Research programme
Automated characterization of quantum devices with interpretable and scalable neural networks
The goal of this research is to automatically analyze and improve quantum devices using smart neural networks, helping to minimize experimental errors and better understand quantum measurements.
- Duration
- 2026 - 2029
- Contact
- Anna Dawid-Lekowska
- Funding
- Quantum Delta Top Talent Initiative
The project proposes automating the characterization of quantum devices by employing interpretable and scalable neural networks, using Rydberg quantum simulators as an example. The automatic characterization here is twofold. On the one hand, we will develop a special, interpretable neural network to detect phases of matter directly from experimental measurements. Thanks to interpretability, the network will also reveal the order parameters driving phase transitions, thereby expanding our understanding of quantum many-body systems. On the other hand, we will develop a network that can learn from well-controlled numerical simulations of smaller quantum systems and extrapolate the learned knowledge to larger quantum systems. We will apply this scalable network to detect the Hamiltonian governing the Rydberg quantum simulator and help decrease errors, e.g., in positioning atoms. With the two-fold automatic characterization of quantum simulators, the results of this project will enhance quantum devices and enable an unbiased interpretation of measurements.