Quantum annealing brings modularity maximization and community detection closer to real-world practice

A new paper titled “Modularity Maximization and Community Detection in Complex Networks Through Recursive and Hierarchical Annealing in the DWAVE Advantage Quantum Processing Units” has just been published.

Its authors are Joan Falcó-Roget (Sano), Kacper Jurek (Cyfronet, AGH), Barbara Wojtarowicz(Sano, AGH), Karol Capała (Sano, AGH), and Katarzyna Rycerz (Sano, Cyfronet, AGH).

The article introduces a new approach to detecting community structure in complex networks using pure quantum annealers, specifically D-Wave Advantage systems, without relying on opaque hybrid quantum–classical systems. The method, designed from the ground up to leverage the strengths of quantum annealing hardware, maximizes the modularity function, a widely used measure of how well a network is partitioned into communities.

Many optimization problems in complex systems are inherently non-binary. This represents a major drawback for the applicability of quantum annealing in real scenarios, which naturally rely on binary encodings. Instead of using the standard one-hot encoding, which requires many additional constraints and careful tuning of penalty terms, the authors propose a hierarchical and recursive decomposition of the problem into binary instances that map naturally onto the quantum processing units. This design avoids the need to guess exact Lagrange multipliers and reduces the risk of suboptimal configurations introduced purely by encoding choices.

In their work, the researchers analyze the variability and robustness of the annealing process as a function of network size, directed versus undirected connections, topology, and the desired resolution of communities. They show that the proposed procedure yields meaningful partitions that are at least on par with state-of-the-art classical community detection algorithms, while retaining computation times that remain tractable and promising from a scalability perspective.

An important advantage of the recursive strategy is that the algorithm naturally produces intermediate structures in the form of dendrograms, representing successive splits of the network. Rather than a single opaque final partition, users obtain a transparent hierarchy of communities, which can be inspected and interpreted at different levels of granularity.

This interpretability is especially relevant for the analysis of brain networks. The dendrograms can be used to uncover normal and pathological hierarchical patterns of connectivity, offering a potential route to clinical applications. In this sense, the work is a promising first step toward practice-oriented use of pure quantum annealing in domains where understanding the structure of complex networks is crucial, such as neuroscience and medicine.

By connecting ideas and tools from network science and quantum computing, the publication contributes to closing the gap between two communities that have until now largely developed in parallel. It also illustrates how emerging quantum technologies can start to address problems that matter outside purely academic benchmarks, moving closer to real-world data and workflows.