web analytics

Model-Free Interferometry Enabled by Machine Learning

Published:
Lead Inventor: Liang Jiang
Print or Save to PDF

SUMMARY

  • Atom interferometry is used in sensors designed to measure gravitational field, acceleration, and angular momentum and is of increasing importance to the emerging field of quantum communications, which shows great promise for fast, secure, and precise relay of information.

  • Interpreting the results of an interferometer requires precise calibration and characterization, typically achieved using a mathematical model of the system under consideration; as such, precise mathematical models are required in order to correctly interpret the output of interferometer. Existing approaches that rely on such models to estimate physical parameters, however, suffer from inherent systematic errors as the underlying mathematical models are only approximations. Further, these models tend to perform poorly when applied to data with low resolution or low signal to noise ratios.

  • The faculty inventor developed a technology using machine learning to infer physical parameters from images generated by an atom interferometer without the need for an exact underlying mathematical model.  Specifically, the technology considers inferring quantities such as acceleration and rotations from images of interference patterns generated by an interferometer. The technology simultaneously estimates quantities of interest such as acceleration and rotation from interference patterns, and the error associated with those estimates sometimes caused by noisy input images. The technology extends the applicability of the interferometer when the resolution is limited, and noise and imperfections are present.

 

FIGURE

Process for using machine learning to build a model-free inference algorithm that requires only patterns with known parameters.

 

ADVANTAGES

ADVANTAGES

  • Improved ability to estimate physical parameters from interferometry 

  • Provides reliable parameter estimates when using poor resolution or low signal-to-noise data

  • Superior to Fourier-based analysis of the interference patterns

  • Superior to the state-of-the-art phase unwrapping algorithm

APPLICATIONS

  • Metrology

  • Quantum sensing

  • Communications / telecommunications

  • Interferometry and metrology