Quantum programming introduces both probabilistic and non-deterministic behavior, making formal verification a challenging task. This work develops a framework for modeling non-deterministic quantum programs using Markov Decision Processes (MDPs). Each program configuration is represented as a combination of the remaining program and its quantum state, while transitions capture both quantum evolution and non-deterministic branching. We define operational semantics for quantum programs in terms of MDPs and prove their equivalence to established denotational semantics. This equivalence ensures consistency between high-level mathematical descriptions and executable models. The approach enables the application of model-checking techniques, originally designed for classical probabilistic systems, to quantum programs. Our results show that MDPs provide a foundation for verifying properties of non-deterministic quantum programs.
Quantum Markov chains are an extension of classical Markov chains which are labelled with super-operators rather than probabilities. They allow to faithfully represent quantum programs and quantum protocols. In this paper, we investigate model checking omega-regular properties, a very general class of properties (including, e.g., LTL properties) of interest, against this model. For classical Markov chains, such properties are usually checked by building the product of the model with a language automaton. Subsequent analysis is then performed on this product. When doing so, one takes into account its graph structure, and for instance performs different analyses per bottom strongly connected component (BSCC). Unfortunately, for quantum Markov chains such an approach does not work directly, because super-operators behave differently from probabilities. To overcome this problem, we transform the product quantum Markov chain into a single super-operator, which induces a decomposition of the state space (the tensor product of classical state space and the quantum one) into a family of BSCC subspaces. Interestingly, we show that this BSCC decomposition provides a solution to the issue of model checking omega-regular properties for quantum Markov chains.
We are currently in the midst of a second quantum revolution: transition from quantum theory to quantum engineering (e.g. quantum computing, communication, sensing). The main purpose of quantum theory is to find fundamental rules governing the existing physical systems. In contrast, quantum engineering aims at designing and implementing new systems (machines, devices, etc.) to achieve some desirable tasks, based on quantum theory. From experience in today's engineering, it is not always easy for a human designer to completely understand the behaviours of the system that she/he is designing, and an error in her/his design may cause serious problems and even disasters. Consequently, theories and methodologies for verification of correctness, safety and reliability of complex engineering systems have been systematically studied in various engineering fields. In particular, computer scientists have developed techniques to verify the correctness of both hardware and software as well as the security of communication protocols. Human intuition is poorly adapted to the quantum world compared to the classical world, which implies that human engineers will make many more mistakes in designing and implementing complex quantum systems such as quantum computer hardware and software and communication protocols. Even worse, because of the essential differences between the classical and quantum worlds, verification techniques for classical systems cannot be directly used for quantum systems. Novel verification techniques will be indispensable for the coming era of quantum engineering and technology. Model-checking is an effective technique to check whether a system satisfies a desired property. The properties that are checked are usually specified in a temporal logic; typical properties are deadlock freedom, invariants, safety, and request–response properties. The systems being checked are mathematically modelled as e.g. (finite-state) automata, transition systems, Markov chains and Markov decision processes [1]. In the last three decades, model-checking has become one of the dominant techniques for verification of computer hardware and software, and has proved mature as witnessed by a large number of successful industrial applications. Techniques of model-checking were even applied in systems biology recently. With the emergence of quantum engineering and quantum technology, a question then naturally arises: is it possible to apply model-checking techniques in verifying the correctness and safety of quantum engineering systems, and how can this be done? Unfortunately, due to some essential differences between classical and quantum systems, it seems unlikely that the classical model-checking techniques can be directly applied to quantum systems. Basically, to make model-checking techniques effective for quantum systems, the following three problems must be systematically addressed: System modelling and property specification: Behaviours of quantum systems cannot be described using classical modelling methods and, consequently, the properties of the quantum systems to be checked cannot be formalized by classical specification languages. As a result, novel conceptual frameworks must be proposed to properly model and consider quantum systems, including formal models and formal descriptions of temporal properties of quantum systems. Quantum measurements: Model-checking is usually applied to check the long-term behaviours of systems. However, to check whether a quantum system satisfies a certain property at a time point, one has to perform a quantum measurement on the system, which can change the state of the system. This makes studying the long-term behaviours of quantum systems much harder than with classical systems. Algorithms: Classical model-checking algorithms normally assume the state spaces to be finite or countably infinite. However, state spaces of quantum systems are inherently continuous. To develop algorithms for model-checking quantum systems, deep mathematical properties of the systems have to be exploited, so that a finite (or countably infinite) number of representative elements in the state spaces will suffice. Note that the state space of any quantum system has a natural linear algebraic structure. A well developed algorithm for verifying quantum systems should make clever use of this structure. Despite the difficulties discussed above, a few model-checking techniques for quantum systems have been developed in the last 10 years. The earliest work mainly targeted checking quantum communication protocols: Taking the probabilism arising from quantum measurements into account, the probabilistic model-checker PRISM was used in [2] to verify the correctness of quantum protocols, including superdense coding, quantum teleportation and quantum error correction. A branching-time temporal extension of exogenous quantum propositional logic was introduced and then the model-checking problem for this logic was studied in [3], with verification of the correctness of quantum key distribution BB84 as an application. A linear temporal extension of exogenous quantum propositional logic was then defined and the corresponding model-checking problem was investigated in [4]. Model-checking techniques were developed in [5] for quantum communication protocols modelled in process algebra CQP (communicating quantum processes) [6]. A model-checker for quantum communication protocols was also developed in [7], where only the protocols that can be modelled as quantum circuits expressible in the stabilizer formalism were considered. This technique was further extended beyond stabilizer states and used to check equivalence of quantum protocols. A research line pursued by the authors and their collaborators is to develop model-checking techniques that can be used not only for quantum communication protocols but also for general quantum systems, including physical systems and quantum programs. Quantum automata were adopted in [8,9] as the model of the systems: Definition 1(Quantum automata [10]). A quantum automaton is a 4-tuple |
A=(H,Act,{Uα:α∈Act},H0)|, where: 1) |
H| is a finite-dimensional Hilbert space, called the state space; 2) |
Act| is a finite set of action names; 3) for each action name |
α∈Act|, Uα is a unitary operator on |
H|; 4) |
H0⊆H| is the subspace of initial states. A quantum automaton behaves as follows: it starts from some initial state in |
H0|, and at each step it performs a unitary transformation Uα for some |
α∈Act|. An algorithm for checking certain linear-time properties (e.g. invariants and safety properties) was proposed in [8], where, following Birkhoff–von Neumann quantum logic, closed subspaces of the state Hilbert space are used as the atomic propositions about the state of the system, and the checked linear-time properties are defined as infinite sequences of sets of atomic propositions. Furthermore, the decidability or undecidability of several reachability problems for quantum automata were established in [9]. The model-checking problem for a larger class of quantum systems than quantum automata, namely quantum Markov chains, was studied in [11]. Note that continuous-time quantum Markov processes have been studied intensively in mathematical physics. Discrete-time quantum Markov chains were recently introduced as a semantic model for quantum programs. Definition 2 (Quantum Markov chains [11]). A quantum Markov chain is a triple |
(H,E,H0)|, where |
H| and |
H0| are the same as in Definition 1, and |
E| is a super-operator on |
H|. A quantum Markov chain starts in an initial state in |
H0|, and at each step it performs the (same) quantum operation modelled by the super-operator |
E|. Note that the (discrete-time) dynamics of closed quantum systems is usually depicted by unitary operators, and the behaviours of open quantum systems are described by super-operators. Obviously, the notion of quantum automata can be generalized by replacing unitary operators Uα in Definition 1 by super-operators |
Eα|. Furthermore, quantum Markov decision processes [12] can be defined by introducing decision strategies into such generalized quantum automata. Several algorithms for checking the reachability of quantum Markov chains and quantum Markov decision processes have been developed. As in checking classical Markov chains and Markov decision processes, graph reachability is a key to these algorithms. However, classical graph theory is not suited to our purpose; instead, a new theory of quantum graphs (i.e. graphs in a Hilbert space with an adjacency relation induced by a super-operator) was developed and, in particular, an algorithm for the BSCC (bottom strongly connected components) decomposition of the state Hilbert spaces was found in [11]. Another decomposition technique, namely periodic decomposition, for quantum Markov chains has recently been proposed. The notion of the super-operator-valued Markov chain is introduced in [13] as a higher-level model of quantum programs and quantum cryptographic protocols. Definition 3 (Super-operator-valued Markov chains [13]). A labelled super-operator-valued Markov chain over a set AP of predefined atomic propositions is a 5-tuple |
(S,s0,H,Q,L)|, where: 1) S is a finite set of classical states with s0 ∈ S being the initial state; 2) |
H| is a finite-dimensional Hilbert space, called the quantum state space; 3) |
Q:S×S→SOH| is a transition super-operator function, where |
SOH| denotes the set of trace-nonincreasing super-operators on |
H| and, for each s ∈ S, ∑t ∈ SQ(s, t) is trace-preserving; and 4) L: S → 2AP is a labelling function. A super-operator-valued Markov chain has two state spaces, a classical one and a quantum one, which are connected through the transition super-operator function. It behaves in a similar manner to classical Markov chains. It starts from the classical initial state s0 but with the quantum initial state unspecified (it can be taken arbitrarily). Then at each step, given the current classical state s and quantum state ρ, it proceeds to a classical state t with probability tr[Q(s, t)(ρ)], and the accompanying quantum state evolves into Q(s, t)(ρ)/tr[Q(s, t)(ρ)] provided that tr[Q(s, t)(ρ)] ≠ 0. The normalization requirement that ∑t ∈ SQ(s, t) is trace-preserving guarantees that the probabilities of going from s to some classical state sum up to 1. As the atomic propositions are taken to be classical (they apply only to classical states), this Markov chain model is suitable for verification of quantum systems against classical properties, such as running time, termination, reachability, etc. One distinct feature of this model, however, is that it allows us to check the properties of the system once and for all; i.e. the verified results apply to all initial quantum states. For example, the model-checking algorithm for the reachability problem essentially calculates a positive operator Π, accounting for all (classical) paths satisfying the concerned property. Then the reachability probability when the Markov chain starts in the initial quantum state ρ is simply tr(Πρ). A corresponding computation tree logic (CTL) for super-operator-valued Markov chains was defined, and algorithms for checking such properties were developed in [13]. A tool implementation of these algorithms has been provided [14] based on a probabilistic model-checker. Algorithms for model-checking ω-regular properties, a general class of properties subsuming linear temporal logic (LTL) formulas, against super-operator-valued Markov chains have been proposed [15], thus allowing analysis of a wide range of properties such as repeated reachability, reachability in a restricted order, and nested Until properties. Furthermore, the reachability problem of a recursive extension of super-operator-valued Markov chains was studied in [16], with the application of analysing quantum programs with procedure calls. As reviewed in previous sections, several theoretical frameworks and algorithms of quantum model-checking have been developed. Certainly, however, quantum model-checking is still at a very early stage of its development; in particular, its applications are only at the level of toy examples. We envisage that, in the future, quantum model-checking techniques can be applied to the following areas: Checking physical systems: Physicists have already considered the algorithmic checking problem of certain properties of quantum systems, e.g. quantum measurement occurrence [17] and reachability of quantum states [18]. Quantum model-checking can offer a systematic view of this line of research. Verification of quantum circuits: Verification of circuits has been one of the major application areas of classical model-checking. But model-checking applied to verification of quantum circuits is an area to be systematically exploited. Analysis and verification of quantum programs: Another important application area of classical model-checking is analysis and verification of programs. Several techniques for analysis and verification of quantum programs have been reported in the last few years [19,20]. However, model-checking techniques specifically designed for quantum programs are still missing. Verification of security of quantum communication protocols: Applications of model-checking mentioned in the section entitled ‘Early research on model-checking of quantum systems' focus on verification of the correctness of quantum communication protocols. However, verification of the security of quantum protocols is much more difficult, and model-checking applied to it is an interesting topic for future research.
Abstract Model-checking techniques have been extended to analyze quantum programs and communication protocols represented as quantum Markov chains, an extension of classical Markov chains. To specify qualitative temporal properties, a subspace-based quantum temporal logic is used, which is built on Birkhoff-von Neumann atomic propositions. These propositions determine whether a quantum state is within a subspace of the entire state space. In this paper, we propose the measurement-based linear-time temporal logic MLTL to check quantitative properties. MLTL builds upon classical linear-time temporal logic (LTL) but introduces quantum atomic propositions that reason about the probability distribution after measuring a quantum state. To facilitate verification, we extend the symbolic dynamics-based techniques for stochastic matrices described by Agrawal et al. (JACM 2015) to handle more general quantum linear operators (super-operators) through eigenvalue analysis. This extension enables the development of an efficient algorithm for approximately model checking a quantum Markov chain against an MLTL formula. To demonstrate the utility of our model-checking algorithm, we use it to simultaneously verify linear-time properties of both quantum and classical random walks. Through this verification, we confirm the previously established advantages discovered by Ambainis et al. (STOC 2001) of quantum walks over classical random walks and discover new phenomena unique to quantum walks.
Most modern (classical) programming languages support recursion. Recursion has also been successfully applied to the design of several quantum algorithms and introduced in a couple of quantum programming languages. So, it can be expected that recursion will become one of the fundamental paradigms of quantum programming. Several program logics have been developed for verification of quantum while-programs. However, there are as yet no general methods for reasoning about (mutual) recursive procedures and ancilla quantum data structure in quantum computing (with measurement). We fill the gap in this paper by proposing a parameterized quantum assertion logic and, based on which, designing a quantum Hoare logic for verifying parameterized recursive quantum programs with ancilla data and probabilistic control. The quantum Hoare logic can be used to prove partial, total, and even probabilistic correctness (by reducing to total correctness) of those quantum programs. In particular, two counterexamples for illustrating incompleteness of non-parameterized assertions in verifying recursive procedures, and, one counterexample for showing the failure of reasoning with exact probabilities based on partial correctness, are constructed. The effectiveness of our logic is shown by three main examples -- recursive quantum Markov chain (with probabilistic control), fixed-point Grover's search, and recursive quantum Fourier sampling.
• Quantum Markov chain for quantum programs based on classical control and quantum data • A QCTL where a proposition is a pair of a classical state and a projection • Sound and complete translation of quantum Markov chains to DTMCs • Convergence and simple cycles make model checking decidable Several quantum Markov chains and temporal quantum logics have been considered in the literature. In this paper we propose a discrete time classical-quantum state quantum Markov chain (cqQMC) with super-operators over a finite dimensional Hilbert space H . The cqQMC is related to previous proposals but yet with a different semantics in that states are pairs of a classical and a quantum state. We define a quantum temporal logic (QCTL) where propositions are pairs of a classical state and a projection over H . We demonstrate how a cqQMC can be translated to a discrete time Markov chain (DTMC). The translation is sound and complete in that model checking any QCTL formula for a cqQMC can equivalently be carried out by checking the same formula for the inferred DTMC. Taking advantage of the finite classical control we show that even though a cqQMC may have infinitely many states then under certain restrictions the inferred DTMC is finite, and hence model checking is decidable.