CoChat

Fast escape of a quantum walker from an integrated photonic maze

Open access

The DOI matches its Crossref record; 61 of the 61 references it cites were found in a registry or an index; 8 of the 14 sources its claims cite were checked for support.

Abstract

Escaping from a complex maze, by exploring different paths with several decision-making branches in order to reach the exit, has always been a very challenging and fascinating task. Wave field and quantum objects may explore a complex structure in parallel by interference effects, but without necessarily leading to more efficient transport. Here, inspired by recent observations in biological energy transport phenomena, we demonstrate how a quantum walker can efficiently reach the output of a maze by partially suppressing the presence of interference. In particular, we show theoretically an unprecedented improvement in transport efficiency for increasing maze size with respect to purely quantum and classical approaches. In addition, we investigate experimentally these hybrid transport phenomena, by mapping the maze problem in an integrated waveguide array, probed by coherent light, hence successfully testing our theoretical results. These achievements may lead towards future bio-inspired photonics technologies for more efficient transport and computation.

Overview

CoChat Summary

TL;DR

This paper theoretically and experimentally shows that partially suppressing interference in a quantum walker can improve escape/transport efficiency through maze-like graphs. In an 18-site integrated photonic waveguide maze, controlled noise increased measured transfer efficiency from 12.5% (uniform) to 22.2% at Δβmax=0.40 mm−1, while theory predicts >5 orders of magnitude enhancement for ~1000-site mazes at p≈0.1.

Key findings

  • Theoretically, for a maze of about one thousand sites, transfer efficiency is more than five orders of magnitude larger when interference is partially suppressed (p≈0.1, ∼10% mixing) than in the purely coherent and fully classical limits.
  • Experimentally, for an 18-site maze at t=60 mm, transfer efficiency was 12.5% for the uniform case, compared with average values of 14.1% for Δβmax=0.12 mm−1 and 22.2% for Δβmax=0.40 mm−1.
  • Experimental curves showed very good agreement with theoretical predictions for both the noiseless fully coherent case and the partially coherent transport when averaging noisy implementations.
  • The sink was implemented as a long chain of 62 waveguides, approximating a one-way irreversible energy transfer process.

Methods

  • Maze structures were generated with a random Depth-First Search algorithm on a square lattice; transport was modeled as a quantum stochastic walk via a Lindblad master equation mixing coherent Hamiltonian dynamics and classical random-walk operators through parameter p, with an irreversible sink at the exit.
  • The photonic simulator used femtosecond-laser-written three-dimensional waveguide arrays in EAGLE2000 glass; 24 structures implemented six propagation lengths for the noiseless case and three noise configurations, with single-mode waveguides at 850 nm, coupling κ=0.40 mm−1, Δβmax=0.40 mm−1, writing speed varied in the 10–40 mm s−1 range, and propagation constant modified every 3 mm. Additional t=60 mm structures included one uniform and 16 noisy implementations (8 at Δβmax=0.12 mm−1 and 8 at Δβmax=0.40 mm−1).
  • Structures were probed with coherent 850 nm light; output distributions were imaged onto a CMOS camera, and light intensity in maze and sink regions was numerically integrated to estimate transfer efficiency, with error contributions from losses and camera discretization analyzed.

Limitations

  • Stated in the paper: mapping arbitrary maze topology onto a waveguide array may not be possible if the maze graph is too complex, and the unfolded geometry is not unique. Noise-induced additional losses in maze waveguides cause overestimation of transfer efficiency; the precise contribution per measurement is hard to retrieve, though simulated systematic error was <3% for 60 mm. The experimental study focused on an 18-site maze as a significant example.
  • Evident from the text: the >5 orders of magnitude enhancement is theoretical for ~1000-site mazes and was not experimentally tested; experiments use coherent light to emulate the single-photon probability distribution; only one maze layout and limited noise strengths were experimentally explored. Data are available from the corresponding authors upon request.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/ncomms11682

Cites

61 of 61 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Maze problem. Pictorial view of a maze with single input (IN) and output (OUT) ports. An ideal walker has to travel from IN to OUT in the shortest possible time.
Figure 1 Maze problem. Pictorial view of a maze with single input (IN) and output (OUT) ports. An ideal walker has to travel from IN to OUT in the shortest possible time.CC BY · PubMed Central
Figure 2 Transport efficiency for different sizes. Left: Transfer efficiency as a function of the size N of the maze, for a time scale t linearly increasing with N, that is, t=10 N. For a maze with N=900 nodes, the optimal mixing p≃0.1 provides a transfer efficiency that is about five orders of magnitude larger than the perfectly coherent (quantum, that is, p=0) and fully noisy (classical, that is, p=1) regimes. The trend of the curves with the maze complexity N, for the different values of p, indicates that even higher speedup can be achieved for increasingly larger mazes. Right: Dwelling time as a function of the size N of the maze, for p=0, 0.1, 1.
Figure 2 Transport efficiency for different sizes. Left: Transfer efficiency as a function of the size N of the maze, for a time scale t linearly increasing with N, that is, t=10 N. For a maze with N=900 nodes, the optimal mixing p≃0.1 provides a transfer efficiency that is about five orders of magnitude larger than the perfectly coherent (quantum, that is, p=0) and fully noisy (classical, that is, p=1) regimes. The trend of the curves with the maze complexity N, for the different values of p, indicates that even higher speedup can be achieved for increasingly larger mazes. Right: Dwelling time as a function of the size N of the maze, for p=0, 0.1, 1.CC BY · PubMed Central
Figure 3 Implementing the maze. (a) Maze structure that is experimentally implemented; (b) unfolding of the maze into an almost linear graph, where each node is represented by a wave guide; (c,d) snapshots of the light diffusion for uniform (c) and noisy structure (d). The latter pictures correspond both to a propagation length of 60 mm. The noisy configuration is noise 3 in Fig. 4.
Figure 3 Implementing the maze. (a) Maze structure that is experimentally implemented; (b) unfolding of the maze into an almost linear graph, where each node is represented by a wave guide; (c,d) snapshots of the light diffusion for uniform (c) and noisy structure (d). The latter pictures correspond both to a propagation length of 60 mm. The noisy configuration is noise 3 in Fig. 4.CC BY · PubMed Central
Figure 4 Role of noise in the transfer efficiency time evolution. Left: Theoretical behaviour of the transfer efficiency as a function of the evolution parameter t for two values of the mixing parameter p, corresponding to QW (p=0) and QSW (p=0.1), for the maze in Fig. 3. Inset: a larger time scale is shown in order to point out the remarkable efficiency enhancement when there is a partial suppression of interference. Centre: Experimental results for both uniform (triangles) and three noisy realizations of the structure reported in Fig. 3. Where not shown the error bars are smaller than the mark size. Right: as in the middle panel, but only with the uniform case (triangles) and with the average efficiency over the noise realizations (circles). In all three panels, the time t is in units of mm because it is experimentally mapped into the propagation length of the three-dimensional waveguide array.
Figure 4 Role of noise in the transfer efficiency time evolution. Left: Theoretical behaviour of the transfer efficiency as a function of the evolution parameter t for two values of the mixing parameter p, corresponding to QW (p=0) and QSW (p=0.1), for the maze in Fig. 3. Inset: a larger time scale is shown in order to point out the remarkable efficiency enhancement when there is a partial suppression of interference. Centre: Experimental results for both uniform (triangles) and three noisy realizations of the structure reported in Fig. 3. Where not shown the error bars are smaller than the mark size. Right: as in the middle panel, but only with the uniform case (triangles) and with the average efficiency over the noise realizations (circles). In all three panels, the time t is in units of mm because it is experimentally mapped into the propagation length of the three-dimensional waveguide array.CC BY · PubMed Central
Figure 5 Averaging over different noise realizations. Experimentally measured transfer efficiency in photonic maze structures with the topology of Fig. 3 and propagation length t=60 mm. Both uniform and different noise maps with Δβmax=0.12 mm−1 and Δβmax=0.40 mm−1 have been implemented. The light grey (left) bars indicate the average over several experimental results obtained for different noise maps with the same amplitude Δβmax (single measurements are reported as open circles). The red (right) bars represent the expected average transfer efficiency, numerically simulated for 1,000 different noise maps with the given Δβmax. Error bars are smaller than the mark size.
Figure 5 Averaging over different noise realizations. Experimentally measured transfer efficiency in photonic maze structures with the topology of Fig. 3 and propagation length t=60 mm. Both uniform and different noise maps with Δβmax=0.12 mm−1 and Δβmax=0.40 mm−1 have been implemented. The light grey (left) bars indicate the average over several experimental results obtained for different noise maps with the same amplitude Δβmax (single measurements are reported as open circles). The red (right) bars represent the expected average transfer efficiency, numerically simulated for 1,000 different noise maps with the given Δβmax. Error bars are smaller than the mark size.CC BY · PubMed Central

Verification

Claims the paper makes

4 Valid · 1 Not valid · 4 Not checked · 30 Own results and interpretations

14 more claims

Create a free account to see all 39 claims and to verify the 6 references still open.

See all claims — free account
  • Probably not a substantial claim. 11
  • It may state two separate claims as one. 13

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Citations

Cites

Show all 61 references

Research works printed but not linked: 2

  1. Shannon, C. Presentation of a maze solving machine. Trans. 8th Conf. Cybernetics: Circular, Causal and Feedback Mechanisms in Biological and Social Systems 169–181New York, USA, (1951).
  2. Shimon, E. Graph Algorithms 2nd ed Cambridge Univ. (2011).

Cited by

See all 105

Metadata

Authors
Caruso, F., Crespi, A., Ciriolo, A. G., Sciarrino, F., & Osellame, R.
Affiliations
  • Quantum Science and Technology in Arcetri (IT)
  • Istituto di Fotonica e Nanotecnologie (IT)
  • Politecnico di Milano (IT)
  • Sapienza University of Rome (IT)
Corresponding author
  • Caruso, Filippo
  • Osellame, Roberto
Venue
Nature Communications
Published
2016-06-01
Volume
7
Issue
1
Pages
11682-11682
ISSN
2041-1723 (ISSN-L)
OpenAlex
W1567283994
Cited by
105
References
61
Citations by year
2026: 7 · 2025: 6 · 2024: 12 · 2023: 9 · 2022: 9 · 2021: 17 · 2020: 8 · 2019: 11 · 2018: 7 · 2017: 10 · 2016: 7
FWCI
4.17
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • Florence Research (University of Florence) (submittedVersion, cc-by)
  • IRIS Research product catalog (Sapienza University of Rome) (submittedVersion, cc-by-nc)
  • Virtual Community of Pathological Anatomy (University of Castilla La Mancha) (submittedVersion, cc-by)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • arXiv (Cornell University)
Funders
  • Nvidia
  • European Commission (307783, FP7 ICT, 600838, EU-FP7, ICT-2011, FP7-ICT-2011-9, 293449, FP7‐ICT‐2011)
  • Ministero dell’Istruzione, dell’Università e della Ricerca (RBFR10M3SB)
  • FP7 Information and Communication Technologies
Topics
  • Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Random lasers and scattering media
  • Physical Sciences › Computer Science › Artificial Intelligence › Quantum Computing Algorithms and Architecture
  • Physical Sciences › Physics and Astronomy › Statistical and Nonlinear Physics › Quantum chaos and dynamical systems
Keywords
quantum walks, transport efficiency, interference suppression

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-05 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.