Kirkwood-Dirac representations of quantum states are increasingly finding use in many areas within quantum theory. Usually, representations of this sort are only applied to provide a representation of quantum states (as complex functions over some set). We show how standard Kirkwood-Dirac representations can be extended to a fully compositional representation of all of quantum theory (including channels, measurements, and so on), and prove that this extension satisfies the essential features of functoriality (namely, that the representation commutes with composition of channels), linearity, and quasistochasticity. Interestingly, the representation of a positive operator-valued measure (POVM) element is uniquely picked out to be the collection of weak values for it relative to the bases defining the representation. We then prove that if one can find any Kirkwood-Dirac representation that is everywhere real and non-negative for a given experimental scenario or fragment of quantum theory, then the scenario or fragment is consistent with the principle of generalized noncontextuality, a key notion of classicality in quantum foundations. We also show that the converse does not hold: even if one verifies that all Kirkwood-Dirac representations (as defined herein) of an experiment require negativity or imaginarity, one cannot generally conclude that the experiment witnesses contextuality.
Kirkwood-Dirac representations of quantum states are increasingly finding use in many areas within quantum theory. Usually, representations of this sort are only applied to provide a representation of quantum states (as complex functions over some set). We show how standard Kirkwood-Dirac representations can be extended to a fully compositional representation of all of quantum theory (including channels, measurements, and so on), and prove that this extension satisfies the essential features of functoriality (namely, that the representation commutes with composition of channels), linearity, and quasistochasticity. Interestingly, the representation of a positive operator-valued measure (POVM) element is uniquely picked out to be the collection of weak values for it relative to the bases defining the representation. We then prove that if one can find any Kirkwood-Dirac representation that is everywhere real and non-negative for a given experimental scenario or fragment of quantum theory, then the scenario or fragment is consistent with the principle of generalized noncontextuality, a key notion of classicality in quantum foundations. We also show that the converse does not hold: even if one verifies that all Kirkwood-Dirac representations (as defined herein) of an experiment require negativity or imaginarity, one cannot generally conclude that the experiment witnesses contextuality.
Quantum theory is fundamentally different from its classical counterpart. This difference has led to striking new insights in the foundations of physics and the theories of computation and information. Nevertheless, characterising the difference between classical and quantum theory has proven to be a difficult problem. In this thesis, I present my progress towards a general solution for this open problem. My main tool is the Kirkwood-Dirac (KD) quasiprobability distribution. The KD distribution provides an alternative description of quantum states, operations and measurements in terms of quasiprobability distributions. The KD quasiprobability distribution obeys all properties of probability distributions except that its values may, in general, lie in the complex unit disc instead of the interval [0,1]. For some experiments, however, all associated KD distributions are genuine probability distributions. For such `KD-positive’ states, operations and measurements, the KD distribution enables the construction of a classical model. The absence of such a construction can provide a natural criterion for the definition of the quantum-classical boundary. In the first part of my thesis, I present my results on the mathematical structure of the KD distribution. I summarise previous literature, and prove several new results, including a proof that the set of KD-positive states is almost always minimal, which was published in Physical Review A; novel bounds on the entries of the KD distribution and KD nonpositivity, a measure of how far a KD distribution is from being positive; and the construction of exotic states, mixed KD-positive states that cannot be written as the convex combination of pure KD-positive states, for Hilbert spaces of any odd dimension. These results lay the foundation for the objective of the remainder of this thesis: the application of the KD distribution in quantum foundations, computation and information. In the foundational setting, I study contextuality, a notion of nonclassicality often used in quantum foundations. An experiment is contextual if it is impossible to describe its indistinguishable procedures identically classically. I design an experiment that is contextual if and only if the input state of the experiment is not KD-positive. By analysing this connection with respect to the aforementioned exotic states, I obtain a quantum protocol that allows an experimenter to verify contextuality using a noncontextual experiment, thereby uncovering a previously unknown feature of contextuality. This section of the thesis is based on my publication in Physical Review A `Verification of contextuality by noncontextual experiments’ KDContextualityPaper. In the computational setting, I study the Delfosse-Guerin-Bian-Raussendorf (DGBR) model of quantum computation with a restricted set of input states. For certain sets of input states, this model is computationally as powerful as a universal quantum computer. However, for other sets of input states, this model is efficiently simulable on a classical computer and thus cannot be as powerful as a universal quantum computer, unless quantum computers offer no advantage over their classical counterparts. I prove that if all states in the set of input states are KD-positive, then the DGBR model is classically efficiently simulable. In this setting, some exotic states are bound-magic states, magic states that are nevertheless simulable input states in the DGBR model. I use my previous work on exotic states to construct the first bound-magic states on qubits. Furthermore, I show that KD nonpositivity is a resource monotone for quantum computation, establishing KD nonpositivity as a necessary resource for quantum computational advantage. This section of the thesis is based on my article `Kirkwood-Dirac nonpositivity is a necessary resource for quantum computing’, which has been accepted for publication in Physical Review Letters. In the information-theoretical setting, I study the uncertainty principle. In modern quantum information theory, the uncertainty principle is formulated as an `entropic lower bound’ on the sum of the entropies of the outcome distributions of two observables. I extend the definition of the standard entropic functions to the KD distribution and the Margenau-Hill (MH) distribution, which is the real part of the KD distribution. I introduce the MH Shannon entropy and the KD Rényi entropy, and establish their basic properties. For the MH Shannon entropy, I prove a generalised subadditivity relation, which yields a state-dependent entropic uncertainty relation that strengthens the previous best bound in most regimes and implies the finite-dimensional analogue of the Van Herstraeten--Cerf conjecture. For the KD Rényi entropy, I prove an entropic uncertainty relation strengthening the Maassen--Uffink bound, in which the KD nonpositivity appears as an additive correction. This yields a new interpretation of KD nonpositivity as a measure of quantum uncertainty, as well as a new entropic witness of KD positivity. Applying these results to cryptographic games, I show that KD nonpositivity is the key resource governing the optimal strategy in each case. This section is based on my article `Kirkwood-Dirac nonpositivity is sufficient for quantum uncertainty'. These results demonstrate that the KD distribution provides an effective formulation of quantum theory for studying the quantum-classical boundary in various settings. My work establishes explicit connections between the KD distribution and quantum foundations, computation, and information, opening up these fields to be studied using the extensive KD toolkit.
We uncover features of generalized contextuality by connecting it to the Kirkwood-Dirac (KD) quasiprobability distribution. Quantum states can be represented by KD distributions, which take values in the complex unit disk. Only for “KD-positive” states are the KD distributions joint probability distributions. A KD distribution can be measured by a series of weak and projective measurements. We design such an experiment and show that it is contextual iff the underlying state is not KD positive. We analyze this connection with respect to mixed KD-positive states that cannot be decomposed as convex combinations of pure KD-positive states. Our result is the construction of a noncontextual experiment that enables an experimenter to verify contextuality.
Abstract There are several mathematical formulations of quantum mechanics. The Schrödinger picture expresses quantum states in terms of wavefunctions over, e.g. position or momentum. Alternatively, phase-space formulations represent states with quasi-probability distributions over, e.g. position and momentum. A quasi-probability distribution resembles a probability distribution but may have negative and non-real entries. The most famous quasi-probability distribution, the Wigner function, has played a pivotal role in the development of a continuous-variable quantum theory that has clear analogues of position and momentum. However, the Wigner function is ill-suited for much modern quantum-information research, which is focused on finite-dimensional systems and general observables. Instead, recent years have seen the Kirkwood–Dirac (KD) distribution come to the forefront as a powerful quasi-probability distribution for analysing quantum mechanics. The KD distribution allows tools from statistics and probability theory to be applied to problems in quantum-information processing. A notable difference to the Wigner function is that the KD distribution can represent a quantum state in terms of arbitrary observables. This paper reviews the KD distribution, in three parts. First, we present definitions and basic properties of the KD distribution and its generalisations. Second, we summarise the KD distribution’s extensive usage in the study or development of measurement disturbance; quantum metrology; weak values; direct measurements of quantum states; quantum thermodynamics; quantum scrambling and out-of-time-ordered correlators; and the foundations of quantum mechanics, including Leggett–Garg inequalities, the consistent-histories interpretation and contextuality. We emphasise connections between operational quantum advantages and negative or non-real KD quasi-probabilities. Third, we delve into the KD distribution’s mathematical structure. We summarise the current knowledge regarding the geometry of KD-positive states (the states for which the KD distribution is a classical probability distribution), describe how to witness and quantify KD non-positivity, and outline relationships between KD non-positivity, coherence and observables’ incompatibility.
Abstract As an extension of classical probability distribution, the Kirkwood-Dirac distribution (KDD) was discussed by Kirkwood in 1933 and Dirac 1945, independently. Recently, it has been proved that nonclassical values (negative and non-real values) of the KDD have the ability of outperforming their classical counterparts in quantum computation, quantum measurement and so on. In this work, by dividing quantum states into KD-real (KD-free) and KD-imaginary (KD-resource) ones based on the KDD of a state, we establish a resource theory for KD-imaginarity with respect to a pair of bases (A, B), called the resource theory of Kirkwood-Dirac imaginarity. This theory is different from the resource theory of imaginarity of quantum states with respect to one basis A, where the free states are those that have real density matrices under the basis A.
Kirkwood-Dirac (KD) quasiprobability is a quantum analog of classical phase space probability. It offers an informationally complete representation of quantum state wherein the quantumness associated with quantum noncommutativity manifests in its nonclassical values, i.e., the nonreal and/or negative values of the real part. This naturally raises a question: how does such form of quantumness comply with the uncertainty principle which also arises from quantum noncommutativity? Here, first, we obtain sufficient conditions for the KD quasiprobability defined relative to a pair of projection-valued measure (PVM) bases to have nonclassical values. Using these nonclassical values, we then introduce two quantities which capture the amount of KD quantumness in a quantum state relative to a single PVM basis. They are defined, respectively, as the nonreality and the classicality---which captures both the nonreality and negativity---of the associated KD quasiprobability over the PVM basis of interest, and another PVM basis, and maximized over all possible choices of the latter. We obtain their lower bounds, and derive trade-off relations respectively reminiscent of the Robertson and Robertson-Schr\"odinger uncertainty relations but with lower bounds maximized over the convex sets of Hermitian operators whose complete sets of eigenprojectors are given by the PVM bases. We discuss their measurement using weak value measurement and classical optimization. We then suggest an information theoretical interpretation of the KD nonreality relative to a PVM basis as a lower bound to the maximum total root-mean-squared error in an optimal estimation of the PVM basis, and thereby obtain a lower bound and a trade-off relation for the root-mean-squared error. Finally, we suggest an interpretation of the KD nonclassicality relative to a PVM basis as a lower bound to the total state disturbance caused by a nonselective projective binary measurement associated with the PVM basis, and derive a lower bound and a trade-off relation for the disturbance.
Abstract Kirkwood discovered in 1933, and Dirac discovered in 1945, a representation of quantum states that has undergone a renaissance recently. The Kirkwood–Dirac (KD) distribution has been employed to study nonclassicality across quantum physics, from metrology to chaos to the foundations of quantum theory. The KD distribution is a quasiprobability distribution, a quantum generalization of a probability distribution, which can behave nonclassically by having negative or nonreal elements. Negative KD elements signify quantum information scrambling and potential metrological quantum advantages. Nonreal elements encode measurement disturbance and thermodynamic nonclassicality. KD distributions’ nonclassicality has been believed to follow necessarily from pairwise noncommutation of operators in the distribution’s definition. We show that noncommutation does not suffice. We prove sufficient conditions for the KD distribution to be nonclassical (equivalently, necessary conditions for it to be classical). We also quantify the KD nonclassicality achievable under various conditions. This work resolves long-standing questions about nonclassicality and may be used to engineer quantum advantages.