The full magnetic characterization of novel materials, be it for fundamental research or technological innovation, calls for a deep understanding of their unique underlying magnetic structures – where the relevant physics happens. Measuring the corresponding generated stray magnetic fields can be done with high accuracy using techniques such as Scanning NV Magnetometry. However, reconstructing the magnetization from the data remains an ill-posed problem that involves complex mathematical modeling and a large number of variables, and requires extensive knowledge of the specific material.
In their recently published paper, A. Dubois (Basel University & Qnami) et al demonstrate how Artificial Intelligence can solve such an ill-posed inverse problem. By…
