157. Qommunity – on the use of quantum technologies in network neuroscience

Abstract:

The community detection problem aims to find groups of nodes that are more connected to each other with respect to the rest of the network. However, it is an NP hard problem and requires heuristic rather than exact solutions. Quantum computing and in particular quantum annealing (QA) can theoretically locate the optimal community structure but evidence for this has been limited to purely academic scenarios. In this talk we will discuss how the problem can be formulated within a combinatorial optimization QUBO framework and addressed with D-Wave [1] solvers. We will discuss how to encode both binary and discrete variables in QA without relying on one-hot constraints (i.e., Lagrange multipliers). Then, we will present Qommunity [2], built on top of QHyper [3], a library for solving optimization problems using quantum methods. Qommunity contains classical, quantum, and hybrid community detection methods. We will also show a newly developed hierarchical partitioning algorithm enabling recursive binary network partitioning and that extends the capabilities of D-Wave’s Advantage quantum solver for real use cases [4]. In summary, we provide a unified description and interface for the implementation of QA methods and the execution of experiments in a hands-on fashion. 

References

[1] https://docs.dwavesys.com/docs/latest/

[2] https://github.com/kacper3615/Qommunity

[3] https://qhyper.readthedocs.io/en/latest/

[4] Joan Falcó-Roget, Kacper Jurek, Barbara Wojtarowicz, Karol Capała, and Katarzyna Rycerz. Modularity maximization and community detection in complex networks through recursive and hierarchical annealing in the D-Wave Advantage quantum processing units.https://arxiv.org/abs/2410.07744

About the authors:

Barbara Wojtarowicz is currently pursuing an MSc in Computer Science at AGH University of Kraków. She’s interested in the application of quantum technologies to nature-based problems and in computational sciences.

Kacper Jurek is a master’s student in Computer Science at AGH University of Krakow. His academic and research interests span theoretical computer science, quantum computing, and modern machine learning methods.

Joan Falcó-Roget holds a Bachelor of Science in fundamental physics from the University of Barcelona and a Master of Science in physics of condensed matter and biological systems from the Autonomous University of Madrid. He researched computational theories of dopamine and animal behavior in Madrid before joining Sano as a PhD student. His current research is centered around neuroimaging studies and methods to study brain lesions and degenerative diseases.