A Noisy Gates-based learning model to characterize Quantum Devices

Quantum computing is an emerging technology that is progressing rapidly and has the potential to significantly impact computational methods in the coming decades. However, current state-of-the-art devices are significantly affected by noise resulting from interactions with the surrounding environment. Over the years, various solutions have been proposed to address this challenge. In this thesis, we introduce a novel learning model based on the concept of noisy gates to characterize the noise affecting quantum devices. The noisy gates framework integrates noise effects directly into gate operations, providing a more accurate simulation of quantum hardware. Building on this approach, we develop a learnable noise model based on a generic noise formulation, incorporating suitable assumptions to ensure tractability. The model is parameterized by damping rates and jump operators in the Lindblad equation. The model is optimized using supervised learning techniques, with a loss function derived from measured observables in quantum circuits. We adapt the Noisy Gates Python package to build the learning model and run simulations. To validate our approach, we conduct experiments on one- and two-qubit systems, demonstrating the effectiveness of the proposed method as a proof of concept. Our results highlight the potential of this model for device characterization in the Noisy Intermediate-Scale Quantum (NISQ) era. This work lays the foundation for future research on scaling the approach to larger quantum systems and refining the optimization process with more sophisticated machine learning algorithms.