Topological network analysis using a programmable photonic quantum processor

Shang Yu, Jinzhao Sun ∗, Zhenghao Li ∗, Ewan Mer, Yazeed K Alwehaibi, Oscar Scholin, Gerard J. Machado, Kuan-Cheng Chen, Aonan Zhang, Raj B Patel, Ying Dong, Ian A. Walmsley, Vlatko Vedral, Ginestra Bianconi · 2025

Read Topological network analysis using a programmable photonic quantum processor by Shang Yu, Jinzhao Sun ∗, Zhenghao Li ∗, Ewan Mer, Yazeed K Alwehaibi, Oscar Scholin, Gerard J. Machado, Kuan-Cheng Chen, Aonan Zhang, Raj B Patel, Ying Dong, Ian A. Walmsley, Vlatko Vedral, Ginestra Bianconi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-08157.txt.

Introduction

Shang Yu and co-authors ran a quantum processor made of light to analyze the hidden topology of networks, in a 2025 paper. The shape of a network, its cliques and holes, matters in fields from brain science to materials. Finding dense fully connected groups, called k-cliques, is hard for classical machines. The team's programmable photonic chip encodes networks with complex weights directly into quantum states of light.

Their sampling method, Gaussian boson sampling, naturally favors heavy, dense subgraphs. From its output they could pick out weighted cliques and estimate Betti numbers, which count holes at each dimension. The entropy of the samples even revealed phase transitions in the network's clique structure. The authors claim the work opens quantum-enhanced analysis of real-world networks, turning an abstract machine into a practical lens on complex data.

Catalog record

CREATORShang Yu, Jinzhao Sun ∗, Zhenghao Li ∗, Ewan Mer, Yazeed K Alwehaibi, Oscar Scholin, Gerard J. Machado, Kuan-Cheng Chen, Aonan Zhang, Raj B Patel, Ying Dong, Ian A. Walmsley, Vlatko Vedral, Ginestra Bianconi
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-08157

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Topological network analysis using a programmable photonic quantum processor Shang Yu, Jinzhao Sun ∗, Zhenghao Li ∗, Ewan Mer, Yazeed K Alwehaibi, Oscar Scholin, Gerard J. Machado, Kuan-Cheng Chen, Aonan Zhang, Raj B Patel, Ying Dong, Ian A. Walmsley, Vlatko Vedral, Ginestra Bianconi Abstract Understanding topological features in networks is crucial for unravelling complex phenomena across fields such as neuroscience, condensed matter, and high-energy physics. However, identifying higher-order topological structures—such as k k -cliques, fundamental building blocks of complex networks—remains a significant challenge. Here we develop a universal programmable photonic quantum processor that enables the encoding of arbitrary complex-weight networks, providing a direct pathway to uncovering their topological structures. We demonstrate how this quantum approach can identify weighted k k -cliques and estimate Betti numbers by leveraging the Gaussian boson sampling algorithm’s ability to preferentially select high-weight, dense subgraphs. The unique capabilities of our programmable quantum processor allow us to observe topological phase transitions and identify clique percolation phenomena directly from the entropy of the sampling results. These findings showcase how photonic quantum computing can be applied to analyse the topological characteristics of real-world complex networks, opening new possibilities for quantum-enhanced data analysis. August 8, 2026 Main A broad spectrum of scientific research employs complex networks to represent intricate interaction mechanisms Vespignani2018 ; Fortunato2022 ; Posfai2024 ; Bianconi2021 , helping illuminate diverse phenomena ranging from brain function Bullmore2009 ; Bassett2017 , protein interactions Wang2020 ; Meng2021 ; Qiu2023 ; Chen2024 to social dynamics Grewal2008 ; Lynn2020 ; Alvarez2021 . Extracting relevant information from the sheer complexity of networked data is a fundamental challenge of the field that requires the integration of approaches coming from different fields of science, including mathematics, physics and computer science. In this context, topological data analysis (TDA) Baas2018 ; Sizemore2019 ; Carlsson2022 ; Harrington , has emerged as a powerful framework, leveraging algebraic topology to uncover the underlying “shape” of data. Through a specific filtering process Baas2018 ; Carlsson2022 that systematically uncovers multi-scale topological structures, this approach identifies global characteristics in complex networks—such as Betti numbers and Euler characteristics. In this way, topological network analysis highlights stable, robust features, that other conventional methods overlook Carlsson2020 . However, a major limitation of this methodology is its inherently high computational complexity (like computing Betti numbers is \#P -hard), leading to an exponential blow-up as the dataset size increases and rendering the analysis of massive networks computationally intractable in practice Basu1999 ; Basu2008 ; Lloyd2016 ; Schmidhuber2023 . Quantum computing presents a novel platform to address this computational challenge with potential advantages over classical methods Harrigan2021 ; Berry2024 . Lloyd et al. initially proposed a quantum machine learning algorithm for estimating Betti numbers Lloyd2016 , which was subsequently demonstrated on a photonic platform Huang2018 . Ongoing research continues to examine the advantages and limitations of quantum algorithms for computing different topological invariants in networks Akhalwaya2024 ; Crichigno2024 ; Gyurik2024 ; Gyurik2022Towards , with recent results suggesting that while its quantum complexity may remain asymptotically hard, quantum algorithms still can provide a polynomial or even superpolynomial speedup over the best-known classical algorithms for certain types of input networks Berry2024 ; Schmidhuber2023 . In particular, the main computational bottleneck of TDA progress is the construction of the simplicial complex from the input network, which requires identifying all k -cliques on the graph Schmidhuber2023 . Gaussian boson sampling (GBS) Lund2014 ; Hamilton2017 ; Zhong2020 ; Madsen2022 ; Yu2023 is a photonic quantum computing scheme that directly encodes an arbitrary complex-weighted graph into photon interference on a linear optical network. Photon-number measurements at the output effectively sample substructures of the graph with a probability distribution that favours higher subgraph-connectivity with heavier edge weights. GBS-based sampling algorithms have shown enhanced speed and performance over classical methods in dense-subgraph searching applications Arrazola2022 ; Santi2022 ; Deng2023 ; Oh2024prxq , precisely the structural characteristics one needs to capture for constructing the clique-complex. GBS therefore provides a useful tool for analysing datasets encoded as a complex network, that is also well suited for near-term noisy intermediate-scale quantum (NISQ) implementations. In this work, we leverage the statistical properties of GBS to develop a quantum approach for analysing topological features in complex networks. Implemented on a universal programmable photonic quantum computing system—featuring a tunable high-gain squeezed-light source and a fully reconfigurable interferometer—this platform is ideally suited for handling weighted network data. By encoding weighted networks into GBS, we identify k -cliques for computing Betti numbers and extract crucial weight information to guide a two-dimensional filtering process. This approach also enables us to compute the Euler characteristic from those cliques and track its evolution, revealing topological phase transitions Dorogovtsev2008 ; Santos2019 . Leveraging the sensitivity of GBS to dense subgraphs, we further investigate clique percolation Derenyi2005 by analysing the entropy of the sampling outcomes, allowing us to probe critical phenomena characterized by abrupt changes in network connectivity. These results demonstrate how quantum techniques can uncover intricate topological properties that are difficult to capture classically. Given that many real-world problems can be represented as graph models, our photonic quantum approach provides a valuable framework for studying large-scale network structures, such as the structural complexity of brain networks Santos2019 . By effectively extracting higher-order topological structures from complex weighted networks, our work highlights the potential of quantum-enhanced methodologies in network science and beyond. Figure 1: Photonic quantum processor for analysing networks with complex-weighted edges. a, The network data is fed into Babbage , a universally programmable GBS machine employing temporal-polarization- hybrid (TPH) encoding. The experimental output aids in identifying all k -cliques within the network. Babbage , featuring expanded mode capacity, comprises three modules: (1) a squeezing source, (2) a quantum processing unit, and (3) a detection system. Details of the network encoding method and the experimental setup are provided in the Methods and Supplementary Information Section II. Abbreviations: periodically poled KTP (ppKTP), electro-optic modulator (EOM), acousto-optic modulator (AOM), half-wave plate (HWP), quarter-wave plate (QWP), (non-polarizing) beam splitter (BS), polarization beam splitter (PBS), right-angle prism mirrors (RAP), piezoelectric mirror (PZT), quantum sequential access memory (QuSAM) and superconducting nanowire single-photon detector (SNSPD). b, The adjacency matrix of a dual-layer random network with complex weights. Dots in the yellow-shaded upper triangle represent the real part of the weighted edges, while those in the blue-shaded lower triangle represent the imaginary part. Lighter and deeper shades of yellow and blue correspond to positive and negative values, respectively. c, Improvement in the success rate of finding k -cliques in the dual-layer network (Fig. 1b) using GBS, compared to uniform sampling and squashed-state sampling as benchmarks, for 5 -, 6 -, and 7 -cliques. Details can be found in the Methods. d, By tracking the birth correlation intensity ( BCI , \omega_{t}^{b} ) and death correlation intensity ( DCI , \omega_{t}^{d} ) of k -cliques at different connectivity thresholds, this diagram illustrates the evolution of the network’s topological structure. Sizes of circles highlight the density of k -cliques, while the distance away from the diagonal signifies their persistent features. Analysing complex-number weighted networks with a photonic quantum computer. We developed a large-scale photonic quantum processor for analysing the topological features of networks. Using a resource-efficient, scalable temporal-polarization-hybrid (TPH) encoding scheme, this hardware enables us to embed arbitrary complex-weighted

Read the complete text