Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

A Quantum Walk Across Biological Networks: A Demonstration on a 40-Qubit Superconducting Processor

inquantio

Published Sep 8, 2026• 0 citations
Publisher

Abstract

A joint team from Algorithmiq and IBM Quantum has run a discrete-time quantum walk (DTQW) on complex networks drawn from protein-protein interaction (PPI) data, using 40 qubits of real quantum hardware. Rather than accept the extreme circuit depth demanded by conventional dense encodings, the researchers traded depth for qubit count, pairing a symmetry-sector encoding with postselection. On a benchmark graph this cut the depth of a single walk step by roughly two orders of magnitude. Applied to an asthma-associated gene network, the walk lifted FLVCR1, a gene placed well down the list by classical benchmark methods, to fourth position. The authors attribute the shift to quantum interference and present the case as a proof of principle for network medicine. [Quantum Biology Society] In network medicine, where disease mechanisms are read off the interaction networks that link proteins and genes, the classical random walk has long been a workhorse. A walker hops between nodes at random, and the resulting visit frequencies serve as a measure of importance. Adding the quantum principles of superposition and interference produces a quantum walk, whose probability spreads through a graph in a way that has no classical counterpart, and researchers have looked to it as a route to a different kind of network analysis. Computing such a walk on a real biological network has been another matter. Biological graphs are heterogeneous, with node degrees that vary from one vertex to the next, and running a quantum walk over that kind of irregular structure on today's noisy, pre-fault-tolerant hardware has been close to out of reach. A paper published online in npj Quantum Information on 24 July 2026, currently available as an accepted but not yet copyedited Article in Press, reports an experiment that pushes past this barrier. Working on IBM Heron superconducting processors with 156 qubits, researchers at Algorithmiq and IBM Quantum used up to 40 qubits to drive a discrete-time quantum walk through seven steps on a graph of 17 nodes and 20 edges. The authors describe it as the largest DTQW implementation on complex graphs on superconducting hardware to date. ■ Circuit Depth as the Bottleneck, and a Symmetry-Sector Way Around It When qubits were scarce, the natural choice was dense encoding, which compresses N walker positions into log₂N qubits plus a coin qubit. The saving comes at a price. Dense encoding leans on multi-controlled gates whose cost climbs steeply, and circuit depth climbs with it. The paper puts a number on this: with the state-of-the-art design of Sato and Saito, a single walk step on an 8-node, 8-edge graph decomposes into 1,218 entangling layers under idealized all-to-all connectivity, and 2,882 once the heavy-hex connectivity of IBM's processors is imposed. At that depth, decoherence dismantles the computation before it can finish. The researchers took the opposite approach. Symmetry-sector encoding: they allocated twice as many qubits as the graph has edges, 2|E| in all, and confined the state to the single-excitation subspace, in which exactly one qubit sits in the excited state |1>. Among 40 qubits, the single qubit that is switched on encodes both where the walker is and which way it is about to hop. Since qubits tend to decay from |1> toward |0>, states of higher Hamming weight are less stable over time, which is part of the authors' reasoning for choosing weight 1. Postselection as noise mitigation: any measurement that returns two or more excitations, or none at all, has violated the single-excitation symmetry and is discarded as a detected error. A further optimization rotates the computational basis into domain-wall states before the coin operator is applied and rotates it back afterwards, which approximately halves both the number of two-qubit native entangling gates and the transpiled depth of that block. The larger share of the depth reduction, however, comes from the fact that the coin and shift operators act in parallel across the qubit register, so their depth does not grow with the number of nodes or edges but with the maximum node degree. On the same 8-node benchmark, one step of the new framework needs 11 entangling layers under all-to-all connectivity and 28 under heavy-hex, roughly a hundredfold reduction against the reference design. ■ Seven Steps on 40 Qubits, and the Hellinger Fidelity That Held The experiments used three subgraphs extracted from BioPlex 3.0, a proteome-scale map of the human interactome: 11 nodes and 12 edges on 24 qubits, 15 nodes and 18 edges on 36 qubits, and 17 nodes and 20 edges on 40 qubits. The 24- and 40-qubit circuits ran on ibm_kingston, a Heron R2 device, and the 36-qubit circuit on ibm_pittsburgh, a Heron R3 device. Without postselection, the measured distributions matched theory poorly from the outset. At the first step, Hellinger fidelity stood at 0.79, 0.61 and 0.57 for the three graphs, which the authors read as evidence that the encoding alone cannot preserve coherence. With postselection applied, the picture changed. Fidelity stayed above 0.95, 0.90 and 0.87 respectively across all seven steps, including for the largest circuit, which reached 287 entangling layers on 40 qubits by step 7. Because a noisy circuit gradually drifts toward a featureless baseline distribution, a high Hellinger fidelity can flatter a result. The authors therefore also report a baseline-corrected fidelity, measured against the stationary distribution in which probability is spread evenly over directed-edge states and so falls on each node in proportion to its degree. That corrected figure remained at or above 0.71, 0.46 and 0.54 for the three graphs, which the authors take as an indication that the postselected results carry genuine signal rather than noise-driven flattening. ■ A Proof of Principle in Asthma Gene Prioritization The team then applied the measured walk dynamics to a disease-gene discovery task. HLA-DQA1, node 7 in the 11-node graph and one of the most strongly asthma-associated genes by genome-wide association study p-value, was chosen as the starting point for the walker, and the 24-qubit experiment was extended to nine steps. At each step the quantum probability at each node was compared with the corresponding classical random-walk probability, and the difference, normalized by the collision probability of the quantum distribution, gave a quantum interference index. Each node's score was its maximum index over time. HLA-C, PON2 and HLA-G took the top three places, all of them close to the seed and visited early in the walk. The result that drew attention was FLVCR1, described in the paper as the choline and heme transporter 1, which came fourth despite having no direct edge to the seed node or its neighbours and being ranked seventh by PageRank and ninth by DIAMOnD. The authors attribute its prioritization to amplitude interference rather than simple topological closeness, and note that FLVCR1 meets the GWAS significance threshold and has been associated with asthma across age groups, having been reported as a risk factor for childhood-onset disease. The partial overlap with the classical rankings, they argue, shows consistency with established methods, while the differences point to information a quantum walk might supply that classical approaches miss. ■ Significance and Limits Running a large quantum walk on the irregular, heterogeneous graphs characteristic of biological data, rather than on the regular lattices and cycle graphs used in earlier demonstrations, is the central achievement here, and it was done on real superconducting hardware rather than in simulation. The authors are explicit about the boundaries of the claim. The asthma case study runs on an 11-node subgraph chosen to fit current hardware limits, which restricts direct real-world applicability and leaves the work at the proof-of-concept stage. All three test graphs share a maximum node degree of 3, and it is that maximum degree, rather than graph size, that governs how the circuit depth scales. The cost of postselection also grows sharply: the fraction of retained bitstrings decays exponentially with depth, and in the 40-qubit run only about 0.2 per cent survived at step 7, roughly 17,000 out of nearly nine million shots. Some errors escape the symmetry check altogether, including phase flips and paired bit flips that leave the excitation count unchanged; the authors estimate the rate of such undetectable gate errors at around 3 × 10⁻⁴. They add that a walk run without postselection would need gate errors of order 10⁻⁴ to reach the same success probability they obtained with it, against the roughly 10⁻³ two-qubit gate errors available today, and suggest that deeper circuits will call for more advanced mitigation, such as tensor-network error mitigation adapted to work alongside postselection. Set against hardware roadmaps that point toward physical qubit counts in the millions, the authors make the case that trading circuit depth for qubits is a more tractable path forward than pushing gate fidelities alone, and one that should let the same framework reach larger and denser biological networks as the hardware improves. #QuantumWalk #DiscreteTimeQuantumWalk #QuantumComputing #ProteinInteractionNetwork #PPINetwork #DiseaseGenePrioritization #NetworkMedicine #IBMQuantum #ErrorMitigation #Postselection #Asthma #npjQuantumInformation #OpenAccess #QuantumBiologySociety Source: https://www.nature.com/articles/s41534-026-01332-w Circuit design code: https://doi.org/10.6084/m9.figshare.32218227

DOI
10.5281/zenodo.22654431

Related papers

Experimental implementation of a discrete-time quantum walk on biological networks

Viacheslav Dubovitskii, Filippo Utro, Aritra Bose +3 more

Quantum walks provide a versatile framework for probing the structural and dynamical properties of complex systems ranging from biological networks to synthetic materials. However, their realization on current noisy pre-fault-tolerant quantum computers is fundamentally limited by decoherence. Conventional dense encodings of graph structures require prohibitively deep circuits, making them incompatible with existing hardware. Here we introduce an algorithm that leverages symmetry-sector encoding and trades circuit depth for the number of qubits, while integrating symmetry-respecting postselection as an effective noise-mitigation strategy. This combination enables us to execute practical quantum-walk circuits for biological network motifs on actual quantum hardware. We benchmark the proposed methodology against known state-of-the-art circuit architectures, highlighting significant reduction of circuit depth in our approach at the cost of moderate qubit overhead. Utilizing 40 qubits, we implement quantum walks on complex graphs containing up to 17 nodes and 20 edges -- the largest experiment on superconducting hardware to date, with the Hellinger fidelity exceeding 87% throughout 7 steps. We present a proof-of-principle case study that illustrates how experimentally obtained quantum-walk dynamics on a protein-protein-interaction network can be applied to prioritizing disease-associated genes. We discuss the framework scalability in the pre-fault-tolerant era and its potential for studying larger biological networks.

quant-pharXiv:2602.24053•npj Quantum Information•Mar 2, 2026•2 citations

Demonstration of Discrete-Time Quantum Walks and Observation of Topological Edge States in a Superconducting Qutrit Chain

Kun Zhou, Jian-Wen Xu, Qi-Ping Su +12 more

Quantum walk serves as a versatile tool for universal quantum computing and algorithmic research. However, the implementation of discrete-time quantum walks (DTQWs) with superconducting circuits is still constrained by some limitations such as operation precision, circuit depth and connectivity. With improved hardware efficiency by using superconducting qutrits (three-level systems), we experimentally demonstrate a scalable DTQW in a superconducting circuit, observing the ballistic spreading of quantum walk in a qutrit chain. The usage of qutrits in our implementation allows hardware efficiently encoding of the walker position and the coin degree of freedom. By exploiting the flexibility and intrinsic symmetries of qutrit-based DTQWs, we successfully prepare two topological phases in the chain. For the first time, particle-hole-symmetry-protected edge states, bounded at the interface between these two topological phases, are observed in the superconducting platform. Measured parameter dependencies further validate the properties of edge states. The scalability and gate-control compatibility of the demonstrated DTQWs enable a versatile tool for superconducting quantum computing and quantum simulation.

quant-pharXiv:2601.01759•Physical Review Letters•Jan 6, 2026•5 citations

Coined Quantum Walks on Complex Networks for Quantum Computers

Rei Sato

We propose a quantum circuit design for implementing coined quantum walks on complex networks. In complex networks, the coin and shift operators depend on the varying degrees of the nodes, which makes circuit construction more challenging than for regular networks. To address this issue, we use a dual-register encoding to enable a simplified shift operator and reduces the resource overhead. We implement the circuit using Qmod, a high-level quantum programming language, and evaluated the performance through numerical simulations on Erdős-Rényi, Watts-Strogatz, and Barabási-Albert models. The results show that the circuit depth scales as approximately $N^{1.9}$ regardless of the network topology. Furthermore, we execute the proposed circuits on the ibm\_torino superconducting quantum processor for Watts-Strogatz models with $N=4$ and $N=8$. The experiments show that hardware-aware optimization slightly improved the variation distance and Hellinger fidelity for the larger network, whereas connectivity constraints imposed overhead for the smaller one. These results indicate that while current NISQ devices are limited to small-scale validations, the polynomial scaling of our framework makes it suitable for larger-scale implementations in the fault-tolerant quantum computing era.

quant-pharXiv:2512.16400•Quantum Information Processing•Dec 19, 2025•0 citations

Gate-level Implementation and Resource Analysis of Lackadaisical Quantum Walk Search

Amit Saha, Debanjan Kola, Nishanka Das +1 more

Lackadaisical quantum walks (LQW) extend discrete-time quantum walks (DTQW) by introducing weighted self-loops, enabling improved spatial-search performance through controlled localization of the walker. Despite substantial theoretical progress, practical gate-level implementations suitable for quantum hardware remain largely unexplored, limiting evaluation under realistic architectural constraints, noise, and resource requirements. In this work, we present a gate-level implementation framework for lackadaisical quantum walk search. The proposed construction encodes the position and coin spaces into qubit registers, and realizes the walk dynamics through oracle, coin, and flip-flop shift operations. We validate the circuit by reproducing the expected search behavior for single and multiple marked vertices and by analyzing the effect of the self-loop weight on the success probability. We further evaluate the implementation under realistic noisy settings using superconducting hardware's noise models. Logical resource analysis shows that, for grids ranging from $8\times8$ to $64\times64$, the transpiled gate count increases from $3.63\times10^{5}$ to $4.38\times10^{6}$ and the circuit depth from $2.13\times10^{5}$ to $2.56\times10^{6}$. Finally, fault-tolerant resource estimates based on a surface-code model using the Microsoft Quantum Resource Estimator demonstrate the substantial space-time trade-off associated with magic-state production. The results establish a practical circuit-level pathway for implementing the LQW search and provide a basis for evaluating its performance.

quant-pharXiv:2608.17136•Aug 19, 2026•0 citations

Efficient implementation of quantum walk-based algorithms into the quantum gate model of computation

Allan Wing Bocanegra

Discrete-time quantum walks provide a universal model of computation, meaning that any computation done by a general purpose quantum computer can also be done by discrete-time quantum walks. In the last decades great progress has been done in this field, developing quantum walk-based algorithms that can outperform classical ones. However, current quantum computers work based on the quantum circuit model of computation, and the mapping from one model to the other is an open problem, thus as an attempt to address this problem, we in this work develop basic theory, experiments and simulations on quantum walks presented in the following way: In the first part of this work we provide a matrix analysis of the evolution operator of Unitary Coined Discrete-Time Quantum Walks (UCDTQW), which leads to the fact that any quantum circuit applied on a bipartite system can be considered as the shift operator of a UCDTQW, and vice versa. Furthermore, we show that the fact that the statevector associated to a quantum walker in a UCDTQW is a bipartite system, inherits the evolution operator the property to be a block matrix, where each block is a Kraus operator. We show that associated to the shift operator of any UCDTQW, there exists a directed multigraph on which the walker moves, thus mimicking the functionality of the transition matrix in a random walk. We also extend the definition of the coin operator in a UCDTQW to a superposition of coins that act on different vertices of the graph on which the UCDTQW takes place, and provide a description of how this can be implemented in circuit form. In the second part of this work, we study the mapping of evolution operators to the quantum circuit model, based on a special case when the shift operator has a block diagonal matrix representation. In this case we propose a mapping process that is straightforward, and apply this method to obtain quantum circuits that generate UCDTQW on the most common topologies found in the literature: the straight line, the cyclic graph, the hypercube and the complete graph. In view that the mapping is not unique, we provide different models for the some topologies, along with techniques to synthesize them. The obtained circuits were implemented on IBM’s quantum computers through the Quantum Composer platform, obtaining mixed results for the different models, but being the circuit associated to the complete graph the one with best performance. Finally, in the third part of this work, we use the theoretical framework proposed in the first two parts to analyze and modify a quantum walk-based search algorithm which proposed over 20 years ago, and then execute it for the first time on a general-purpose quantum computer provided by IBM.

•Jan 1, 2023•0 citations

Quantum Network Biology: A Network-First Perspective on Quantum-Enabled Biological Communication Network Analysis

Swarup Roy, Jugal Kumar Kalita

Modern biology is increasingly represented through graphs, including protein–protein interaction graphs, cell–cell communication networks, host–pathogen interactomes, gene regulatory networks, and disease–gene association networks. These graphs help capture how molecular and cellular entities interact, signal, regulate, and perturb one another. Classical graph algorithms and graph neural networks have taken this a long way. However, several tasks remain difficult, including network alignment, conserved subgraph discovery, disease–module identification, and therapeutic target prioritization. These challenges are not merely software-engineering issues; many arise from hard combinatorial problem structures and become even harder in practice because biological networks are sparse, noisy, and incompletely annotated. This letter proposesQuantum Network Biology(QNB) as a vocabulary, a scope definition and, above all, a benchmark agenda for asking whether quantum and hybrid quantum–classical methods have anything to offer here. We view biological networks as communication systems. In this view, propagation asks how information moves through a network, and a quantum walk offers a different model of that movement, not just a faster version of a classical algorithm. We separate QNB from quantum biology, quantum bioinformatics and quantum network medicine, pair four quantum primitives with specific graph problems, and specify six benchmarks with named datasets, fixed classical baselines and reported metrics. We do not claim demonstrated quantum advantage. Instead, we argue that the criteria for testing such a claim should be defined now.

IEEE Transactions on Molecular Biological and Multi-Scale Communications•Jan 1, 2026•0 citations
A joint team from Algorithmiq and IBM Quantum has run a discrete-time quantum walk (DTQW) on complex networks drawn from protein-protein interaction (PPI) data, using 40 qubits of real quantum hardware. Rather than accept the extreme circuit depth demanded by conventional dense encodings, the researchers traded depth for qubit count, pairing a symmetry-sector encoding with postselection. On a benchmark graph this cut the depth of a single walk step by roughly two orders of magnitude. Applied to an asthma-associated gene network, the walk lifted FLVCR1, a gene placed well down the list by classical benchmark methods, to fourth position. The authors attribute the shift to quantum interference and present the case as a proof of principle for network medicine. [Quantum Biology Society] In network medicine, where disease mechanisms are read off the interaction networks that link proteins and genes, the classical random walk has long been a workhorse. A walker hops between nodes at random, and the resulting visit frequencies serve as a measure of importance. Adding the quantum principles of superposition and interference produces a quantum walk, whose probability spreads through a graph in a way that has no classical counterpart, and researchers have looked to it as a route to a different kind of network analysis. Computing such a walk on a real biological network has been another matter. Biological graphs are heterogeneous, with node degrees that vary from one vertex to the next, and running a quantum walk over that kind of irregular structure on today's noisy, pre-fault-tolerant hardware has been close to out of reach. A paper published online in npj Quantum Information on 24 July 2026, currently available as an accepted but not yet copyedited Article in Press, reports an experiment that pushes past this barrier. Working on IBM Heron superconducting processors with 156 qubits, researchers at Algorithmiq and IBM Quantum used up to 40 qubits to drive a discrete-time quantum walk through seven steps on a graph of 17 nodes and 20 edges. The authors describe it as the largest DTQW implementation on complex graphs on superconducting hardware to date. ■ Circuit Depth as the Bottleneck, and a Symmetry-Sector Way Around It When qubits were scarce, the natural choice was dense encoding, which compresses N walker positions into log₂N qubits plus a coin qubit. The saving comes at a price. Dense encoding leans on multi-controlled gates whose cost climbs steeply, and circuit depth climbs with it. The paper puts a number on this: with the state-of-the-art design of Sato and Saito, a single walk step on an 8-node, 8-edge graph decomposes into 1,218 entangling layers under idealized all-to-all connectivity, and 2,882 once the heavy-hex connectivity of IBM's processors is imposed. At that depth, decoherence dismantles the computation before it can finish. The researchers took the opposite approach. Symmetry-sector encoding: they allocated twice as many qubits as the graph has edges, 2|E| in all, and confined the state to the single-excitation subspace, in which exactly one qubit sits in the excited state |1>. Among 40 qubits, the single qubit that is switched on encodes both where the walker is and which way it is about to hop. Since qubits tend to decay from |1> toward |0>, states of higher Hamming weight are less stable over time, which is part of the authors' reasoning for choosing weight 1. Postselection as noise mitigation: any measurement that returns two or more excitations, or none at all, has violated the single-excitation symmetry and is discarded as a detected error. A further optimization rotates the computational basis into domain-wall states before the coin operator is applied and rotates it back afterwards, which approximately halves both the number of two-qubit native entangling gates and the transpiled depth of that block. The larger share of the depth reduction, however, comes from the fact that the coin and shift operators act in parallel across the qubit register, so their depth does not grow with the number of nodes or edges but with the maximum node degree. On the same 8-node benchmark, one step of the new framework needs 11 entangling layers under all-to-all connectivity and 28 under heavy-hex, roughly a hundredfold reduction against the reference design. ■ Seven Steps on 40 Qubits, and the Hellinger Fidelity That Held The experiments used three subgraphs extracted from BioPlex 3.0, a proteome-scale map of the human interactome: 11 nodes and 12 edges on 24 qubits, 15 nodes and 18 edges on 36 qubits, and 17 nodes and 20 edges on 40 qubits. The 24- and 40-qubit circuits ran on ibm_kingston, a Heron R2 device, and the 36-qubit circuit on ibm_pittsburgh, a Heron R3 device. Without postselection, the measured distributions matched theory poorly from the outset. At the first step, Hellinger fidelity stood at 0.79, 0.61 and 0.57 for the three graphs, which the authors read as evidence that the encoding alone cannot preserve coherence. With postselection applied, the picture changed. Fidelity stayed above 0.95, 0.90 and 0.87 respectively across all seven steps, including for the largest circuit, which reached 287 entangling layers on 40 qubits by step 7. Because a noisy circuit gradually drifts toward a featureless baseline distribution, a high Hellinger fidelity can flatter a result. The authors therefore also report a baseline-corrected fidelity, measured against the stationary distribution in which probability is spread evenly over directed-edge states and so falls on each node in proportion to its degree. That corrected figure remained at or above 0.71, 0.46 and 0.54 for the three graphs, which the authors take as an indication that the postselected results carry genuine signal rather than noise-driven flattening. ■ A Proof of Principle in Asthma Gene Prioritization The team then applied the measured walk dynamics to a disease-gene discovery task. HLA-DQA1, node 7 in the 11-node graph and one of the most strongly asthma-associated genes by genome-wide association study p-value, was chosen as the starting point for the walker, and the 24-qubit experiment was extended to nine steps. At each step the quantum probability at each node was compared with the corresponding classical random-walk probability, and the difference, normalized by the collision probability of the quantum distribution, gave a quantum interference index. Each node's score was its maximum index over time. HLA-C, PON2 and HLA-G took the top three places, all of them close to the seed and visited early in the walk. The result that drew attention was FLVCR1, described in the paper as the choline and heme transporter 1, which came fourth despite having no direct edge to the seed node or its neighbours and being ranked seventh by PageRank and ninth by DIAMOnD. The authors attribute its prioritization to amplitude interference rather than simple topological closeness, and note that FLVCR1 meets the GWAS significance threshold and has been associated with asthma across age groups, having been reported as a risk factor for childhood-onset disease. The partial overlap with the classical rankings, they argue, shows consistency with established methods, while the differences point to information a quantum walk might supply that classical approaches miss. ■ Significance and Limits Running a large quantum walk on the irregular, heterogeneous graphs characteristic of biological data, rather than on the regular lattices and cycle graphs used in earlier demonstrations, is the central achievement here, and it was done on real superconducting hardware rather than in simulation. The authors are explicit about the boundaries of the claim. The asthma case study runs on an 11-node subgraph chosen to fit current hardware limits, which restricts direct real-world applicability and leaves the work at the proof-of-concept stage. All three test graphs share a maximum node degree of 3, and it is that maximum degree, rather than graph size, that governs how the circuit depth scales. The cost of postselection also grows sharply: the fraction of retained bitstrings decays exponentially with depth, and in the 40-qubit run only about 0.2 per cent survived at step 7, roughly 17,000 out of nearly nine million shots. Some errors escape the symmetry check altogether, including phase flips and paired bit flips that leave the excitation count unchanged; the authors estimate the rate of such undetectable gate errors at around 3 × 10⁻⁴. They add that a walk run without postselection would need gate errors of order 10⁻⁴ to reach the same success probability they obtained with it, against the roughly 10⁻³ two-qubit gate errors available today, and suggest that deeper circuits will call for more advanced mitigation, such as tensor-network error mitigation adapted to work alongside postselection. Set against hardware roadmaps that point toward physical qubit counts in the millions, the authors make the case that trading circuit depth for qubits is a more tractable path forward than pushing gate fidelities alone, and one that should let the same framework reach larger and denser biological networks as the hardware improves. #QuantumWalk #DiscreteTimeQuantumWalk #QuantumComputing #ProteinInteractionNetwork #PPINetwork #DiseaseGenePrioritization #NetworkMedicine #IBMQuantum #ErrorMitigation #Postselection #Asthma #npjQuantumInformation #OpenAccess #QuantumBiologySociety Source: https://www.nature.com/articles/s41534-026-01332-w Circuit design code: https://doi.org/10.6084/m9.figshare.32218227
Quantum walks provide a versatile framework for probing the structural and dynamical properties of complex systems ranging from biological networks to synthetic materials. However, their realization on current noisy pre-fault-tolerant quantum computers is fundamentally limited by decoherence. Conventional dense encodings of graph structures require prohibitively deep circuits, making them incompatible with existing hardware. Here we introduce an algorithm that leverages symmetry-sector encoding and trades circuit depth for the number of qubits, while integrating symmetry-respecting postselection as an effective noise-mitigation strategy. This combination enables us to execute practical quantum-walk circuits for biological network motifs on actual quantum hardware. We benchmark the proposed methodology against known state-of-the-art circuit architectures, highlighting significant reduction of circuit depth in our approach at the cost of moderate qubit overhead. Utilizing 40 qubits, we implement quantum walks on complex graphs containing up to 17 nodes and 20 edges -- the largest experiment on superconducting hardware to date, with the Hellinger fidelity exceeding 87% throughout 7 steps. We present a proof-of-principle case study that illustrates how experimentally obtained quantum-walk dynamics on a protein-protein-interaction network can be applied to prioritizing disease-associated genes. We discuss the framework scalability in the pre-fault-tolerant era and its potential for studying larger biological networks.
Quantum walk serves as a versatile tool for universal quantum computing and algorithmic research. However, the implementation of discrete-time quantum walks (DTQWs) with superconducting circuits is still constrained by some limitations such as operation precision, circuit depth and connectivity. With improved hardware efficiency by using superconducting qutrits (three-level systems), we experimentally demonstrate a scalable DTQW in a superconducting circuit, observing the ballistic spreading of quantum walk in a qutrit chain. The usage of qutrits in our implementation allows hardware efficiently encoding of the walker position and the coin degree of freedom. By exploiting the flexibility and intrinsic symmetries of qutrit-based DTQWs, we successfully prepare two topological phases in the chain. For the first time, particle-hole-symmetry-protected edge states, bounded at the interface between these two topological phases, are observed in the superconducting platform. Measured parameter dependencies further validate the properties of edge states. The scalability and gate-control compatibility of the demonstrated DTQWs enable a versatile tool for superconducting quantum computing and quantum simulation.
We propose a quantum circuit design for implementing coined quantum walks on complex networks. In complex networks, the coin and shift operators depend on the varying degrees of the nodes, which makes circuit construction more challenging than for regular networks. To address this issue, we use a dual-register encoding to enable a simplified shift operator and reduces the resource overhead. We implement the circuit using Qmod, a high-level quantum programming language, and evaluated the performance through numerical simulations on Erdős-Rényi, Watts-Strogatz, and Barabási-Albert models. The results show that the circuit depth scales as approximately N1.9N^{1.9}N1.9 regardless of the network topology. Furthermore, we execute the proposed circuits on the ibm\_torino superconducting quantum processor for Watts-Strogatz models with N=4N=4N=4 and N=8N=8N=8. The experiments show that hardware-aware optimization slightly improved the variation distance and Hellinger fidelity for the larger network, whereas connectivity constraints imposed overhead for the smaller one. These results indicate that while current NISQ devices are limited to small-scale validations, the polynomial scaling of our framework makes it suitable for larger-scale implementations in the fault-tolerant quantum computing era.
Lackadaisical quantum walks (LQW) extend discrete-time quantum walks (DTQW) by introducing weighted self-loops, enabling improved spatial-search performance through controlled localization of the walker. Despite substantial theoretical progress, practical gate-level implementations suitable for quantum hardware remain largely unexplored, limiting evaluation under realistic architectural constraints, noise, and resource requirements. In this work, we present a gate-level implementation framework for lackadaisical quantum walk search. The proposed construction encodes the position and coin spaces into qubit registers, and realizes the walk dynamics through oracle, coin, and flip-flop shift operations. We validate the circuit by reproducing the expected search behavior for single and multiple marked vertices and by analyzing the effect of the self-loop weight on the success probability. We further evaluate the implementation under realistic noisy settings using superconducting hardware's noise models. Logical resource analysis shows that, for grids ranging from 8×88\times88×8 to 64×6464\times6464×64, the transpiled gate count increases from 3.63×1053.63\times10^{5}3.63×105 to 4.38×1064.38\times10^{6}4.38×106 and the circuit depth from 2.13×1052.13\times10^{5}2.13×105 to 2.56×1062.56\times10^{6}2.56×106. Finally, fault-tolerant resource estimates based on a surface-code model using the Microsoft Quantum Resource Estimator demonstrate the substantial space-time trade-off associated with magic-state production. The results establish a practical circuit-level pathway for implementing the LQW search and provide a basis for evaluating its performance.
Discrete-time quantum walks provide a universal model of computation, meaning that any computation done by a general purpose quantum computer can also be done by discrete-time quantum walks. In the last decades great progress has been done in this field, developing quantum walk-based algorithms that can outperform classical ones. However, current quantum computers work based on the quantum circuit model of computation, and the mapping from one model to the other is an open problem, thus as an attempt to address this problem, we in this work develop basic theory, experiments and simulations on quantum walks presented in the following way: In the first part of this work we provide a matrix analysis of the evolution operator of Unitary Coined Discrete-Time Quantum Walks (UCDTQW), which leads to the fact that any quantum circuit applied on a bipartite system can be considered as the shift operator of a UCDTQW, and vice versa. Furthermore, we show that the fact that the statevector associated to a quantum walker in a UCDTQW is a bipartite system, inherits the evolution operator the property to be a block matrix, where each block is a Kraus operator. We show that associated to the shift operator of any UCDTQW, there exists a directed multigraph on which the walker moves, thus mimicking the functionality of the transition matrix in a random walk. We also extend the definition of the coin operator in a UCDTQW to a superposition of coins that act on different vertices of the graph on which the UCDTQW takes place, and provide a description of how this can be implemented in circuit form. In the second part of this work, we study the mapping of evolution operators to the quantum circuit model, based on a special case when the shift operator has a block diagonal matrix representation. In this case we propose a mapping process that is straightforward, and apply this method to obtain quantum circuits that generate UCDTQW on the most common topologies found in the literature: the straight line, the cyclic graph, the hypercube and the complete graph. In view that the mapping is not unique, we provide different models for the some topologies, along with techniques to synthesize them. The obtained circuits were implemented on IBM’s quantum computers through the Quantum Composer platform, obtaining mixed results for the different models, but being the circuit associated to the complete graph the one with best performance. Finally, in the third part of this work, we use the theoretical framework proposed in the first two parts to analyze and modify a quantum walk-based search algorithm which proposed over 20 years ago, and then execute it for the first time on a general-purpose quantum computer provided by IBM.
Modern biology is increasingly represented through graphs, including protein–protein interaction graphs, cell–cell communication networks, host–pathogen interactomes, gene regulatory networks, and disease–gene association networks. These graphs help capture how molecular and cellular entities interact, signal, regulate, and perturb one another. Classical graph algorithms and graph neural networks have taken this a long way. However, several tasks remain difficult, including network alignment, conserved subgraph discovery, disease–module identification, and therapeutic target prioritization. These challenges are not merely software-engineering issues; many arise from hard combinatorial problem structures and become even harder in practice because biological networks are sparse, noisy, and incompletely annotated. This letter proposesQuantum Network Biology(QNB) as a vocabulary, a scope definition and, above all, a benchmark agenda for asking whether quantum and hybrid quantum–classical methods have anything to offer here. We view biological networks as communication systems. In this view, propagation asks how information moves through a network, and a quantum walk offers a different model of that movement, not just a faster version of a classical algorithm. We separate QNB from quantum biology, quantum bioinformatics and quantum network medicine, pair four quantum primitives with specific graph problems, and specify six benchmarks with named datasets, fixed classical baselines and reported metrics. We do not claim demonstrated quantum advantage. Instead, we argue that the criteria for testing such a claim should be defined now.