Machine learning techniques for frequency comb optimization

We use supervised and unsupervised machine learning techniques to investigate how optical frequency combs may be used to identify gas molecules in the atmosphere. Dual comb heterodyne detection with GHz and THz repetition rates are often used as spectroscopic methods for probing ro-vibrational IR spectra of small molecules, proving to be promising in dispersive atmospheric sensing regimes. We simulate and recover an intensity spectrum from the molecule-specific IR absorption in the region spanned by the comb bandwidth. In practice, conventional frequency combs contain a narrow range of frequencies that may not span the spectral range associated with typical atmospheric target signals, prohibiting convenient molecule identification. We present several inexpensive, efficient machine learning methods to analyze optimal comb frequencies to maximize the ability to determine the molecular composition of a sample, based only on the interactions at those frequencies with respect to various channels of noise. Using a synthetic dataset, these methods are able to accurately identify molecules using simulated frequency combs, comparable to the ones used in practice. Furthermore, these methods withstand added noise (based on the imperfections of the experimental equipment) beyond what is expected in real applications. We also investigate the optimal comb generation parameters that provide the most informational value, and provide an analysis of the change in accuracy based on the number and size of allocated combs. The robustness of these methods suggests the techniques presented in this paper may be integrated into a laboratory setup and even in on-chip comb systems deployed in an atmospheric setting. This approach may be used as a detection framework for future quantum-enhanced systems via squeezed bi-photon comb inputs.