QuantGraph: A Receding-Horizon Quantum Graph Solver
Extracted source text · arXiv 2512.15476
\documentclass[a4paper, amsfonts, amssymb, amsmath, reprint, showkeys, superscriptaddress, nofootinbib,twoside, 10pt, aps, nobibnotes]{revtex4-2}
\usepackage[english]{babel}
\usepackage[utf8]{inputenc}
\usepackage[normalem]{ulem}
\usepackage{amsmath,bm}
\usepackage{mathtools}
\usepackage{epstopdf}
\usepackage{mathtools}
\usepackage{extarrows}
\usepackage{booktabs}
\usepackage{graphicx}
\usepackage{subcaption}
\usepackage[export]{adjustbox}
\usepackage[colorinlistoftodos, color=green!40, prependcaption]{todonotes}
\usepackage{graphicx}
\usepackage{hyperref}
\usepackage{braket}
\usepackage{enumitem} %
\usepackage[linesnumbered,ruled,vlined]{algorithm2e}
\usepackage[noend]{algpseudocode} %
\SetNlSkip{1.0em}
\renewcommand{\algorithmicrequire}{\textbf{Input:}}
\renewcommand{\algorithmicensure}{\textbf{Output:}}
\newcommand{\AP}[1]{\textcolor{red}{#1}}
\newcommand{\PV}[1]{\textcolor{blue}{#1}}
\renewcommand{\thesection}{\Roman{section}}
\renewcommand{\thesubsection}{\thesection.\Roman{subsection}}
\renewcommand{\thesubsubsection}{\thesubsection.\Roman{subsubsection}}
\bibliographystyle{apsrev4-1}
\preprint{APS/123-QED}
\begin{document}
\title{QuantGraph: A Receding-Horizon Quantum Graph Solver}
\author{Pranav Vaidhyanathan}
\thanks{Equal Contribution.\\
\texttt{\{pranav, aristotelis\}@robots.ox.ac.uk}}
\affiliation{Department of Engineering Science, University of Oxford, Oxford OX1 3PJ, United Kingdom}
\author{Aristotelis Papatheodorou}
\thanks{Equal Contribution.\\
\texttt{\{pranav, aristotelis\}@robots.ox.ac.uk}}
\affiliation{Department of Engineering Science, University of Oxford, Oxford OX1 3PJ, United Kingdom}
\author{David R.\ M.\ Arvidsson-Shukur}
\affiliation{Hitachi Cambridge Laboratory, J. J. Thomson Avenue, Cambridge CB3 0HE, United Kingdom}
\author{Mark T.\ Mitchison}
\affiliation{School of Physics, Trinity College Dublin, College Green, Dublin D02 K8N4, Ireland}
\affiliation{Department of Physics, King's College London, Strand, London WC2R 2LS, United Kingdom}
\author{Natalia Ares}
\affiliation{Department of Engineering Science, University of Oxford, Oxford OX1 3PJ, United Kingdom}
\author{Ioannis Havoutis}
\affiliation{Department of Engineering Science, University of Oxford, Oxford OX1 3PJ, United Kingdom}
\begin{abstract}
Dynamic programming~\cite{DP_2010} is a cornerstone of graph-based optimization. While effective, it scales unfavorably with problem size. In this work, we present QuantGraph, a two-stage quantum-enhanced framework that casts local and global graph-optimization problems as quantum-searches over discrete trajectory spaces. The solver is designed to operate efficiently by first finding a sequence of locally optimal transitions in the graph (local stage), without considering full trajectories. The accumulated cost of these transitions acts as a threshold that prunes the search space (up to $60\%$ reduction for certain examples). The subsequent global stage, based on this threshold, refines the solution. Both stages utilize variants of the Grover-adaptive-search algorithm~\cite{Gilliam_2021}. To achieve scalability and robustness, we draw on principles from control theory and embed QuantGraph’s global stage within a receding-horizon model-predictive-control scheme. This classical layer stabilizes and guides the quantum search, improving precision and reducing computational burden. In practice, the resulting closed-loop system exhibits robust behavior and lower overall complexity. Notably, for a fixed query budget, QuantGraph attains a $2\times$ increase in control-discretization precision while still benefiting from Grover-search’s inherent quadratic speedup compared to classical methods.
\end{abstract}
\keywords{quantum-enhanced dynamic programming, two-stage Grover-adaptive-search, quantum-enhanced trajectory optimization, quantum-enhanced model-predictive-control}
\maketitle
\section{Introduction}
\label{sec:intro}
Many decision-making problems in science and engineering can be framed as finding the minimum-cost path through graph-like structures~\cite{blum1997fast, bertsekas2012dynamic, powell_dp}. This simple abstraction underlies applications ranging from autonomous driving and robotics to energy dispatch and logistics. In all such problems, a system must move from one place to another while minimizing effort, risk, or resource use along the way. Classical methods that leverage this structure have long been the workhorse for such problems. Yet, they are challenged by modern applications that commonly live in big-data regimes and require long planning horizons.
Robotics provides a particularly compelling illustration of these challenges. Robots have become one of the most transformative technologies of the modern era, supporting society in areas ranging from healthcare and manufacturing to disaster response~\cite{trevelyan2016robotics}, agriculture~\cite{sparrow2021robots}, and home assistance~\cite{yamazaki2012home}. By operating in places that are hazardous, remote, or physically demanding, they extend human reach and enable new forms of service and mobility. Their growing presence in everyday life relies fundamentally on the ability to make rapid and trustworthy decisions in complex, uncertain environments. As these systems take on broader responsibilities, the number of possibilities they must evaluate in real time expands sharply~\cite{papatheodorou2024momentumawaretrajectoryoptimisationusing,vaidhyanathan2025metasymsymplecticmetalearningframework}. This places increasing pressure on planning and decision-making algorithms, underscoring the need for computational tools capable of exploring vast decision spaces both efficiently and reliably.
Quantum computing offers a promising avenue to address these computational bottlenecks~\cite{preskill2018quantum, cerezo2021variational}. Owing to the principles of superposition and entanglement, quantum processors can evaluate many possible transition sequences simultaneously. Algorithms like Grover’s search~\cite{grover1996fastquantummechanicalalgorithm} increase the chance of finding trajectories that meet a chosen metric or constraint, offering an advantage over classical search (see Section~\ref{sec:background}). However, direct implementations that operate on full trajectories require circuits that are too large for current quantum hardware~\cite{shukla2019trajectory}.
In this work, we introduce QuantGraph, a two-stage quantum-enhanced solver for minimum-cost path problems. QuantGraph is designed to scale beyond current classical approaches. The method builds on principles that guide effective decision making in control and planning. It uses receding-horizon feedback and warm-start priors~\cite{mattingley2011receding} to steer the search toward useful regions of the solution space. QuantGraph adapts these ideas to the quantum setting by allocating computation where it matters most. This design aligns with insights from model-predictive-control (see Section~\ref{sec:robotics_background}), reinforcement learning~\cite{SuttonRL} and evolutionary algorithms~\cite{EC}, which show that many real tasks do not require exploring every possible decision to obtain strong solutions.
Building on these principles, the framework operates in two phases. At first, it identifies a small set of promising next actions and computes their combined cost. This produces a baseline that serves as a warm-start prior. Then, the algorithm evaluates only those trajectories that improve on this baseline. It does so by considering the full-planning horizon of the problem. QuantGraph focuses computation on the most relevant parts of the search space instead of examining all possible paths, while preserving solution quality.
To ensure scalability beyond current quantum-search methods, the second stage of QuantGraph follows a receding-horizon strategy~\cite{kouvaritakis2016model}. Instead of solving the entire problem at once, it optimizes a short control window, applies the first action, and then shifts the window forward as new information becomes available. This approach is standard in control theory and allows the solver to correct errors and adapt to changing conditions at every step. As the window advances, each iteration begins from the updated state and uses a newly refined threshold. This allows QuantGraph to focus its resources on the most relevant region of the search space at that moment. Our empirical results show that this targeted allocation reduces computational load and improves reliability over long tasks (see Section~\ref{sec:analysis}).
To demonstrate the capabilities of our proposed framework, we focus on the challenging problem of trajectory optimization for discrete-time dynamical systems. In such problems, a system evolves along a trajectory represented as a path through a temporal graph of states (nodes) and actions (edges). By recasting the underlying problem into a quadratic-unconstrained-binary-optimization formulation~\cite{Qubo}, we are able to utilize QuantGraph's scalability and inherent quadratic speedup with profound implications across many scientific fields. Section~\ref{sec:results} presents our findings in detail.
\subsection{Background}
\label{sec:background}
To frame the operation of QuantGraph, we begin with a concise overview of the core ideas and notation used by the robotics and quantum computing communities. The aim is to build an intuitive understanding of the underlying principles without resorting to formal derivations.
\subsubsection*{A. Robot dynamics and trajectory optimization}\label{sec:robotics_background}
A core problem in robotics is to drive a robot, or more generally, a dynamical system, toward a desired goal. Among model-based approaches, trajectory optimization provides a systematic framework to achieve this. Such methods rest on two fundamental pillars: the \textbf{dynamics model}, which predicts how the system evolves under control inputs, and the \textbf{optimization solver}, which determines the sequence of control inputs that minimize a surrogate objective describing the task or goal given the system's dynamics.
\paragraph*{\textbf{Dynamics modeling:}}
A robot’s motion evolves according to physical laws that relate its state ($\mathbf x_k$) (e.g., positions, velocities, orientations) and \textbf{control input} ($\mathbf u_k$) (e.g., motor torques, wheel forces) across time. In discrete time, the state-evolution for a time step ($k$) is described by the state-space model~\cite{siciliano2008springer},
\begin{equation}
\mathbf x_{k+1} = f(\mathbf x_k, \mathbf u_k),
\label{nonlinear_dynamics}
\end{equation}
where $f(\cdot,\cdot)$ can be nonlinear. When linear or linearized around an operating point, we obtain the \textbf{linear time-invariant} form~\cite{willems1986time, oppenheim1997signals},
\begin{equation}
\mathbf x_{k+1} = \mathbf A \mathbf x_k + \mathbf B \mathbf u_k,
\label{LTI}
\end{equation}
which captures many robotic systems such as manipulators or wheeled platforms near equilibrium \cite{sastry2011adaptive}. These dynamics define how the system moves from a state to the next, and thus determine the transitions that form a trajectory. The collection of all possible trajectories can be visualized as a graph. Each node represents a state ($\mathbf x_k$), and each edge represents the transition to the next state ($\mathbf x_{k+1}$). By encoding the problem with this graph structure, we can represent every admissible path the system can follow over time.
\begin{figure*}[ht!]
\centering
\begin{subfigure}[t]{0.52\textwidth}
\centering
\includegraphics[valign=t,width=\linewidth]{Images/iter1.pdf}
\caption{Iteration 1 of the model predictive controller}
\label{fig:MPC_iter1}
\end{subfigure}
\hfill
\begin{subfigure}[t]{0.47\textwidth}
\centering
\includegraphics[valign=t,width=\linewidth]{Images/iter2.pdf}
\vspace{3mm}
\caption{Iteration 2 of the model predictive controller}
\label{fig:MPC_iter2}
\end{subfigure}
\caption{Subsequent model-predictive-control (MPC) iterations: (a) In early model-predictive-control steps, discretized states deviate from the continuous optimum, with compounding errors $\delta x_1 < \delta x_2$ along the horizon. To limit computation over the full horizon $T$, the model-predictive-control framework optimizes only the first $N_c \ll T$ control actions, while beyond $N_c$ the control-inputs are held constant (the \emph{Fixed Horizon} block). Although this truncation increases discretization drift, receding-horizon re-optimization corrects the tail error at each iteration, applying only the first control input before shifting the horizon forward, while there are significant gains in terms of computational efficiency. (b) In subsequent iterations, the model-predictive-control framework recomputes controls from the updated state, compensating for prior deviations and refining the trajectory. Through this iterative correction, the closed-loop trajectory converges toward the continuous-time optimum while maintaining tractable computational cost.}
\label{fig:MPC}
\end{figure*}
\paragraph*{\textbf{Trajectory optimization:}}
The goal of trajectory optimization is to find the optimal control sequence $\{\mathbf u^*_k\}_{k=0}^{T-1}$ which, given the system dynamics in Eq.~\eqref{nonlinear_dynamics}, steers the system over a finite horizon of length $T$~\cite{rao2014trajectory,betts1998survey}. This horizon can be represented as a $T$-layered tree, where each layer $k=0,\dots,T$ collects all states ($\mathbf x_k$) that are reachable at time step $k$, and each edge between layers corresponds to applying an admissible control input ($\mathbf u_k$) and propagating the dynamics to the next state ($\mathbf x_{k+1}$) (see Fig.~\ref{fig:MPC}). Any root-to-leaf path in this tree is thus a dynamically feasible trajectory. Among all such trajectories, we seek the one that minimizes an accumulated performance index,
\begin{equation}
J = \sum_{k=0}^{T-1} \ell_k(\mathbf x_k, \mathbf u_k) + \phi(\mathbf x_T),
\end{equation}
where $\ell_k(\mathbf x_k, \mathbf u_k)$ is the \textit{stage-wise cost} and $\phi(\mathbf x_T)$ is the terminal penalty. Common quadratic cost functions~\cite{mastalli20crocoddyl} are,
\begin{equation}
\begin{aligned}
\ell_k(\mathbf x_k, \mathbf u_k) &= (\mathbf x_k - \mathbf x_k^{\mathrm{ref}})^{\top} \mathbf Q (\mathbf x_k - \mathbf x_k^{\mathrm{ref}}) \\
&\quad + (\mathbf u_k - \mathbf u_k^{\mathrm{ref}})^{\top} \mathbf R (\mathbf u_k - \mathbf u_k^{\mathrm{ref}}),\\
\phi(\mathbf x_T) &= (\mathbf x_T - \mathbf x_T^{\mathrm{ref}})^{\top} \mathbf P (\mathbf x_T - \mathbf x_T^{\mathrm{ref}}),
\label{eq:cost}
\end{aligned}
\end{equation}
where $\mathbf Q, \mathbf R, \mathbf P \succeq 0$ are weighting matrices that balance tracking accuracy against control effort. Moreover, $\mathbf x^{\mathrm{ref}}, \mathbf u^{\mathrm{ref}}$ denote a reference trajectory that regularizes the problem and encourages smooth convergence of the solver to a desirable motion. Intuitively, the quadratic terms penalize deviations from the reference state and input at each time step, while the terminal cost pushes the system to reach the intended terminal state. Minimizing $J$ therefore selects, among all feasible paths in the $T$-layered tree, the dynamically consistent trajectory that remains close to the desired motion while using control inputs efficiently. Note that in our case, we use only the terminal cost to penalize deviation from the goal state ($\mathbf x_T^{ref}$), while the remaining stage-wise costs penalize large state and control magnitudes.
\paragraph*{\textbf{Dynamic programming:}}
The trajectories or transitions in the aforementioned graphs are evaluated recursively by algorithms such as classical dynamic programming~\cite{DP_2010}, which relies on Bellman's principle of optimality. Intuitively, the principle states that any suffix of an optimal trajectory must itself be optimal for the corresponding intermediate state. Formally, the optimal cost-to-go $J_k^*(\mathbf x_k)$ satisfies
\begin{equation}
\begin{aligned}
&J_k^*(\mathbf x_k) = \min_{\mathbf u_k}\big[\ell_k(\mathbf x_k,\mathbf u_k) + J_{k+1}^*(f(\mathbf x_k,\mathbf u_k))\big],\\
&J_T^*(\mathbf x_T)=\phi(\mathbf x_T),
\end{aligned}
\end{equation}
which expresses the decomposition of the global-optimization problem into a sequence of stage-wise decisions. The recursion states that each optimal action minimizes its immediate cost plus the best future cost resulting from that action. This allows efficient backward traversal over the most promising state transitions and guarantees a globally optimal trajectory without enumerating every path. However, global optimality comes at a cost. Even with this recursion, dynamic programming suffers from the curse of dimensionality. For highly combinatorial tasks, its complexity can grow exponentially with problem size, making real-time control impractical.
\paragraph*{\textbf{Model predictive control:}}
This challenge can be mitigated by solving a \textit{finite-horizon} optimal control problem at each time step. We distinguish the \emph{predictive horizon} $N_p$ from a shorter \emph{control horizon} ($N_c$) and the total number of time steps of the task ($T$), with $N_c \le N_p \leq T$. Over $N_p$ steps we predict states and accumulate cost, but we only optimize the first $N_c$ control actions.
Beyond $t + N_c - 1$, the inputs are constrained by a simple \emph{terminal (tail) policy}, for example, applying the last optimized input $\mathbf u_{t + N_c - 1}$. This \emph{fixed} tail reduces decision variables while the receding-horizon recalculations correct any long-range approximation errors (see Figure~\ref{fig:MPC}).
Starting from the current state ($\mathbf x_t$), the controller minimizes over the stage-wise control inputs ($\mathbf u_k$) the following problem:
\begin{equation}
\begin{aligned}
\min_{\{\mathbf u_k\}_{k=t}^{t+N_c-1}} &\sum_{k=t}^{T-1} \ell_k(\mathbf x_k, \mathbf u_k) + \phi(\mathbf x_{T}),\\
&\quad \text{s.t.} \quad \mathbf x_{k+1}=f(\mathbf x_k,\mathbf u_k)\\
&\mathbf u_k =
\begin{cases}
\mathbf u_k, & k \le t + N_c - 1, \\
\mathbf u_{t + N_c - 1}, & k \ge t + N_c,
\end{cases}
\label{eq:MPC_objective}
\end{aligned}
\end{equation}
over a shorter \textbf{control horizon} ($N_c \ll T$). Only the first optimal control ($\mathbf u_t^\star$) is applied to the system, after which new measurements update the state and the optimization repeats (i.e. \textit{receding-horizon} principle). By continually re-optimizing in a sliding-window fashion with fresh data, model-predictive-control naturally adapts to model errors, disturbances, and changing goals. This, effectively, provides a feedback mechanism that bridges optimal planning and real-time control. Figure~\ref{fig:MPC} illustrates the underlying operation principles~\cite{Kober2014,SuttonRL}.
In this work, we reinterpret trajectory optimization and model-predictive-control through a \textbf{quantum-enhanced lens}, framing the problem as a graph search over discrete trajectories. This allows us to leverage quantum-search algorithms to achieve scalable, efficient, and adaptive control for complex robotic systems.
\subsubsection*{B. Quantum search and optimization}\label{sec:quantum_background}
The limits of classical dynamic programming and large-scale graph searches motivate the use of quantum-search algorithms, which can explore large discrete spaces more efficiently. To clarify this connection, we briefly introduce the notation and principles underlying quantum computation.
A quantum computer encodes information in a \emph{state vector} $|\psi\rangle$ that resides in a complex Hilbert space $\mathcal{H} \cong \mathbb{C}^{N}$, where $N = 2^{M}$ is the size of the computational basis for a register of $M$ qubits. This state is a normalized superposition of basis vectors,
\[
|\psi\rangle = \sum_{x=0}^{N-1} \psi_x |x\rangle, \qquad \sum_{x=0}^{N-1} |\psi_x|^2 = 1,
\]
with each complex amplitude $\psi_x$ representing the contribution of basis state $|x\rangle$~\cite{borrelli2010dirac}. In our setting, each basis state can be viewed as encoding one discrete trajectory through a graph.
Quantum computation proceeds through the application of unitary operators $U$, which play a role analogous to reversible transition maps in classical systems. Because $U^\dagger U = I$, these operators preserve norms and ensure deterministic state evolution:
\[
|\psi_{\mathrm{final}}\rangle = U\,|\psi_{\mathrm{initial}}\rangle.
\]
Measurement is where quantum mechanics departs most clearly from classical computation. A classical register returns a definite value. On the contrary, a quantum state must be sampled probabilistically. Upon measurement, the outcome is a bitstring $x$ drawn with probability
\[
P(x) = |\langle x | \psi \rangle|^2,
\]
according to the Born rule~\cite{nielsen2010quantum}. Thus, while the computation evolves deterministically, the readout is inherently random, reflecting the amplitudes accumulated along different encoded trajectories.
\paragraph*{\textbf{Grover's algorithm:}}
Grover's quantum search algorithm is a cornerstone of quantum optimization, offering a provable quadratic speedup for unstructured-search tasks~\cite{grover1996fastquantummechanicalalgorithm}. Whereas a classical search through an unsorted database of size N requires $O(N)$ queries, Grover's algorithm can identify a marked element in just $O(\sqrt{N})$ queries. The algorithm's power stems from three core quantum-mechanical steps:
\begin{enumerate}
\item \textbf{State preparation:} The process begins by preparing an $n$-qubit register, where $N=2^n$, in a uniform superposition of all possible computational basis states. This is typically achieved by applying a Hadamard gate to each qubit in the $\ket{0}^{\otimes n}$ state, yielding the state $\ket{s} = \frac{1}{\sqrt{N}}\sum_{x=0}^{N-1}\ket{x}$~\cite{borrelli2010dirac}. This initial state represents all candidate solutions simultaneously with equal amplitude. In the context of trajectory optimization, this superposition encodes the entire search space of possible discretized control sequences (trajectories) the dynamical system can execute over the planning horizon.
\item \textbf{The quantum oracle ($\mathcal{O}$):} The heart of the algorithm is an oracle, which is a unitary operator designed to recognize solutions. For a search problem defined by a function $m(x)$ that returns 1 for a solution state $\ket{x_w}$ and 0 otherwise, the oracle applies a conditional phase shift, such that $\ket{x} \mapsto (-1)^{m(x)}\ket{x}$. This operation effectively \emph{marks} the solution state by inverting its phase (e.g., $\ket{x_w} \mapsto -\ket{x_w}$) while leaving all other states unchanged. Crucially, the oracle does this without collapsing the superposition~\cite{kashefi2002comparison}.
\item \textbf{Amplitude amplification ($\mathcal{D}$):} Following the oracle's application, a diffusion operator, $\mathcal{D} = 2\ket{s}\bra{s} - I$ (where $I$ is the identity), is applied. This operator performs an inversion about the mean of the amplitudes of all states. Geometrically, this operation amplifies the amplitude of the marked state (which has a negative phase) while shrinking the amplitudes of all other states.
\end{enumerate}
The sequential application of the oracle and the diffusion operator constitutes a single \emph{Grover iteration}. Each time the marked state with the highest amplitude is used as a theshold for the subsequent iteration. By repeating these iterations approximately $\frac{\pi}{4}\sqrt{N}$ times, the amplitude of the solution state is driven close to 1, ensuring that a final measurement will yield the correct solution with high probability.
\paragraph*{\textbf{Grover-adaptive-search:}}
While the canonical algorithm is designed for finding a specific known item, it can be adapted for optimization through methods such as Grover-adaptive-search~\cite{Gilliam_2021}. Grover-adaptive-search reframes the optimization problem as a series of quantum searches. It iteratively searches for any solution with a cost below a dynamically updated threshold, $\tau$. Upon finding a superior solution, the threshold is lowered to the new best-known cost, progressively converging toward the global minimum. This transforms Grover's algorithm from a simple search primitive into a powerful minimum-finding solver.
\paragraph*{\textbf{Implications and challenges:}}
The implications of this quadratic speedup are profound for discretized continuum problems, such as the trajectory optimization tasks central to robotics. For a fixed-query budget, a quantum search can interrogate a quadratically larger solution space than its classical counterpart. As shown in Section \ref{sec:analysis}, the quantum advantage allows for a doubling of the bits of precision ($M_{quantum} \approx 2 \cdot M_{classical}$) in control discretization. Effectively, this translates to a quadratic reduction in discretization error and a more accurate approximation of the true continuous optimum.
\subsection{Overview of our approach}\label{sec:contributions}
In what follows, we describe QuantGraph's main components and capabilities. Primarily, we introduce a quantum-enhanced solver for generic graph optimization problems, with a particular focus on minimum-cost path search. This reframes trajectory optimization problems that are typically addressed via dynamic programming into a more general graph-search setting with a quadratic-unconstrained-binary-optimization formulation~\cite{chella2022quantum}.
Our framework applies Grover-adaptive-search to this formulation, inheriting its quadratic query-complexity speedup over dynamic-programming solvers. The advantage here is significant, since instead of evaluating the entire decision tree, the quantum search directly promotes low-cost trajectories. As a result, the global task of identifying high-quality paths from an exponentially large set can be carried out without exhaustive enumeration. This allows the solver to explore more paths with finer discretizations than classical optimal-control algorithms.
QuantGraph is organized as a two-stage, local–global architecture. The local stage considers only a single-step horizon that requires fewer qubits to quickly identify promising immediate actions. The accumulated cost of these single-step transitions serves as a threshold that warm-starts the global Grover-adaptive-search. The global stage then refines the solution efficiently, leveraging the cost prior from the local stage to guide the search and accelerate the discovery of globally optimal trajectories.
Finally, we embed QuantGraph within a receding-horizon model-predictive-control framework to manage large, dynamically-evolving state spaces. Instead of constructing and searching the full set of possible trajectories at once, the solver optimizes over a sliding window, substantially reducing memory demands. This provably retains robust long-horizon performance. As shown in Section~\ref{sec:analysis}, this design converts the exponential query complexity of the original Grover-adaptive-search algorithm into a linear dependence on the task horizon for a fixed window size. This in turn yields measurable gains in terms of precision and accuracy for trajectory optimization problems.
\section{Results}\label{sec:results}
We evaluate QuantGraph on a range of graph-based traversal and path-planning problems. We begin with a static example to illustrate the roles of the local and global stages and their impact on runtime and convergence. We then examine more challenging, robotics-inspired settings involving systems with dynamically evolving trajectories, such as the linear time-invariant double integrator and the highly nonlinear cart pole. Each benchmark highlights different strengths and limitations of the framework. All simulations are carried out using IBM’s Qiskit~\cite{qiskit}, while further task-specific details are provided in the following sections.
\subsection{Static graph}\label{sec:static_graph}
To illustrate how our approach operates, we begin with a small, static navigation problem.
This problem is described by the static directed-acyclic-graph illustrated in Figure~\ref{fig:static_graph}. The graph consists of nodes connected with transitions. Each transition has an associated cost. The task is to traverse the graph and accumulate the lowest cost possible from node~\textbf{a} to node ~\textbf{h}.
\begin{figure}[htb!]
\includegraphics[width=0.51\textwidth]{Images/StaticGraph.pdf}
\caption{A simple navigation problem as a directed graph with weighted transitions. QuantGraph is called to find the path associated with the lowest cost possible from node \textbf{a} to node \textbf{h}. The local-search step acts as a sub-optimal threshold that warm starts the global search effectively minimizing the required Grover-adaptive-search iterations needed for convergence.}
\label{fig:static_graph}
\end{figure}
In our example, there are five possible paths that the solver could take. Let $\{\mathbf a, \dots, \mathbf h\} \in \mathcal X$ represent a trajectory inside the set of all feasible trajectories $\mathcal X$. The cost associated for each trajectory is shown in Table~\ref{tab:static_graph_traj_cost}. The local stage acts as a threshold to the global stage, effectively eliminating three out of five trajectories ($60\%$ search-space reduction). The threshold reduces the candidates for the global stage, leading to accelerated convergence. Even though the static graph is a toy problem used to build intuition, the performance advantage of our two-stage process becomes significant as the graphs scale to several million nodes with multiple local minima, such the ones found in social media platforms~\cite{facebook_anatomy}.
\begin{table}[ht]
\centering
\begin{tabular}{@{} l | r @{}}
\toprule
Trajectories & Cost \\
\hline\hline
\midrule
$\{\mathbf a, \mathbf d, \mathbf e, \mathbf h\}$ & 23 \\
$\{\mathbf a, \mathbf d, \mathbf e, \mathbf f, \mathbf g, \mathbf h\}$ & \textbf{18} \\
$\{\mathbf a, \mathbf b, \mathbf c, \mathbf f, \mathbf g, \mathbf h\}$ & 22 \\
$\{\mathbf a, \mathbf b, \mathbf c, \mathbf d, \mathbf e, \mathbf h\}$ & 34 \\
$\{\mathbf a, \mathbf b, \mathbf c, \mathbf d, \mathbf e, \mathbf f, \mathbf g, \mathbf h\}$ & 29 \\
\bottomrule
\end{tabular}
\caption{List of all possible trajectories from Figure \ref{fig:static_graph} and their associated costs.}
\label{tab:static_graph_traj_cost}
\end{table}
\subsection{Double integrator}\label{sec:double_int}
To demonstrate the scalability of QuantGraph, we benchmark it against a highly combinatorial, dynamically evolving graph associated with trajectory optimization problems. More specifically, the problem concerns a discretized linear time-invariant double-integrator system. For example, the system can be thought of as a box of unit mass ($1$ kg) on a frictionless one-dimensional plane. Its dynamics are described in Eq.~\eqref{eq:double_int}, in which $q$ represents the position of the system and $u$ the driving force~\cite{ren2008consensus}, which is bound constrained by the maximum admissible force $u_{max}$. Each time step of the discretized model is represented by the index $k$.
\begin{figure*}[ht!]
\includegraphics[width=1.0\textwidth]{Images/DoubleInt.pdf}
\caption{Planned trajectories for the double integrator: The local stage warm starts the global stage of QuantGraph that operates in a receding horizon fashion. The objective is to drive the double integrator to the target position. To demonstrate the consistency of our framework, we plot the results for ten runs. The local stage converges to a sub-optimal trajectory as expected, warm-starting the global stage that converges to a smooth optimal trajectory with low variance across runs. }
\label{fig:double_integrator}
\end{figure*}
\begin{equation}
\begin{aligned}
\ddot{q}&=u(t), \quad|u(t)| \leq u_{\text{max}} \Rightarrow \\
\dot{\mathbf x} &= \overbrace{\begin{bmatrix} 0 & 1 \\0 & 0\end{bmatrix}}^{\mathbf A}\mathbf x + \overbrace{\begin{bmatrix} 0 \\ 1\end{bmatrix}}^{\mathbf B}u(t) \text{, with } \mathbf x=\begin{bmatrix} q \\ \dot{q} \end{bmatrix} \Rightarrow \\
\mathbf x_{k+1} &= e^{\mathbf Adt}\mathbf x_k + \int_{t_k}^{t_k+dt}e^{\mathbf A\left( t_k+dt-\tau\right)}\mathbf Bu\; d\tau \Rightarrow \\
\mathbf x_{k+1} &= \begin{bmatrix} 1 & dt \\0 & 1\end{bmatrix}\mathbf x_k + \begin{bmatrix} \frac{1}{2}dt^2 \\ dt\end{bmatrix}u_k.\\
\end{aligned}
\label{eq:double_int}
\end{equation}
In terms of complexity the dynamics are linear, but the set of feasible trajectories forms a large, dynamically evolving tree. QuantGraph’s receding-horizon global stage mitigates this problem by enabling stable and efficient operation without searching the full tree. As in classical model-predictive-control, short horizons may yield solutions that are not fully optimal, but increasing the window length brings the result arbitrarily close to the global optimum. The trade-off between tractability and full-horizon optimality is inherent to large graph optimization problems, where memory limits naturally motivate a sliding-window approach. The results for the double integrator example are illustrated in Figure~\ref{fig:double_integrator}.
\subsection{Cart pole}\label{sec:cartpole}
To demonstrate the scalability and computational performance of our framework, we evaluate it on a cart-pole trajectory-optimization task. In this problem, an underactuated pole with a mass is attached on top of an actuated cart moving freely on a frictionless one-dimensional plane. The goal is to raise the pole to a standing position. The system is described in Eq.~\ref{eq:cart_pole}, where $m$ represents the mass ($0.1$ kg) ontop of the pole ($l$) with length $0.5$ m, $M$ the cart's mass ($1$ kg) and $F$ the force driving the cart~\cite{geva2002cartpole}:
\begin{figure*}[ht!]
\centering
\begin{subfigure}[t]{0.68\textwidth}
\centering
\includegraphics[width=\linewidth]{Images/CartPole.pdf}
\caption{}
\label{fig:planned_trajectories}
\end{subfigure}
\hfill
\begin{subfigure}[t]{0.31\textwidth}
\centering
\includegraphics[width=\linewidth]{Images/CartPoleSystem.pdf}
\caption{}
\label{fig:system}
\end{subfigure}
\caption{(a) Planned trajectories for the cart pole system. The primary objective is to swing up the underactuated pole mass to the upright position while minimizing displacement of the cart from its initial location (secondary objective). The plot shows the mean and standard deviation over ten independent runs. The negligible variance in the evolution of the pole angle $\left(\theta\right)$ highlights the consistency and reliability of our planning approach. (b) Schematic of the physical cart pole system, annotated with the variables used in Eq.~\eqref{eq:cart_pole}.}
\label{fig:cartpole}
\end{figure*}
\begin{equation}
\begin{aligned}
&\ddot{\theta} = \frac{1}{\mu(\theta)} \big( \frac{\cos \theta}{l} F + \frac{(m+M)g}{l} \sin(\theta)\\&\qquad \qquad \qquad- m \cos(\theta) \sin(\theta) \dot{\theta}^2\big),\\
&\ddot{q} = \frac{1}{\mu(\theta)} \big( F + m \cos(\theta) \sin(\theta) g - m l \sin(\theta) \dot{\theta} \big),
\end{aligned}
\label{eq:cart_pole}
\end{equation}
and where $\mu(\theta) := M + m \sin(\theta)^2$ designates the effective mass of the cart pole.
The nonlinearity of this system makes it a meaningful benchmark for assessing QuantGraph’s capabilities. The local stage, designed for linear time-invariant systems with quadratic costs, is less effective in handling the cart pole's nonlinearities. Future extensions could relax this limitation by incorporating ancilla qubits~\cite{AncillaQubitsRef}, though at the cost of increased circuit complexity. This in turn will provide efficient warm-start capabilities, accelerating the convergence of the overall algorithm.
On the other hand, QuantGraph’s global stage is inherently capable of addressing general nonlinear dynamics and non-quadratic cost landscapes. This allows it to successfully solve challenging control tasks such as the cart-pole problem, even in the absence of guidance from the local stage, thereby underscoring the robustness and versatility of our approach.
The results for the cart-pole, path-planning problem are presented in Figure~\ref{fig:cartpole}. In this experiment, the solver's global stage is initially cold-started, with each subsequent iteration warm-started by the obtained cost of the previous one. Notably, the cold start imposes minimal computational overhead. This is attributed to the global stage's quadratic speedup that is inherent to Grover search algorithm.
More specifically, the global stage operates directly on the cost function and system’s native nonlinearities. By executing this procedure in a receding-horizon manner, our framework demonstrates strong convergence properties and robustness to initialization. This highlights its effectiveness in handling complex, nonlinear control tasks even without the local stage.
\section{Discussion}
\label{conclusion}
In this work, we introduced QuantGraph, a quantum-enhanced framework for solving graph-structured optimization problems, including trajectory planning in robotics and control. By reformulating dynamic programming as quantum search over discrete-trajectory spaces, we showed that quantum-accelerated methods can provide computational gains (see Section~\ref{sec:results}). Our two-stage architecture uses Grover-adaptive-search at its core. As shown in Section~\ref{sec:analysis}, our framework delivers the expected quadratic speedup while remaining computationally tractable. Empirical results on the double-integrator and cart-pole systems demonstrate that QuantGraph performs well on both linear and nonlinear dynamics.
From a theoretical perspective, QuantGraph links quantum optimization with practical control. While we have focused on linearized dynamics described by Eq.~\eqref{LTI}, the local search could be extended to nonlinear dynamics by embedding this into a linear evolution over ancilla qubits~\cite{dattani2019quadratizationdiscreteoptimizationquantum}, albeit at higher circuit cost. Another direction is to incorporate quantum algorithms for continuous optimization, which would mitigate the quantization errors inherent in the current discretization scheme \cite{jin2025fixed}. A detailed noise analysis would also clarify robustness to both sensor noise and quantum hardware imperfections. Importantly, the rapid and improved convergence of the model-predictive-control paradigm highlights how classical control methods can bolster quantum computing primitives, just as quantum-inspired algorithms have become a source of new ideas in classical computing (e.g.~see Ref.~\cite{arrazola2020} and references therein). Advances in control can guide the development of quantum algorithms, and quantum resources may in turn strengthen the performance of established control methods~\cite{chella2022quantum}.
The receding-horizon structure of QuantGraph also offers practical advantages on quantum hardware. Optimizing over a short control horizon $N_c$ keeps circuit depth and CNOT count proportional to $N_c$ rather than the full horizon $T$, reducing the decoherence burden, due to the short time horizon~\cite{sahay2025error}. At each sliding-window iteration, the quantum register is reinitialized, which clears accumulated stochastic phase errors before they can propagate. The model-predictive-control loop enables corrective actions by re-optimizing from the current state, correcting errors introduced in earlier iterations. Progressive tightening of the threshold $\tau$ further shrinks the search space and reduces the number of required Grover iterations.
As quantum hardware continues to improve, we anticipate that QuantGraph and similar hybrid algorithms will find increasing applications in real-world optimization problems ranging from autonomous-vehicle trajectory planning to supply-chain optimization and energy-grid management~\cite{gachnang2022quantum, blenninger2024q}. By carefully integrating quantum primitives with classical frameworks and exploiting natural error mitigation through feedback, we can develop algorithms that deliver meaningful speedups. This work shows that the intersection of quantum computing, optimization, and control offers promising avenues for developing methods that combine the strengths of each field to address increasingly complex real-world problems~\cite{preskill2018quantum}.
\section{Methods}\label{sec:methods}
\subsection{Problem formulation}
QuantGraph considers the linear time-invariant discrete dynamics of Eq.~\eqref{LTI}, with the states $\mathbf x_k\in\mathbb{R}^{n_x}$ and controls $\mathbf u_k\in\mathbb{R}^{n_u}$. There are different costs driving each stage of the algorithm:
\begin{itemize}
\item \textbf{Local stage}: The solver optimizes one-step transitions that act as threshold to the global stage. Hence the associated cost that provides a surrogate objective for this task is given by,
\begin{equation}
\begin{aligned}
\quad\quad \ell_{local,k} =& (\mathbf x_k - \mathbf x_k^{\mathrm{ref}})^{\top} \mathbf Q (\mathbf x_k - \mathbf x_k^{\mathrm{ref}}) \\
&+ (\mathbf u_k - \mathbf u_k^{\mathrm{ref}})^{\top} \mathbf R (\mathbf u_k - \mathbf u_k^{\mathrm{ref}})\\
&+ (\mathbf x_{k+1} - \mathbf x_{T}^{\mathrm{ref}})^{\top} \mathbf P (\mathbf x_{k+1} - \mathbf x_{T}^{\mathrm{ref}})
\end{aligned}
\end{equation}
\item \textbf{Global stage}: The solver minimizes the cost given by Eq.~\eqref{eq:cost} and more specifically the model-predictive-control objective in Eq.~\eqref{eq:MPC_objective} that increases robustness and handles highly combinatorial problems more effectively.
\end{itemize}
Each control component is encoded in fixed-point binary format with \(M\) bits,
yielding a binary vector $\mathbf b_k\in\{0,1\}^{M}$.
Stacking all control inputs across time produces
$$
\mathbf b=\bigl[\mathbf b_0^{\!\top},\dots, \mathbf b_{N_c-1}^{\!\top}\bigr]^{\!\top}
\in\{0,1\}^{MN_c},
$$
and straightforward algebra collects the cumulative cost in the quadratic-unconstrained-binary-optimization form
\begin{equation}
L(\mathbf b)= \mathbf b^{\top} \mathbf Q \mathbf b + \mathbf q^{\top} \mathbf b + c ,
\label{eq:qubo}
\end{equation}
with $\mathbf Q\in\mathbb{R}^{MN_c\times MN_c}$ and $\mathbf q\in\mathbb{R}^{MN_c}$. For linear time-invariant systems, these matrices can be efficiently synthesized (in polynomial time) at runtime based on the current state and system dynamics.
\subsection{Two-stage search architecture}
The optimization objective stated in Eq.~\eqref{eq:qubo} proceeds in two stages as seen in Algorithm \ref{alg:quantgraph}.
\paragraph*{\textbf{Local stage}.} The horizon is split into \(T\) single-step sub-problems, each using \(M_{\mathrm{loc}}\) qubits. Algorithm \ref{alg:grovermin} is executed on every sub-problem, while summing the resulting costs yields a bound \(\tau_0\) that prunes the subsequent global search.
\begin{algorithm}[htb!]
\caption{\textsc{QuantGraph}}
\label{alg:quantgraph}
\SetAlgoLined
\SetNlSty{}{}{\hspace{-1em}} %
\DontPrintSemicolon %
\KwIn{Initial state $x_0$; total horizon $T$, control horizon $N_c \ll T$,
\#bits~$(M_{\mathrm{loc}},M_{\mathrm{glob}})$;}
\KwOut{Optimal state trajectory $\mathbf X^\star$, control sequence $\mathbf U^\star$;}
\textbf{Setup:}
Discretize the control function into $2^{M_{\mathrm{loc}}}$ (local) and
$2^{M_{\mathrm{glob}}}$ (global) binary symbols; Pre-compute quadratic unconstrained binary optimization matrices
$\mathbf Q$ for one-step costs $\ell_k$. Let $\ell_{\mathrm{acc}}$ be the accumulated cost;
\textbf{Local Stage}\Comment{Local Warm-Start}\;
$\tau\gets+\infty$;\; $\ell_{\mathrm{acc}}\gets0$;\;
\For{$k\gets0$ \KwTo $T-1$}{
$(\mathbf u_k^{\mathrm{loc}},\ell_k)\gets
\textsc{GroverMin}\!\bigl(\mathbf Q,M_{\mathrm{loc}},\tau\bigr)$\;
$\ell_{\mathrm{acc}}\gets \ell_{\mathrm{acc}}+\ell_k$;\;
\If{$\ell_{\mathrm{acc}}<\tau$}{\ $\tau\gets \ell_{\mathrm{acc}}$\;}
$\mathbf x_{k+1}\gets \mathbf A \mathbf x_k + \mathbf B \mathbf u_k^{\mathrm{loc}}$ \Comment{Linear-time-invariant dynamics only}
}
$\ell_{\mathrm{acc}}\gets \ell_{\mathrm{acc}}+\ell_{T}$;\;
\textbf{Global Stage (Receding-Horizon Grover-adaptive-search).}\;
$t\gets0$;\; $X^\star\gets[\mathbf x_0]$;\; $U^\star\gets[\;]$;\;
$\tau \gets \ell_{\mathrm{acc}}$ (Warm start);\;
\While{$t\le T-N_c-1$}{
Synthesize $N_c$-horizon quadratic unconstrained optimization problem $(\mathbf{Q}_t, \mathbf{q}_t)$ starting at state $\mathbf{x}$.\;
$(\mathbf{b}_{t:t+N_c},\ell_{\min})\gets
\textsc{GroverMin}\!\bigl(\mathbf{Q}_t, \mathbf{q}_t,\,M_{\mathrm{glob}}N_c,\,\tau\bigr)$\;
$\tau\gets\min(\tau,\ell_{\min})$.\;
Decode first block of $\mathbf{b}_{t:t+N_c}$ to $\hat{\mathbf{u}}_t$.\;
$\mathbf{x}\gets \mathbf{f}(\mathbf{x}, \hat{\mathbf{u}}_t)$ \Comment{Apply first input; general dynamics}\;
Append $\hat{\mathbf{u}}_t$ to $\mathbf{U}^\star$ and $\mathbf{x}$ to $\mathbf{X}^\star$;\;
$t\gets t+1$;\;
}
\textbf{Stage 3 (Return).}\;
\Return{$\mathbf{X}^\star, \mathbf{U}^\star$ and $\ell_{min}$}\;
\end{algorithm}
\paragraph*{\textbf{Receding-horizon global stage}.} At control instant \(t\), starting from state $\mathbf{x}_t$, the linear time-invariant dynamics (see Eq.~\eqref{LTI}) and quadratic costs over the control horizon $N_c \ll T$ are synthesized into a quadratic-unconstrained-binary-optimization formulation (Eq.~\eqref{eq:qubo}). Grover-adaptive-search is then applied using an implicit oracle derived from this quadratic unconstrained optimization problem definition (Algorithm \ref{alg:grovermin}), searching the space of \(2^{M_{\mathrm{glob}}N_c}\) control sequences with the incumbent bound \(\tau_t\). Only the first element of the current best sequence is applied to the plant, after which the window shifts and the process repeats. The required register therefore never exceeds $M_{\mathrm{glob}}N_c$ plus the ancillary qubits required for the quantum arithmetic in the implicit oracle. Importantly, this procedure can handle trajectories of arbitrary temporal lengths.
\subsection{Grover adaptive search}
As seen in Algorithm \ref{alg:grovermin},
Grover-adaptive-search maintains a classical threshold \(\tau\) and iterates:
\begin{enumerate}[label=(\alph*)]
\item \textbf{Implicit oracle construction.}
Constructs a quantum circuit (the implicit oracle) $\mathcal{O}_\tau$ that evaluates the quadratic-unconstrained-binary-optimization's cost function in superposition and performs a conditional phase flip: $\lvert \mathbf{b}\rangle \mapsto (-1)^{[L(\mathbf{b})\le\tau]}\lvert \mathbf{b}\rangle$. This requires quantum arithmetic circuits to compute $L(\mathbf{b})$ and a quantum comparator.
\item \textbf{Randomized amplification.}
Draws \(k\sim\mathcal{U}\{0,\dots,K-1\}\) and apply $k$ Grover iterations $G=D\mathcal{O}_\tau$, where \(D=2\lvert s\rangle\langle s\rvert - I\) and $\lvert s\rangle$ is the uniform superposition. Failure to improve the cost triggers $K \leftarrow 2K$.
\item \textbf{Measurement and update.}
If the measured candidate cost is below~\(\tau\), updates \(\tau\) and records the index. Otherwise the algorithm doubles \(K\). The loop terminates after one full-doubling cycle with no improvement, yielding the global minimum with probability $1-1/N$.
\end{enumerate}
\begin{algorithm}[H]
\caption{\textsc{GroverMin}$(\mathbf{Q}, \mathbf{q}, m, \tau)$}
\label{alg:grovermin}
\SetAlgoLined
\SetNlSty{}{}{\hspace{-1em}} %
\SetInd{0.0em}{0.5em}
\KwIn{Quadratic unconstrained optimization matrices $\mathbf{Q}, \mathbf{q}$; qubit count $m$; threshold $\tau$}
\KwOut{Optimal bitstring $\mathbf{b}^\star$ and cost $L(\mathbf{b}^\star)$}
Initialize uniform superposition on $m$ qubits\;
\While{improvement observed}{
Construct implicit oracle $O_\tau$ that evaluates $L(\mathbf{b}) = \mathbf{b}^{\top} \mathbf{Q} \mathbf{b} + \mathbf{q}^{\top} \mathbf{b}$ and flips phase if $L(\mathbf{b})\le\tau$.\;
Apply diffusion operator $D$ and repeat $(O_\tau;D)$ for $k$ iterations (using randomized amplification schedule)\;
Measure register to obtain $\mathbf{b}_{\mathrm{cand}}$\;
$\ell_{\mathrm{cand}}\gets L(\mathbf{b}_{\mathrm{cand}})$ \Comment{Evaluate classically}\;
\If{$\ell_{\mathrm{cand}}<\tau$}{
$\tau\gets \ell_{\mathrm{cand}}$\;
$\mathbf{b}^\star\gets \mathbf{b}_{\mathrm{cand}}$\;
}\Else{
increase iteration budget $K$\;
}
}
\Return{$\mathbf{b}^\star,\tau$}\;
\end{algorithm}
\section{Computational Analysis}\label{sec:analysis}
We analyze the computational advantages of the QuantGraph framework, focusing on its query-complexity scaling and the impact of the quantum-search advantage on the precision of solutions for discretized continuum problems.
\subsection{Query complexity and scalability}
The query complexity of the proposed solver is determined by the combination of the Grover-adaptive-search algorithm and the model-predictive-control architecture~\cite{nayak1999quantum}. We compare this complexity to a monolithic application of Grover-adaptive-search (``original Grover-adaptive-search'') over the entire optimization horizon $T$.
Recall that the control input at each step is discretized using $M$ bits ($M_{\mathrm{glob}}$ in the global stage), and the predictive horizon has length $N_p = T - t$, where the control-input is optimized only for $N_c$ timesteps (the control horizon).
\paragraph*{\textbf{Monolithic search} (original Grover-adaptive-search).}
If a single Grover search were applied to optimize the entire trajectory simultaneously, the total search-space size would be $N_{\mathrm{vanilla}} = (2^M)^T = 2^{MT}$. The query complexity, benefiting from the quadratic speedup, is:
\begin{equation}
O(\sqrt{N_{\mathrm{vanilla}}}) = O(\sqrt{2^{MT}}) = O(2^{MT/2}).
\label{eq:complexity_vanilla}
\end{equation}
This scales exponentially with the total horizon $T$, making it intractable for long-duration tasks and demanding significant quantum resources.
\paragraph*{\textbf{QuantGraph framework}.}
QuantGraph operates in a receding-horizon manner. At each time step, the global stage optimizes over the window $N_c$. The search-space size per step is $N_{\mathrm{step}} = 2^{MN_c}$. The query complexity for the Grover-adaptive-search algorithm at each step is $O(\sqrt{N_{\mathrm{step}}}) = O(2^{MN_c/2})$.
The model-predictive-control loop runs for approximately $T-N_c$ iterations. Therefore, the total query complexity for QuantGraph is:
\begin{equation}
O((T-N_c) \cdot 2^{MN_c/2}) \approx O(T \cdot 2^{MN_c/2}).
\label{eq:complexity_quantgraph}
\end{equation}
Crucially, the complexity varies exponentially with the moving horizon $N_c$. However, if $N_c$ is kept small and fixed, the complexity scales only \textit{linearly} with the total time $T$. This shift from exponential to linear scaling in $T$ is the primary computational advantage of the model-predictive-control framework, allowing QuantGraph to handle large $T$ without incurring the exponential cost associated with monolithic search. It is important to note that the associated costs with this complexity are heavily influenced by various resources such as current quantum hardware implementations, error correction techniques and qubit coherence times \cite{montanaro2016quantum}.
\subsection{Precision in discretized continuum problems}
When continuous control problems (e.g., cart pole or double integrator) are discretized into a graph, the precision relates to how closely the best path found approximates the true optimum of the continuous problem. The precision is inherently linked to the discretization granularity $M$~\cite{szablowski2021understanding}. A larger $M$ yields a finer resolution $\Delta u \propto 1/2^M$, reducing the discretization error $\epsilon$.
We analyze the achievable precision compared to the classical case (e.g., dynamic programming) for a fixed query budget $\mathcal{Q}$.
Classical search requires $O(N)$ queries, while quantum search requires $O(\sqrt{N})$. For a budget $\mathcal{Q}$, the maximum searchable-space sizes are:
\begin{equation}
N_{\mathrm{classical}} \approx \mathcal{Q}, \qquad N_{\mathrm{quantum}} \approx \mathcal{Q}^2.
\end{equation}
We relate the search space size $N$ back to the bits of precision $M$ over a horizon $H$ (where $H$ could be $T$ for monolithic search or $N_c$ for receding-horizon formulation): $N = 2^{MH}$.
\begin{align}
2^{M_{\mathrm{classical}}H} &\approx \mathcal{Q}, \\
2^{M_{\mathrm{quantum}}H} &\approx \mathcal{Q}^2.
\end{align}
Taking the logarithm:
\begin{align}
M_{\mathrm{classical}}H &\approx \log_2(\mathcal{Q}), \\
M_{\mathrm{quantum}}H &\approx \log_2(\mathcal{Q}^2) = 2 \cdot \log_2(\mathcal{Q}).
\end{align}
Therefore, we find that $M_{\mathrm{quantum}} \approx 2 \cdot M_{\mathrm{classical}}$. For a fixed number of queries, the quantum approach allows for \textbf{twice the number of bits of precision} in the control discretization compared to the classical approach.
This translates into a quadratic improvement in the discretization error $\epsilon$:
\begin{equation}
\epsilon_{\mathrm{quantum}} \propto \frac{1}{2^{M_{\mathrm{quantum}}}} \approx \left(\frac{1}{2^{M_{\mathrm{classical}}}}\right)^2 \propto (\epsilon_{\mathrm{classical}})^2.
\label{eq:precision_quadratic}
\end{equation}
This analysis has demonstrated how QuantGraph drastically reduces the discretization error (see Eq.~\ref{eq:precision_quadratic}), providing a significantly better approximation of the continuous optimum within the constraints of the horizon $N_c$.
\section*{Acknowledgements}
P.V. is supported by the United States Army Research Office under Award No. W911NF-21-S-0009-2. A.P is supported by University of Oxford’s Clarendon Fund. M.T.M. is supported by a Royal Society University Research Fellowship. N.A. acknowledges support from the European Research Council (grant agreement 948932) and the Royal Society (URF-R1-191150).
\bibliography{references}
\end{document}
%merlin.mbs apsrev4-1.bst 2010-07-25 4.21a (PWD, AO, DPC) hacked
%Control: key (0)
%Control: author (72) initials jnrlst
%Control: editor formatted (1) identically to author
%Control: production of article title (-1) disabled
%Control: page (0) single
%Control: year (1) truncated
%Control: production of eprint (0) enabled
\begin{thebibliography}{45}%
\makeatletter
\providecommand \@ifxundefined [1]{%
\@ifx{#1\undefined}
}%
\providecommand \@ifnum [1]{%
\ifnum #1\expandafter \@firstoftwo
\else \expandafter \@secondoftwo
\fi
}%
\providecommand \@ifx [1]{%
\ifx #1\expandafter \@firstoftwo
\else \expandafter \@secondoftwo
\fi
}%
\providecommand \natexlab [1]{#1}%
\providecommand \enquote [1]{``#1''}%
\providecommand \bibnamefont [1]{#1}%
\providecommand \bibfnamefont [1]{#1}%
\providecommand \citenamefont [1]{#1}%
\providecommand \href@noop [0]{\@secondoftwo}%
\providecommand \href [0]{\begingroup \@sanitize@url \@href}%
\providecommand \@href[1]{\@@startlink{#1}\@@href}%
\providecommand \@@href[1]{\endgroup#1\@@endlink}%
\providecommand \@sanitize@url [0]{\catcode `\\12\catcode `\$12\catcode `\&12\catcode `\#12\catcode `\^12\catcode `\_12\catcode `\%12\relax}%
\providecommand \@@startlink[1]{}%
\providecommand \@@endlink[0]{}%
\providecommand \url [0]{\begingroup\@sanitize@url \@url }%
\providecommand \@url [1]{\endgroup\@href {#1}{\urlprefix }}%
\providecommand \urlprefix [0]{URL }%
\providecommand \Eprint [0]{\href }%
\providecommand \doibase [0]{http://dx.doi.org/}%
\providecommand \selectlanguage [0]{\@gobble}%
\providecommand \bibinfo [0]{\@secondoftwo}%
\providecommand \bibfield [0]{\@secondoftwo}%
\providecommand \translation [1]{[#1]}%
\providecommand \BibitemOpen [0]{}%
\providecommand \bibitemStop [0]{}%
\providecommand \bibitemNoStop [0]{.\EOS\space}%
\providecommand \EOS [0]{\spacefactor3000\relax}%
\providecommand \BibitemShut [1]{\csname bibitem#1\endcsname}%
\let\auto@bib@innerbib\@empty
%</preamble>
\bibitem [{\citenamefont {Bellman}(2010)}]{DP_2010}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {R.}~\bibnamefont {Bellman}},\ }\href@noop {} {\emph {\bibinfo {title} {Dynamic programming}}},\ \bibinfo {edition} {first princeton landmarks in mathematics edition}\ ed.,\ Princeton landmarks in mathematics\ (\bibinfo {publisher} {Princeton University Press},\ \bibinfo {address} {Princeton},\ \bibinfo {year} {2010})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Gilliam}\ \emph {et~al.}(2021)\citenamefont {Gilliam}, \citenamefont {Woerner},\ and\ \citenamefont {Gonciulea}}]{Gilliam_2021}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Gilliam}}, \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Woerner}}, \ and\ \bibinfo {author} {\bibfnamefont {C.}~\bibnamefont {Gonciulea}},\ }\href {\doibase 10.22331/q-2021-04-08-428} {\bibfield {journal} {\bibinfo {journal} {Quantum}\ }\textbf {\bibinfo {volume} {5}},\ \bibinfo {pages} {428} (\bibinfo {year} {2021})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Blum}\ and\ \citenamefont {Furst}(1997)}]{blum1997fast}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {A.~L.}\ \bibnamefont {Blum}}\ and\ \bibinfo {author} {\bibfnamefont {M.~L.}\ \bibnamefont {Furst}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Artificial intelligence}\ }\textbf {\bibinfo {volume} {90}},\ \bibinfo {pages} {281} (\bibinfo {year} {1997})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Bertsekas}(2012)}]{bertsekas2012dynamic}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {D.}~\bibnamefont {Bertsekas}},\ }\href@noop {} {\emph {\bibinfo {title} {Dynamic programming and optimal control: Volume I}}},\ Vol.~\bibinfo {volume} {4}\ (\bibinfo {publisher} {Athena scientific},\ \bibinfo {year} {2012})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Powell}(2011)}]{powell_dp}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {W.~B.}\ \bibnamefont {Powell}},\ }\href {\doibase 10.1002/9781118029176} {\emph {\bibinfo {title} {Approximate {Dynamic} {Programming}: {Solving} the {Curses} of {Dimensionality}}}},\ \bibinfo {edition} {1st}\ ed.,\ Wiley {Series} in {Probability} and {Statistics}\ (\bibinfo {publisher} {Wiley},\ \bibinfo {year} {2011})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Trevelyan}\ \emph {et~al.}(2016)\citenamefont {Trevelyan}, \citenamefont {Hamel},\ and\ \citenamefont {Kang}}]{trevelyan2016robotics}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Trevelyan}}, \bibinfo {author} {\bibfnamefont {W.~R.}\ \bibnamefont {Hamel}}, \ and\ \bibinfo {author} {\bibfnamefont {S.-C.}\ \bibnamefont {Kang}},\ }in\ \href@noop {} {\emph {\bibinfo {booktitle} {Springer handbook of robotics}}}\ (\bibinfo {publisher} {Springer},\ \bibinfo {year} {2016})\ pp.\ \bibinfo {pages} {1521--1548}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Sparrow}\ and\ \citenamefont {Howard}(2021)}]{sparrow2021robots}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {R.}~\bibnamefont {Sparrow}}\ and\ \bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Howard}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {precision agriculture}\ }\textbf {\bibinfo {volume} {22}},\ \bibinfo {pages} {818} (\bibinfo {year} {2021})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Yamazaki}\ \emph {et~al.}(2012)\citenamefont {Yamazaki}, \citenamefont {Ueda}, \citenamefont {Nozawa}, \citenamefont {Kojima}, \citenamefont {Okada}, \citenamefont {Matsumoto}, \citenamefont {Ishikawa}, \citenamefont {Shimoyama},\ and\ \citenamefont {Inaba}}]{yamazaki2012home}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {K.}~\bibnamefont {Yamazaki}}, \bibinfo {author} {\bibfnamefont {R.}~\bibnamefont {Ueda}}, \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Nozawa}}, \bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Kojima}}, \bibinfo {author} {\bibfnamefont {K.}~\bibnamefont {Okada}}, \bibinfo {author} {\bibfnamefont {K.}~\bibnamefont {Matsumoto}}, \bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Ishikawa}}, \bibinfo {author} {\bibfnamefont {I.}~\bibnamefont {Shimoyama}}, \ and\ \bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Inaba}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Proceedings of the IEEE}\ }\textbf {\bibinfo {volume} {100}},\ \bibinfo {pages} {2429} (\bibinfo {year} {2012})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Papatheodorou}\ \emph {et~al.}(2024)\citenamefont {Papatheodorou}, \citenamefont {Merkt}, \citenamefont {Mitchell},\ and\ \citenamefont {Havoutis}}]{papatheodorou2024momentumawaretrajectoryoptimisationusing}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Papatheodorou}}, \bibinfo {author} {\bibfnamefont {W.}~\bibnamefont {Merkt}}, \bibinfo {author} {\bibfnamefont {A.~L.}\ \bibnamefont {Mitchell}}, \ and\ \bibinfo {author} {\bibfnamefont {I.}~\bibnamefont {Havoutis}},\ }\href {https://arxiv.org/abs/2310.06074} {\enquote {\bibinfo {title} {Momentum-aware trajectory optimisation using full-centroidal dynamics and implicit inverse kinematics},}\ } (\bibinfo {year} {2024}),\ \Eprint {http://arxiv.org/abs/2310.06074} {arXiv:2310.06074 [cs.RO]} \BibitemShut {NoStop}%
\bibitem [{\citenamefont {Vaidhyanathan}\ \emph {et~al.}(2025)\citenamefont {Vaidhyanathan}, \citenamefont {Papatheodorou}, \citenamefont {Mitchison}, \citenamefont {Ares},\ and\ \citenamefont {Havoutis}}]{vaidhyanathan2025metasymsymplecticmetalearningframework}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {P.}~\bibnamefont {Vaidhyanathan}}, \bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Papatheodorou}}, \bibinfo {author} {\bibfnamefont {M.~T.}\ \bibnamefont {Mitchison}}, \bibinfo {author} {\bibfnamefont {N.}~\bibnamefont {Ares}}, \ and\ \bibinfo {author} {\bibfnamefont {I.}~\bibnamefont {Havoutis}},\ }\href {https://arxiv.org/abs/2502.16667} {\enquote {\bibinfo {title} {Metasym: A symplectic meta-learning framework for physical intelligence},}\ } (\bibinfo {year} {2025}),\ \Eprint {http://arxiv.org/abs/2502.16667} {arXiv:2502.16667 [cs.LG]} \BibitemShut {NoStop}%
\bibitem [{\citenamefont {Preskill}(2018)}]{preskill2018quantum}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Preskill}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Quantum}\ }\textbf {\bibinfo {volume} {2}},\ \bibinfo {pages} {79} (\bibinfo {year} {2018})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Cerezo}\ \emph {et~al.}(2021)\citenamefont {Cerezo}, \citenamefont {Arrasmith}, \citenamefont {Babbush}, \citenamefont {Benjamin}, \citenamefont {Endo}, \citenamefont {Fujii}, \citenamefont {McClean}, \citenamefont {Mitarai}, \citenamefont {Yuan}, \citenamefont {Cincio} \emph {et~al.}}]{cerezo2021variational}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Cerezo}}, \bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Arrasmith}}, \bibinfo {author} {\bibfnamefont {R.}~\bibnamefont {Babbush}}, \bibinfo {author} {\bibfnamefont {S.~C.}\ \bibnamefont {Benjamin}}, \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Endo}}, \bibinfo {author} {\bibfnamefont {K.}~\bibnamefont {Fujii}}, \bibinfo {author} {\bibfnamefont {J.~R.}\ \bibnamefont {McClean}}, \bibinfo {author} {\bibfnamefont {K.}~\bibnamefont {Mitarai}}, \bibinfo {author} {\bibfnamefont {X.}~\bibnamefont {Yuan}}, \bibinfo {author} {\bibfnamefont {L.}~\bibnamefont {Cincio}}, \emph {et~al.},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Nature Reviews Physics}\ }\textbf {\bibinfo {volume} {3}},\ \bibinfo {pages} {625} (\bibinfo {year} {2021})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Grover}(1996)}]{grover1996fastquantummechanicalalgorithm}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {L.~K.}\ \bibnamefont {Grover}},\ }\href {https://arxiv.org/abs/quant-ph/9605043} {\enquote {\bibinfo {title} {A fast quantum mechanical algorithm for database search},}\ } (\bibinfo {year} {1996}),\ \Eprint {http://arxiv.org/abs/quant-ph/9605043} {arXiv:quant-ph/9605043 [quant-ph]} \BibitemShut {NoStop}%
\bibitem [{\citenamefont {Shukla}\ and\ \citenamefont {Vedula}(2019)}]{shukla2019trajectory}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Shukla}}\ and\ \bibinfo {author} {\bibfnamefont {P.}~\bibnamefont {Vedula}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Journal of Global Optimization}\ }\textbf {\bibinfo {volume} {75}},\ \bibinfo {pages} {199} (\bibinfo {year} {2019})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Mattingley}\ \emph {et~al.}(2011)\citenamefont {Mattingley}, \citenamefont {Wang},\ and\ \citenamefont {Boyd}}]{mattingley2011receding}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Mattingley}}, \bibinfo {author} {\bibfnamefont {Y.}~\bibnamefont {Wang}}, \ and\ \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Boyd}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {IEEE Control Systems Magazine}\ }\textbf {\bibinfo {volume} {31}},\ \bibinfo {pages} {52} (\bibinfo {year} {2011})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Sutton}\ and\ \citenamefont {Barto}(2018)}]{SuttonRL}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {R.~S.}\ \bibnamefont {Sutton}}\ and\ \bibinfo {author} {\bibfnamefont {A.~G.}\ \bibnamefont {Barto}},\ }\href@noop {} {\emph {\bibinfo {title} {Reinforcement Learning: An Introduction}}}\ (\bibinfo {publisher} {A Bradford Book},\ \bibinfo {address} {Cambridge, MA, USA},\ \bibinfo {year} {2018})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Back}\ \emph {et~al.}(1997)\citenamefont {Back}, \citenamefont {Fogel},\ and\ \citenamefont {Michalewicz}}]{EC}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {T.}~\bibnamefont {Back}}, \bibinfo {author} {\bibfnamefont {D.~B.}\ \bibnamefont {Fogel}}, \ and\ \bibinfo {author} {\bibfnamefont {Z.}~\bibnamefont {Michalewicz}},\ }\href@noop {} {\emph {\bibinfo {title} {Handbook of Evolutionary Computation}}},\ \bibinfo {edition} {1st}\ ed.\ (\bibinfo {publisher} {IOP Publishing Ltd.},\ \bibinfo {address} {GBR},\ \bibinfo {year} {1997})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Kouvaritakis}\ and\ \citenamefont {Cannon}(2016)}]{kouvaritakis2016model}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Kouvaritakis}}\ and\ \bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Cannon}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Switzerland: Springer International Publishing}\ }\textbf {\bibinfo {volume} {38}},\ \bibinfo {pages} {7} (\bibinfo {year} {2016})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Kochenberger}\ \emph {et~al.}(2014)\citenamefont {Kochenberger}, \citenamefont {Hao}, \citenamefont {Glover}, \citenamefont {Lewis}, \citenamefont {L\"{u}}, \citenamefont {Wang},\ and\ \citenamefont {Wang}}]{Qubo}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {G.}~\bibnamefont {Kochenberger}}, \bibinfo {author} {\bibfnamefont {J.-K.}\ \bibnamefont {Hao}}, \bibinfo {author} {\bibfnamefont {F.}~\bibnamefont {Glover}}, \bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Lewis}}, \bibinfo {author} {\bibfnamefont {Z.}~\bibnamefont {L\"{u}}}, \bibinfo {author} {\bibfnamefont {H.}~\bibnamefont {Wang}}, \ and\ \bibinfo {author} {\bibfnamefont {Y.}~\bibnamefont {Wang}},\ }\href {\doibase 10.1007/s10878-014-9734-0} {\bibfield {journal} {\bibinfo {journal} {J. Comb. Optim.}\ }\textbf {\bibinfo {volume} {28}},\ \bibinfo {pages} {58–81} (\bibinfo {year} {2014})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Siciliano}\ \emph {et~al.}(2008)\citenamefont {Siciliano}, \citenamefont {Khatib},\ and\ \citenamefont {Kr{\"o}ger}}]{siciliano2008springer}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Siciliano}}, \bibinfo {author} {\bibfnamefont {O.}~\bibnamefont {Khatib}}, \ and\ \bibinfo {author} {\bibfnamefont {T.}~\bibnamefont {Kr{\"o}ger}},\ }\href@noop {} {\emph {\bibinfo {title} {Springer handbook of robotics}}},\ Vol.\ \bibinfo {volume} {200}\ (\bibinfo {publisher} {Springer},\ \bibinfo {year} {2008})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Willems}(1986)}]{willems1986time}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {J.~C.}\ \bibnamefont {Willems}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Automatica}\ }\textbf {\bibinfo {volume} {22}},\ \bibinfo {pages} {561} (\bibinfo {year} {1986})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Oppenheim}\ \emph {et~al.}(1997)\citenamefont {Oppenheim}, \citenamefont {Willsky},\ and\ \citenamefont {Nawab}}]{oppenheim1997signals}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {A.~V.}\ \bibnamefont {Oppenheim}}, \bibinfo {author} {\bibfnamefont {A.~S.}\ \bibnamefont {Willsky}}, \ and\ \bibinfo {author} {\bibfnamefont {S.~H.}\ \bibnamefont {Nawab}},\ }\href@noop {} {\emph {\bibinfo {title} {Signals \& systems}}}\ (\bibinfo {publisher} {Pearson Educaci{\'o}n},\ \bibinfo {year} {1997})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Sastry}\ and\ \citenamefont {Bodson}(2011)}]{sastry2011adaptive}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Sastry}}\ and\ \bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Bodson}},\ }\href@noop {} {\emph {\bibinfo {title} {Adaptive control: stability, convergence and robustness}}}\ (\bibinfo {publisher} {Courier Corporation},\ \bibinfo {year} {2011})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Rao}(2014)}]{rao2014trajectory}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {A.~V.}\ \bibnamefont {Rao}},\ }in\ \href@noop {} {\emph {\bibinfo {booktitle} {Optimization and optimal control in automotive systems}}}\ (\bibinfo {publisher} {Springer},\ \bibinfo {year} {2014})\ pp.\ \bibinfo {pages} {3--21}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Betts}(1998)}]{betts1998survey}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {J.~T.}\ \bibnamefont {Betts}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Journal of guidance, control, and dynamics}\ }\textbf {\bibinfo {volume} {21}},\ \bibinfo {pages} {193} (\bibinfo {year} {1998})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Mastalli}\ \emph {et~al.}(2020)\citenamefont {Mastalli}, \citenamefont {Budhiraja}, \citenamefont {Merkt}, \citenamefont {Saurel}, \citenamefont {Hammoud}, \citenamefont {Naveau}, \citenamefont {Carpentier}, \citenamefont {Righetti}, \citenamefont {Vijayakumar},\ and\ \citenamefont {Mansard}}]{mastalli20crocoddyl}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {C.}~\bibnamefont {Mastalli}}, \bibinfo {author} {\bibfnamefont {R.}~\bibnamefont {Budhiraja}}, \bibinfo {author} {\bibfnamefont {W.}~\bibnamefont {Merkt}}, \bibinfo {author} {\bibfnamefont {G.}~\bibnamefont {Saurel}}, \bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Hammoud}}, \bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Naveau}}, \bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Carpentier}}, \bibinfo {author} {\bibfnamefont {L.}~\bibnamefont {Righetti}}, \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Vijayakumar}}, \ and\ \bibinfo {author} {\bibfnamefont {N.}~\bibnamefont {Mansard}},\ }in\ \href@noop {} {\emph {\bibinfo {booktitle} {IEEE International Conference on Robotics and Automation (ICRA)}}}\ (\bibinfo {year} {2020})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Kober}\ and\ \citenamefont {Peters}(2014)}]{Kober2014}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Kober}}\ and\ \bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Peters}},\ }\enquote {\bibinfo {title} {Reinforcement learning in robotics: A survey},}\ in\ \href {\doibase 10.1007/978-3-319-03194-1_2} {\emph {\bibinfo {booktitle} {Learning Motor Skills: From Algorithms to Robot Experiments}}}\ (\bibinfo {publisher} {Springer International Publishing},\ \bibinfo {address} {Cham},\ \bibinfo {year} {2014})\ pp.\ \bibinfo {pages} {9--67}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Borrelli}(2010)}]{borrelli2010dirac}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Borrelli}},\ }in\ \href@noop {} {\emph {\bibinfo {booktitle} {Styles of Thinking in Science and Technology. Proceedings of the 3rd International Conference of the European Society for the History of Science}}}\ (\bibinfo {year} {2010})\ pp.\ \bibinfo {pages} {361--371}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Nielsen}\ and\ \citenamefont {Chuang}(2010)}]{nielsen2010quantum}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {M.~A.}\ \bibnamefont {Nielsen}}\ and\ \bibinfo {author} {\bibfnamefont {I.~L.}\ \bibnamefont {Chuang}},\ }\href@noop {} {\emph {\bibinfo {title} {Quantum computation and quantum information}}}\ (\bibinfo {publisher} {Cambridge university press},\ \bibinfo {year} {2010})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Kashefi}\ \emph {et~al.}(2002)\citenamefont {Kashefi}, \citenamefont {Kent}, \citenamefont {Vedral},\ and\ \citenamefont {Banaszek}}]{kashefi2002comparison}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {E.}~\bibnamefont {Kashefi}}, \bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Kent}}, \bibinfo {author} {\bibfnamefont {V.}~\bibnamefont {Vedral}}, \ and\ \bibinfo {author} {\bibfnamefont {K.}~\bibnamefont {Banaszek}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Physical Review A}\ }\textbf {\bibinfo {volume} {65}},\ \bibinfo {pages} {050304} (\bibinfo {year} {2002})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Chella}\ \emph {et~al.}(2022)\citenamefont {Chella}, \citenamefont {Gaglio}, \citenamefont {Pilato}, \citenamefont {Vella},\ and\ \citenamefont {Zammuto}}]{chella2022quantum}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Chella}}, \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Gaglio}}, \bibinfo {author} {\bibfnamefont {G.}~\bibnamefont {Pilato}}, \bibinfo {author} {\bibfnamefont {F.}~\bibnamefont {Vella}}, \ and\ \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Zammuto}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Mathematics}\ }\textbf {\bibinfo {volume} {10}},\ \bibinfo {pages} {2475} (\bibinfo {year} {2022})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Javadi-Abhari}\ \emph {et~al.}(2024)\citenamefont {Javadi-Abhari}, \citenamefont {Treinish}, \citenamefont {Krsulich}, \citenamefont {Wood}, \citenamefont {Lishman}, \citenamefont {Gacon}, \citenamefont {Martiel}, \citenamefont {Nation}, \citenamefont {Bishop}, \citenamefont {Cross}, \citenamefont {Johnson},\ and\ \citenamefont {Gambetta}}]{qiskit}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Javadi-Abhari}}, \bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Treinish}}, \bibinfo {author} {\bibfnamefont {K.}~\bibnamefont {Krsulich}}, \bibinfo {author} {\bibfnamefont {C.~J.}\ \bibnamefont {Wood}}, \bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Lishman}}, \bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Gacon}}, \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Martiel}}, \bibinfo {author} {\bibfnamefont {P.~D.}\ \bibnamefont {Nation}}, \bibinfo {author} {\bibfnamefont {L.~S.}\ \bibnamefont {Bishop}}, \bibinfo {author} {\bibfnamefont {A.~W.}\ \bibnamefont {Cross}}, \bibinfo {author} {\bibfnamefont {B.~R.}\ \bibnamefont {Johnson}}, \ and\ \bibinfo {author} {\bibfnamefont {J.~M.}\ \bibnamefont {Gambetta}},\ }\href {\doibase 10.48550/arXiv.2405.08810} {\enquote {\bibinfo {title} {Quantum computing with {Q}iskit},}\ } (\bibinfo {year} {2024}),\ \Eprint {http://arxiv.org/abs/2405.08810} {arXiv:2405.08810 [quant-ph]}
\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Ugander}\ \emph {et~al.}(2011)\citenamefont {Ugander}, \citenamefont {Karrer}, \citenamefont {Backstrom},\ and\ \citenamefont {Marlow}}]{facebook_anatomy}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Ugander}}, \bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Karrer}}, \bibinfo {author} {\bibfnamefont {L.}~\bibnamefont {Backstrom}}, \ and\ \bibinfo {author} {\bibfnamefont {C.}~\bibnamefont {Marlow}},\ }\href {https://arxiv.org/abs/1111.4503} {\enquote {\bibinfo {title} {The anatomy of the facebook social graph},}\ } (\bibinfo {year} {2011}),\ \Eprint {http://arxiv.org/abs/1111.4503} {arXiv:1111.4503 [cs.SI]} \BibitemShut {NoStop}%
\bibitem [{\citenamefont {Ren}(2008)}]{ren2008consensus}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {W.}~\bibnamefont {Ren}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {IEEE Transactions on Automatic Control}\ }\textbf {\bibinfo {volume} {53}},\ \bibinfo {pages} {1503} (\bibinfo {year} {2008})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Geva}\ and\ \citenamefont {Sitte}(2002)}]{geva2002cartpole}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Geva}}\ and\ \bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Sitte}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {IEEE Control Systems Magazine}\ }\textbf {\bibinfo {volume} {13}},\ \bibinfo {pages} {40} (\bibinfo {year} {2002})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Nüßlein}\ \emph {et~al.}(2024)\citenamefont {Nüßlein}, \citenamefont {Sünkel}, \citenamefont {Stein}, \citenamefont {Rohe}, \citenamefont {Schuman}, \citenamefont {Feld}, \citenamefont {O'Meara}, \citenamefont {Cortiana},\ and\ \citenamefont {Linnhoff-Popien}}]{AncillaQubitsRef}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Nüßlein}}, \bibinfo {author} {\bibfnamefont {L.}~\bibnamefont {Sünkel}}, \bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Stein}}, \bibinfo {author} {\bibfnamefont {T.}~\bibnamefont {Rohe}}, \bibinfo {author} {\bibfnamefont {D.}~\bibnamefont {Schuman}}, \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Feld}}, \bibinfo {author} {\bibfnamefont {C.}~\bibnamefont {O'Meara}}, \bibinfo {author} {\bibfnamefont {G.}~\bibnamefont {Cortiana}}, \ and\ \bibinfo {author} {\bibfnamefont {C.}~\bibnamefont {Linnhoff-Popien}},\ }\href {https://arxiv.org/abs/2412.17841} {\enquote {\bibinfo {title} {Reducing qubo density by factoring out semi-symmetries},}\ } (\bibinfo {year} {2024}),\ \Eprint {http://arxiv.org/abs/2412.17841} {arXiv:2412.17841 [quant-ph]} \BibitemShut {NoStop}%
\bibitem [{\citenamefont {Dattani}(2019)}]{dattani2019quadratizationdiscreteoptimizationquantum}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {N.}~\bibnamefont {Dattani}},\ }\href {https://arxiv.org/abs/1901.04405} {\enquote {\bibinfo {title} {Quadratization in discrete optimization and quantum mechanics},}\ } (\bibinfo {year} {2019}),\ \Eprint {http://arxiv.org/abs/1901.04405} {arXiv:1901.04405 [quant-ph]} \BibitemShut {NoStop}%
\bibitem [{\citenamefont {Jin}\ \emph {et~al.}(2025)\citenamefont {Jin}, \citenamefont {Huang}, \citenamefont {Wu}, \citenamefont {Zhou}, \citenamefont {Zou}, \citenamefont {Sun},\ and\ \citenamefont {Wang}}]{jin2025fixed}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Jin}}, \bibinfo {author} {\bibfnamefont {Y.}~\bibnamefont {Huang}}, \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Wu}}, \bibinfo {author} {\bibfnamefont {G.}~\bibnamefont {Zhou}}, \bibinfo {author} {\bibfnamefont {C.-L.}\ \bibnamefont {Zou}}, \bibinfo {author} {\bibfnamefont {L.}~\bibnamefont {Sun}}, \ and\ \bibinfo {author} {\bibfnamefont {X.}~\bibnamefont {Wang}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Science China Physics, Mechanics \& Astronomy}\ }\textbf {\bibinfo {volume} {68}},\ \bibinfo {pages} {1} (\bibinfo {year} {2025})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Arrazola}\ \emph {et~al.}()\citenamefont {Arrazola}, \citenamefont {Delgado}, \citenamefont {Bardhan},\ and\ \citenamefont {Lloyd}}]{arrazola2020}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {J.~M.}\ \bibnamefont {Arrazola}}, \bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Delgado}}, \bibinfo {author} {\bibfnamefont {B.~R.}\ \bibnamefont {Bardhan}}, \ and\ \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Lloyd}},\ }\href {\doibase 10.22331/q-2020-08-13-307} {\ \textbf {\bibinfo {volume} {4}},\ \bibinfo {pages} {307}},\ \Eprint {http://arxiv.org/abs/1905.10415} {1905.10415} \BibitemShut {NoStop}%
\bibitem [{\citenamefont {Sahay}\ \emph {et~al.}(2025)\citenamefont {Sahay}, \citenamefont {Lin}, \citenamefont {Huang}, \citenamefont {Brown},\ and\ \citenamefont {Puri}}]{sahay2025error}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {K.}~\bibnamefont {Sahay}}, \bibinfo {author} {\bibfnamefont {Y.}~\bibnamefont {Lin}}, \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Huang}}, \bibinfo {author} {\bibfnamefont {K.~R.}\ \bibnamefont {Brown}}, \ and\ \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Puri}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {PRX quantum}\ }\textbf {\bibinfo {volume} {6}},\ \bibinfo {pages} {020326} (\bibinfo {year} {2025})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Gachnang}\ \emph {et~al.}(2022)\citenamefont {Gachnang}, \citenamefont {Ehrenthal}, \citenamefont {Hanne},\ and\ \citenamefont {Dornberger}}]{gachnang2022quantum}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {P.}~\bibnamefont {Gachnang}}, \bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Ehrenthal}}, \bibinfo {author} {\bibfnamefont {T.}~\bibnamefont {Hanne}}, \ and\ \bibinfo {author} {\bibfnamefont {R.}~\bibnamefont {Dornberger}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Asian Journal of Logistics Management}\ }\textbf {\bibinfo {volume} {1}},\ \bibinfo {pages} {57} (\bibinfo {year} {2022})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Blenninger}\ \emph {et~al.}(2024)\citenamefont {Blenninger}, \citenamefont {Bucher}, \citenamefont {Cortiana}, \citenamefont {Ghosh}, \citenamefont {Mohseni}, \citenamefont {N{\"u}{\ss}lein}, \citenamefont {O’Meara}, \citenamefont {Porawski},\ and\ \citenamefont {Wimmer}}]{blenninger2024q}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Blenninger}}, \bibinfo {author} {\bibfnamefont {D.}~\bibnamefont {Bucher}}, \bibinfo {author} {\bibfnamefont {G.}~\bibnamefont {Cortiana}}, \bibinfo {author} {\bibfnamefont {K.}~\bibnamefont {Ghosh}}, \bibinfo {author} {\bibfnamefont {N.}~\bibnamefont {Mohseni}}, \bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {N{\"u}{\ss}lein}}, \bibinfo {author} {\bibfnamefont {C.}~\bibnamefont {O’Meara}}, \bibinfo {author} {\bibfnamefont {D.}~\bibnamefont {Porawski}}, \ and\ \bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Wimmer}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {KI-K{\"u}nstliche Intelligenz}\ }\textbf {\bibinfo {volume} {38}},\ \bibinfo {pages} {339} (\bibinfo {year} {2024})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Nayak}\ and\ \citenamefont {Wu}(1999)}]{nayak1999quantum}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Nayak}}\ and\ \bibinfo {author} {\bibfnamefont {F.}~\bibnamefont {Wu}},\ }in\ \href@noop {} {\emph {\bibinfo {booktitle} {Proceedings of the thirty-first annual ACM symposium on Theory of computing}}}\ (\bibinfo {year} {1999})\ pp.\ \bibinfo {pages} {384--393}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Montanaro}(2016)}]{montanaro2016quantum}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Montanaro}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {npj Quantum Information}\ }\textbf {\bibinfo {volume} {2}},\ \bibinfo {pages} {1} (\bibinfo {year} {2016})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Szab{\l}owski}(2021)}]{szablowski2021understanding}%
\BibitemOpen
\bibfield {author} {\bibinfo {author} {\bibfnamefont {P.~J.}\ \bibnamefont {Szab{\l}owski}},\ }\href@noop {} {\bibfield {journal} {\bibinfo {journal} {Quantum Information Processing}\ }\textbf {\bibinfo {volume} {20}},\ \bibinfo {pages} {191} (\bibinfo {year} {2021})}\BibitemShut {NoStop}%
\end{thebibliography}%