Parts | Days | Selection | Search | Updates | Downloads | Help

MON: Monday Contributed Sessions

MON 23: Poster Session: Fundamental Aspects and Model Systems

MON 23.15: Poster

Monday, September 8, 2025, 18:30–20:30, ZHG Foyer 1. OG

Scalable Entanglement Quantification in Quantum Many-Body Systems with a Graph Neural Network — •Susanna Bräu, Martina Jung, and Martin Gärttner — Institut für Festkörpertheorie und Optik, Friedrich-Schiller-Universität Jena, Max-Wien-Platz 1, 07743 Jena

Entanglement is a fundamental feature of quantum mechanics and plays a central role in quantum computing, quantum communication, and quantum information theory. Quantifying entanglement between different parts of a system - via measures such as entanglement entropy or quantum mutual information - usually requires full knowledge of the quantum state. However, due to the curse of dimensionality, quantum state tomography is infeasible for larger systems, limiting the accessible system sizes. Therefore, we propose a supervised machine learning approach to estimate entanglement features based on a set of measurement snapshots of the system. For that, we develop a permutation invariant graph neural network (GNN) that is parameter-efficient, being linear in the system size. Our scalable GNN incorporates the mini-set architecture, developed by Kim et al. [1], who divided the input into smaller sets which the model processes in parallel. By attending the output of each mini-set in a permutation invariant manner, high order correlations can be extracted. In this way, we aim to improve the scaling such that the GNN can be applied to larger data sets or be used to increase the time over which the model can accurately predict entanglement features in the future.

[1] Kim, H. et al. arXiv:2405.11632 [quant-ph] (Nov. 2024).

Keywords: Machine learning; Entanglement quantification; Graph neural networks; Quantum many-body systems