Single-Particle Discrete-Time Quantum Walk

• Single-particle discrete-time quantum walk is defined as a unitary evolution on a discrete lattice that combines a two-level coin and spatial degrees of freedom to generate coherent superpositions and interference.
• Its dynamics rely on repeated coin operations and conditional shifts, resulting in ballistic spreading, coin–position entanglement, and the simulation of continuum models like the Dirac equation.
• Experimental implementations across photonics, trapped ions, and superconducting systems showcase DTQWs as versatile platforms for quantum transport, computation, and metrological applications.

Searching arXiv for recent and foundational papers on single-particle discrete-time quantum walks. A single-particle discrete-time quantum walk (DTQW) is a unitary, stepwise evolution of one quantum particle on a discrete graph or lattice, with dynamics defined on a composite Hilbert space that combines a finite-dimensional internal degree of freedom—the coin—and a position degree of freedom. In its standard one-dimensional form, the walk is generated by repeated application of a coin operation and a conditional shift, producing coherent superpositions over paths, coin–position entanglement, ballistic spreading, and graph-dependent interference patterns. Across the literature, the single-particle DTQW functions simultaneously as a model of quantum transport, a simulator of continuum dynamics such as Dirac and Schrödinger equations, a substrate for universal quantum computation, and a platform for experimentally probing coherence, entanglement, and more recently quantum magic and sensing protocols (Shikano, 2013).

1. Formal structure and kinematics

The basic DTQW is defined on a tensor-product Hilbert space written either as $\mathcal{H}=\mathcal{H}_c\otimes\mathcal{H}_p$ or $\mathcal{H}=\mathcal{H}_p\otimes\mathcal{H}_c$, depending on convention. The coin space is two-dimensional, with basis states such as $\{|0\rangle,|1\rangle\}$, $\{|L\rangle,|R\rangle\}$, or $\{|\uparrow\rangle,|\downarrow\rangle\}$, while the position space is spanned by lattice-site states $\{|n\rangle:n\in\mathbb{Z}\}$ on a line or by vertex states on finite graphs and cycles. A generic state is a superposition of coin–position basis vectors, and the measured position distribution at step $t$ is obtained by summing coin-resolved probabilities at each site (Shikano, 2013).

One step of the walk is a unitary of the form

$U = S\,(C\otimes I_p),$

or equivalently $U=T\cdot S$ in notations where $S$ denotes the coin rotation and $\mathcal{H}=\mathcal{H}_p\otimes\mathcal{H}_c$0 the conditional translation. The coin is commonly chosen from a general SU(2) family, and special choices recover familiar single-qubit gates such as the Hadamard coin. The conditional shift moves the walker left or right according to the coin state; in the standard one-dimensional formulation, amplitudes in $\mathcal{H}=\mathcal{H}_p\otimes\mathcal{H}_c$1 shift to $\mathcal{H}=\mathcal{H}_p\otimes\mathcal{H}_c$2 and amplitudes in $\mathcal{H}=\mathcal{H}_p\otimes\mathcal{H}_c$3 shift to $\mathcal{H}=\mathcal{H}_p\otimes\mathcal{H}_c$4. This coin-controlled transport is the elementary mechanism behind all subsequent interference, spreading, and computational behavior (Su et al., 2022).

A particularly useful formulation employs the most general coin

$\mathcal{H}=\mathcal{H}_p\otimes\mathcal{H}_c$5

which embeds the Hadamard, Grover-type, and Fourier coins as special cases. In this parametrization, $\mathcal{H}=\mathcal{H}_p\otimes\mathcal{H}_c$6 controls the amount of mixing between coin states, $\mathcal{H}=\mathcal{H}_p\otimes\mathcal{H}_c$7 controls asymmetry in the walker’s probability distribution, and $\mathcal{H}=\mathcal{H}_p\otimes\mathcal{H}_c$8 does not affect the probability distribution even though it appears in the wavefunction. This separation of roles is central to parameter engineering in one-dimensional DTQWs (Jayakody et al., 2021).

The same kinematic template extends beyond the infinite line. Single-particle DTQWs have been formulated on finite cycles, four-site closed graphs, body-centered cubic lattices, honeycomb lattices, triangular lattices, and tetrahedral tessellations. In each case, the internal space remains finite-dimensional and the one-step rule remains local and unitary, but the geometry of the underlying graph determines which conditional translations are available and which continuum limits emerge (D'Ariano et al., 2017).

2. Spreading, asymptotics, and localization

The most basic dynamical distinction between a single-particle DTQW and a classical random walk is the scaling of spread. For homogeneous, time-independent walks on the line, the position random variable $\mathcal{H}=\mathcal{H}_p\otimes\mathcal{H}_c$9 obeys a weak limit theorem due to Konno: $\{|0\rangle,|1\rangle\}$0 with a non-Gaussian limiting density supported on $\{|0\rangle,|1\rangle\}$1, where $\{|0\rangle,|1\rangle\}$2 is a coin parameter. The limiting law is inverted-bell-shaped, typically with enhanced weight near the edges of its support, and the variance scales as $\{|0\rangle,|1\rangle\}$3. This ballistic regime contrasts with the classical diffusive scaling $\{|0\rangle,|1\rangle\}$4 and underlies the transport advantage of the quantum walk (Shikano, 2013).

The asymptotic distribution is not determined by the coin alone. A common misconception is that a symmetric coin automatically yields a symmetric probability profile. In fact, spatial symmetry depends jointly on the coin and the initial coin state. For the Hadamard walk, $\{|0\rangle,|1\rangle\}$5 at the origin gives strong skewness, whereas $\{|0\rangle,|1\rangle\}$6 can yield a symmetric distribution. In the general-coin formulation, $\{|0\rangle,|1\rangle\}$7 acts as a bias-control parameter even for an unbiased initial coin state, while $\{|0\rangle,|1\rangle\}$8 determines the effective support width and hence the maximal propagation speed (Jayakody et al., 2021).

Analytically, single-particle DTQWs admit multiple exact treatments. Fourier analysis and spectral decomposition yield moments, characteristic functions, and weak limits in the homogeneous case. A complementary path-integral formulation rewrites the amplitude at a given coin–position basis state as a sum over forward/backward path strings, classified by the number of switches and associated parity factors. This provides closed-form amplitudes for arbitrary step number and endpoint, and makes the interference structure explicit at the level of individual paths (Joshi et al., 2018).

Localization introduces a qualitatively different regime. On a homogeneous line, the walk spreads ballistically and the return probability at the origin tends to zero. But a single-site phase defect can create exponentially localized stationary states of the two-step evolution operator $\{|0\rangle,|1\rangle\}$9. In the defect model, each traversal of $\{|L\rangle,|R\rangle\}$0 contributes a phase $\{|L\rangle,|R\rangle\}$1, producing up to two localized eigenstates $\{|L\rangle,|R\rangle\}$2 with decay parameters $\{|L\rangle,|R\rangle\}$3. Their existence depends on $\{|L\rangle,|R\rangle\}$4, and the long-time localized fraction depends on the overlap of the initial coin state $\{|L\rangle,|R\rangle\}$5 with these bound states through factors proportional to $\{|L\rangle,|R\rangle\}$6. This shows that a single defect, not only extended disorder, can induce coin-selective trapping (Wojcik et al., 2011).

3. Continuum, relativistic, and geometric limits

A major line of research treats the single-particle DTQW as a quantum dynamical simulator of continuum physics. In one dimension, suitable scaling of the coin can produce the $\{|L\rangle,|R\rangle\}$7-dimensional Dirac equation. With a coin

$\{|L\rangle,|R\rangle\}$8

and a quasi-continuum limit where $\{|L\rangle,|R\rangle\}$9 is identified with lattice spacing, the walk approximates

$\{|\uparrow\rangle,|\downarrow\rangle\}$0

which is the $\{|\uparrow\rangle,|\downarrow\rangle\}$1-dimensional Dirac equation in spinor form. This observation places the DTQW above both relativistic and non-relativistic lattice dynamics in a continuum hierarchy (Shikano, 2013).

The same review establishes a bridge from DTQW to continuous-time quantum walk (CTQW). In the final-time-dependent DTQW, the coin depends on the total number of steps $\{|\uparrow\rangle,|\downarrow\rangle\}$2 through a parameter $\{|\uparrow\rangle,|\downarrow\rangle\}$3. Under the scaling $\{|\uparrow\rangle,|\downarrow\rangle\}$4, the walk becomes asymptotically equivalent to a linear combination of CTQWs with hopping amplitudes $\{|\uparrow\rangle,|\downarrow\rangle\}$5. By tuning $\{|\uparrow\rangle,|\downarrow\rangle\}$6, one obtains a crossover from standard ballistic DTQW behavior ($\{|\uparrow\rangle,|\downarrow\rangle\}$7) to CTQW-like arcsine distributions ($\{|\uparrow\rangle,|\downarrow\rangle\}$8) and then to localization ($\{|\uparrow\rangle,|\downarrow\rangle\}$9) (Shikano, 2013).

These continuum constructions extend to non-square geometries. On honeycomb and triangular lattices, a single-particle DTQW with local unitaries can reproduce the Dirac equation in $\{|n\rangle:n\in\mathbb{Z}\}$0 dimensions. The construction rewrites the kinetic term as directional momentum components along lattice axes and realizes them by sequences of conditional shifts and fixed coins adapted to the lattice symmetry. A notable implication is that simulating Dirac dynamics need not rely on square grids; the triangular formulation also suggests a route toward discrete surfaces and curved geometries (Arrighi et al., 2018).

An even more geometric extension places the walk on tetrahedral tessellations of three-dimensional space. There, a four-component state is associated with tetrahedral facets, and local shift-plus-coin rules yield a continuum limit equal to the $\{|n\rangle:n\in\mathbb{Z}\}$1-dimensional Dirac equation. The same framework admits an interpretation on the dual 4-valent graph of the tetrahedral complex, making contact with spin-network-like structures and suggesting an ordered scheme for propagating matter over a spin network (Nzongani et al., 2024).

Curved-spacetime transport appears in a distinct but related line of work on discrete metrics. In the $\{|n\rangle:n\in\mathbb{Z}\}$2-dimensional massless case, grouped and encoded QWs yield the scalar transport equation

$\{|n\rangle:n\in\mathbb{Z}\}$3

with the local speed $\{|n\rangle:n\in\mathbb{Z}\}$4 dictated by the local coin. The notable feature is that this tunable speed can be implemented using only a finite number of coin operators, rather than a continuous coin family. This suggests a discrete representation of the metric field in which geometry is encoded by a finite gate alphabet and its spacetime pattern (Arrighi et al., 2017).

4. Single-particle DTQW as a computation model

Single-particle DTQWs are not only simulators of transport but also models of universal quantum computation. One route encodes logical qubits into the coin and the position space of a single walker. In the one-dimensional construction for two- and three-qubit systems, the coin encodes the first logical qubit and a two-site or four-site position graph encodes the remaining qubits. Phase and Hadamard gates on the first qubit are direct coin operations, while gates on the position-encoded qubits are realized by position-dependent coin operations and conditional shifts. Within this encoding, the universal gate set $\{|n\rangle:n\in\mathbb{Z}\}$5 is realized, and nontrivial gates such as Toffoli and controlled-$\{|n\rangle:n\in\mathbb{Z}\}$6 are also implemented (Singh et al., 2019).

A more scalable variant employs closed graphs. The Hilbert space is $\{|n\rangle:n\in\mathbb{Z}\}$7, with one physical two-level coin and a composite position space formed from multiple closed graphs. For odd $\{|n\rangle:n\in\mathbb{Z}\}$8, the construction uses $\{|n\rangle:n\in\mathbb{Z}\}$9 four-site closed graphs; for even $t$0, it uses $t$1 four-site graphs plus one two-site graph. Each 4-site graph encodes two logical qubits, and the total position basis is a tensor product of these graph-local position states. Position-selective walk operators built from Pauli operations, conditional shifts, and projectors then realize Hadamard, phase, and CNOT gates on arbitrary logical qubits (Chawla et al., 2020).

Within this closed-graph model, standard quantum algorithms are mapped to sequences of single-particle DTQW steps. Grover’s search on three logical qubits is constructed from DTQW-based Hadamards, a position-dependent oracle, and a diffusion operator. Quantum Fourier transform is assembled from controlled-swap and controlled-phase primitives derived from the walk. Phase estimation uses the coin as an eigenstate register and the position space as a control register, with controlled-$t$2 powers implemented as position-dependent coin operations followed by inverse QFT in the walk model. The same work also presents an elementary implementation of error detection and correction, together with an analysis of space and time complexity (Chawla et al., 2020).

From a circuit-compilation perspective, efficient implementations matter. For a Hadamard DTQW on the $t$3-cycle, diagonalization of the conditional shift operator yields a scalable circuit with only $t$4 two-qubit gates for $t$5 time steps, compared to $t$6 for the most efficient implementation based on quantum Fourier transforms. In the Fourier basis, the shift becomes diagonal, so the QFT is applied only at the beginning and the end, while each walk step reduces to a coin and a layer of controlled phases. This directly targets the main obstacle in NISQ implementations: excessive two-qubit gate count and depth (Razzoli et al., 2024).

5. Physical implementations and experimental realizations

The physical realization of single-particle DTQWs traditionally uses distinct degrees of freedom for coin and walker. Photonic systems, trapped ions, optical lattices, and superconducting architectures all instantiate the same logical structure: a two-level coin and a multi-site walker. A particularly unconventional implementation encodes both coin and walker into the two-dimensional Hilbert space of a single qubit. In that “one-qubit” approach, a logical DTQW state

$t$7

is mapped to

$t$8

so the walker position is stored in phase factors while the qubit basis encodes the coin. The coin becomes a single-qubit $t$9-rotation and the shift becomes a $U = S\,(C\otimes I_p),$0-phase gate (Su et al., 2022).

That single-qubit encoding has been implemented experimentally with single photons. The physical qubit is polarization, $U = S\,(C\otimes I_p),$1 and $U = S\,(C\otimes I_p),$2, and one DTQW step is realized as

$U = S\,(C\otimes I_p),$3

using a QWP–HWP–QWP sequence. A 7-step one-particle DTQW was implemented in this way, and the full logical coin+walker state was reconstructed by repeated single-qubit tomography with different phase encodings. This gave access not just to marginal position distributions but to the complete high-dimensional state of the logical DTQW (Su et al., 2022).

On quantum computers, the optimized $U = S\,(C\otimes I_p),$4-cycle circuit has been tested on an IBM device for Hadamard DTQWs on the 4- and 8-cycle. These walks show periodic dynamics and recurrent generation of maximally entangled single-particle states between the coin and position degrees of freedom. Because the circuit uses the QFT only once at the beginning and once at the end, experimental results remain meaningful well beyond the regime of very few time steps, and the Hellinger fidelity of the measured position distributions remains substantial over many steps (Razzoli et al., 2024).

These implementations clarify that “single-particle” refers to the number of walkers, not to the dimensionality of the effective state space. A single walker can occupy a large effective Hilbert space through coin–position structure, and physical encodings may separate or merge those degrees of freedom. This suggests that resource counting in DTQW architectures must distinguish physical qubits from logical dimension; a plausible implication is that different platforms may optimize very different hardware bottlenecks while realizing the same logical walk.

6. Correlations, magic, and metrological applications

Once the full state of a single-particle DTQW is accessible, the walk becomes a laboratory for information-theoretic resources. In the one-qubit photonic implementation, the reconstructed pure joint state of coin and walker after 7 steps allowed direct evaluation of quantum mutual information, entanglement, relative entropy of coherence, and correlated coherence. For initial coin states

$U = S\,(C\otimes I_p),$5

the measured ordering

$U = S\,(C\otimes I_p),$6

held across $U = S\,(C\otimes I_p),$7, while the coherences satisfied

$U = S\,(C\otimes I_p),$8

All correlation and coherence measures reached extremal values near $U = S\,(C\otimes I_p),$9, where the initial coin is maximally coherent, and their behavior tracked the position variance $U=T\cdot S$0 (Su et al., 2022).

A recent extension studies non-stabilizer resources in DTQWs through the Stabilizer Rényi Entropy (SRE). For a single walker on a one-dimensional lattice with Hadamard coin, the reduced coin state $U=T\cdot S$1 is used to define magic. Even when the initial coin state is a stabilizer state such as $U=T\cdot S$2, the walk generates non-zero magic after a few steps through coin–position entanglement and interference. The long-time relation between coin magic and coin–position entanglement is not monotone: the two exhibit complementary patterns, so highly entangled reduced coin states need not be highly magical, and coin states with SRE near the single-qubit upper bound $U=T\cdot S$3 need not maximize entanglement (Mittal et al., 21 Jun 2025).

This corrects another common misconception: Clifford walk primitives do not imply trivial resource generation. In the magic analysis, the Hadamard coin is itself a Clifford gate, yet the reduced coin state can become non-stabilizer under the full DTQW dynamics because interference in the enlarged Hilbert space reorganizes the resource content of the subsystem (Mittal et al., 21 Jun 2025).

Single-particle DTQWs also support metrological protocols. In quantum magnetometry using a spin-$U=T\cdot S$4 walker on a bounded one-dimensional lattice, a homogeneous magnetic field modifies the effective coin, and the field strength is estimated from the resulting position and spin statistics. The position variance identifies directions of maximal influence, while the quantum Fisher information and Fisher information quantify sensitivity. For an electron implementing a 50-step walk, the reported estimate had a root-mean-square error of the order of 0.1 picoTesla, and the sensitivity could be tuned to measure a desired magnetic field by adjusting the walk parameters (Shukla et al., 2023).

Across these applications, the single-particle DTQW appears less as a narrowly defined transport model than as a structured interferometric engine. It generates coin–position entanglement, coherence, and magic; it supports controllable bias, localization, and continuum limits; and, through suitable encodings, it realizes both computation and sensing. This suggests that the unifying object is not a specific lattice or coin, but a local unitary architecture in which one walker explores a graph while internal and external degrees of freedom are continually re-entangled.