Quantum Walk Encodings via Linear Optics

• Quantum walk encodings via linear optics are techniques that map photonic degrees of freedom to walker and coin states using time-bin, spatial mode, and OAM implementations.
• They utilize programmable, passive optical elements to execute discrete-time quantum walks with tunable coin and shift operations for high fidelity.
• Recent advances demonstrate scalable and reconfigurable platforms that achieve low-loss, high-dimensional quantum simulations and topologically nontrivial state engineering.

Quantum walk encodings via linear optics are a central paradigm in photonic quantum information processing, enabling the simulation, measurement, and engineering of high-dimensional quantum systems using only passive and/or programmable linear optical elements. These protocols leverage various optical degrees of freedom—such as time bins, spatial modes, orbital angular momentum (OAM), frequency, and polarization—to efficiently represent the quantum walk position ("walker") and coin degrees of freedom, and to implement coin and shift dynamics. Recent developments have demonstrated high-fidelity, scalable, and reconfigurable linear-optical platforms capable of executing large-scale, multi-dimensional, and topologically nontrivial quantum walks, as well as advanced measurement and state engineering tasks.

1. Fundamental Principles of Linear-Optical Quantum Walk Encodings

Linear-optical quantum walk encodings map the walker’s position and coin Hilbert spaces onto photonic degrees of freedom, implementing the discrete-time coined-walk model through interferometric, polarization, or time-bin control.

1.1. Position (Walker) Encoding

• Time-bin encoding: The spatial position is mapped to the arrival time of an optical pulse within coupled fiber cavities. Each basis state $|t_n\rangle$ represents the $n$-th time slot, with the slot length set by the difference in cavity delays ($\tau$) (Boutari et al., 2016).
• Transverse spatial modes: Distinct walk positions are realized as separate spatial beam paths labeled by $x \in \mathbb{Z}$, using birefringent beam displacers or waveguide circuits (Fang et al., 2022).
• OAM encoding: The walker state utilizes the OAM eigenmode $|m\rangle$ (the "twisted photon"), mapping position to the integer OAM quantum number. All positions are contained in a single collinear paraxial beam (Cardano et al., 2014, Innocenti et al., 2017).
• Multi-DOF encoding for multidimensional walks: Each spatial lattice direction is encoded in a distinct optical DOF, e.g., OAM for $x$, time-bin for $y$, frequency for $z$ axes. This architecture supports large $d$-dimensional walks (Goyal et al., 2015).

1.2. Coin Encoding and Operation

• Polarization as coin: The binary coin is represented as $|H\rangle \equiv |0\rangle$ and $n$0 or circularly polarized $n$1, $n$2 states (Cardano et al., 2014, Fang et al., 2022).
• Path encoding: For multidimensional coins ($n$3), spatial paths or additional DOFs encode each coin subspace (Goyal et al., 2015).
• Tunable unitary operation: The coin operation is realized by passive elements—fiber couplers with variable reflectivity (for time-bin), birefringent waveplates (for polarization), or multiport interferometric devices (for multidegree coins), supporting arbitrary SU(2) or U($n$4) transformations (Boutari et al., 2016, Osawa et al., 2018).

1.3. Shift and Step Operators

• The shift operator $n$5 acts conditionally on the coin state, advancing or delaying (in time-bin encoding), routing spatial modes (in spatial mode encoding), or incrementing/decrementing OAM (in OAM encoding). The single-step operator is $n$6 (Boutari et al., 2016, Cardano et al., 2014, Fang et al., 2022).
• In directionally-unbiased multiports, a single optical unit implements the combined coin and shift (Osawa et al., 2018).

2. Linear-Optical Implementations: Architectures and Methods

2.1. Time-Bin Encoded Walks in Fiber Cavities

• Architecture: Coupled fiber cavities with distinct round-trip delays $n$7, $n$8 enable mapping of walker's position onto time-bins, with coin operation by a variable fiber coupler ($n$9) (Boutari et al., 2016).
• Scalability: Loss per round-trip ($\tau$0 dB) allows $\tau$162 steps with $\tau$2 fidelity. Multi-particle injection and multidimensional lattice extension are enabled by multiplying loop number (Boutari et al., 2016).

2.2. Integrated Spatial-Path and Polarization Walks

• Spatial modes: A multi-stage interferometer using beam displacers for conditional spatial translation and waveplates for coin operations carries out the walk (Fang et al., 2022).
• Arbitrary coin control: Stepwise tuning of waveplate orientation implements arbitrary coin sequences (e.g., Hadamard, identity), supporting programmable and optimal entanglement generation (Fang et al., 2022).

2.3. OAM-Based and Multi-DOF Encodings

• Single-beam OAM walks: No interferometers, single collinear beam, each step realized as a waveplate–q-plate–waveplate sequence; OAM increment/decrement acts as shift conditional on polarization (Cardano et al., 2014, Innocenti et al., 2017).
• State engineering: Any $\tau$3-level qudit can be realized as a walker superposition via $\tau$4 sequential steps, with coin dynamically programmed using a waveplate sequence (Innocenti et al., 2017).

2.4. Multiport and Directionally Unbiased Networks

• Directionally-unbiased $\tau$5-ports implement reversible coin+shift unitaries on graphs; resource count only scales linearly with the number of vertices, crucial for topological and many-body quantum simulation (Osawa et al., 2018, Simon et al., 2017).

2.5. Quantum Walks for Generalized Measurement

• Quantum walks are employed as programmable generators of arbitrary POVMs (e.g., semi-SIC), with site-dependent coins and spatially resolved measurements mapping to the required generalized measurement outcomes (Xu et al., 2 Mar 2026).

3. Performance Metrics, Fidelity, and Experimental Benchmarks

3.1. Loss, Transmission, and Step Count

• Time-bin platforms: Transmission after $\tau$6 steps: $\tau$7, $\tau$8 typically 0.11 (fractional), supporting $\tau$9 steps with high fidelity (Boutari et al., 2016).
• OAM and spatial mode platforms: Insertion losses per q-plate $x \in \mathbb{Z}$0–$x \in \mathbb{Z}$1, overall fidelity $x \in \mathbb{Z}$2 for $x \in \mathbb{Z}$3 steps in bulk setups (Cardano et al., 2014). Integrated devices can potentially support many more steps with on-chip optics (Osawa et al., 2018).

3.2. Fidelity with Ideal Quantum Walk

• Statistical fidelity $x \in \mathbb{Z}$4 remains $x \in \mathbb{Z}$5 up to $x \in \mathbb{Z}$6 steps in fiber architectures (Boutari et al., 2016).
• OAM and path-encoded state-fidelity typically exceeds 0.99 for 10–16 dimensional reconstructed states (Su et al., 2018).

3.3. Quantum Correlations and Entanglement

• Maximal coin-position entanglement (entropy $x \in \mathbb{Z}$7) at $x \in \mathbb{Z}$8 steps can be achieved universally via appropriate coin programming; process fidelity for implemented coin channel $x \in \mathbb{Z}$9 (Fang et al., 2022).

3.4. Programmability and Versatility

• Walkers initialized in arbitrary superpositions using compact single-photon encoding schemes support full tomography, large-dimensional state preparation, and controllable transport dynamics with experimental fidelities $|m\rangle$0 (Su et al., 2018).

4. Resource Scaling, Dimensionality, and Extensions

4.1. Resource Comparison and Scalability

Architecture	Element Scaling	Notes
Multi-path interferometer (BS-trees)	$|m\rangle$1	Exponential in steps, severe hardware footprint
Time-bin fiber cavity, OAM, spatial	$|m\rangle$2	Linear in steps $|m\rangle$3, high stability
Multiport/Unbiased multiport networks	$|m\rangle$4	$|m\rangle$5 number of sites, efficient for complex graphs
Classical light/DOF-based multidim walks	$|m\rangle$6	$|m\rangle$7 lattice dimension (multidimensional walks)

4.2. Multidimensional and Topological Walks

• Multi-DOF encoding: OAM, time-bin, and frequency shifts support up to $|m\rangle$8–$|m\rangle$9 dimensions; coin is realized by integrating coin qubits in multiple paths with global U($x$0) operations (Goyal et al., 2015).
• Topological simulation: Directionally-unbiased multiports support lattice walks with nontrivial band topology, e.g., SSH Hamiltonian simulation with topological boundary states and winding numbers (Simon et al., 2017).
• Tensor network simulations: Cascaded multiport networks implement walks over arbitrary graph or tensor structures with efficient linear optics resource scaling (Osawa et al., 2018).

5. State Engineering, Quantum Information Tasks, and Advanced Encodings

5.1. High-Dimensional State Preparation

• Arbitrary $x$1-level superpositions (qudit states) can be prepared as quantum walk output by inverse engineering the coin sequence for reachability, absorbing boundary conditions, and local amplitude control in the coin (Innocenti et al., 2017).
• Encoding arbitrary initial walker states in a single qubit (polarization) via nonorthogonal basis mapping and repeated tomographic protocols achieves high-fidelity, resource-efficient qudit embedding (Su et al., 2018).

5.2. Quantum Measurement and POVM Implementation

• Discrete-time quantum walk networks serve as programmable engines for generalized qubit POVMs (including semi-SIC), mapping quantum-walk outputs to measurement outcomes. Semi-device-independent certification of these POVMs is achievable by prepare-and-measure protocols fully compatible with photonic platforms (Xu et al., 2 Mar 2026).
• Number of POVM outcomes scales as $x$2 for an $x$3-step walk, enabling complex measurement design.

5.3. Harmonic Oscillator Encodings via Quantum Walk

• Grid-state qubits (approximate Gottesman-Kitaev-Preskill codewords) are encoded via quantum walks in phase space using ancilla cat states as coins and conditional Mach–Zehnder displacement circuits, with post-selection ensuring high-fidelity grid state formation. Resource requirements and success probabilities are compatible with near-term photonic and superconducting platforms (Wu et al., 2024).

6. Experimental Demonstrations and Notable Results

6.1. Large-Scale, Low-Loss Walks

• Time-bin encoding in fiber supports over 62 walk steps with per-step loss $x$4 (fractional), achieving $x$5 and robust coin programmability (biases 0.2, 0.5, 0.8; coin angle $x$6) (Boutari et al., 2016).

6.2. Maximal Entanglement Protocols

• Universal coin-position entanglement at all $x$7 via optimized Hadamard/identity coin sequences, confirmed via full state/process tomography in spatial-path linear-optical networks; entanglement entropy $x$8 (Fang et al., 2022).

6.3. Programmable POVM Engine

• Realization and self-testing of four-outcome (semi-SIC) POVMs via two- and five-step quantum walks in linear optics; semi-device-independent certification via prepare-and-measure witness exceeding quantum-optimal bounds (Xu et al., 2 Mar 2026).

6.4. Topological States and Quantum Transport

• Simulation of SSH and higher-dimensional Hamiltonians, demonstration of topologically protected boundary states, and resource scaling quadratic improvement ($x$9vs$y$0) in multiport platforms (Simon et al., 2017, Osawa et al., 2018).

7. Outlook and Future Directions

Linear-optical quantum-walk encodings provide a highly versatile, scalable, and programmable approach to the simulation, measurement, and control of complex quantum systems. The demonstrated ability to engineer multidimensional walks, programmable coin operations, high-fidelity state preparation, and generalized quantum measurements with efficient resource scaling positions these platforms as key enablers for photonic quantum simulation, topological state engineering, high-dimensional quantum information processing, and quantum metrology. Ongoing integration with on-chip platforms, enhanced photon source/detector technologies, and multi-DOF control will broaden the accessible complexity and algorithmic space, fostering further advances in quantum technologies (Boutari et al., 2016, Cardano et al., 2014, Fang et al., 2022, Osawa et al., 2018, Innocenti et al., 2017, Goyal et al., 2015, Xu et al., 2 Mar 2026, Wu et al., 2024, Simon et al., 2017).