining whether a transformation is possible within a certain error tolerance can be solved in polynomial time relative to the dimension of the quantum states, making the theory practically applicable for numerical analysis.
Limitations and Future Work
While the RTAD provides a robust framework, several avenues for future research remain:
- Metric Sensitivity: The results are somewhat dependent on using trace distance as the measure of approximation. Alternative measures like infidelity might yield different second-order terms in asymptotic expansions.
- Channel Resource Theory: Extending these results from quantum states to quantum channels (processes) is a natural next step.
- Multichotomies: The current theory is restricted to pairs of states. Expanding the framework to handle distinguishability between three or more states remains an open challenge.
Summary updated