Speckle-learned Complex Coefficient Prediction of superposed Orbital Angular Momentum States of Light

Orbital Angular Momentum (OAM) beams are a special type of light that can carry information in unique ways, making them valuable for things like boosting data capacity in communication systems and for use in quantum technologies. Their capacity to carry more information compared to traditional light beams makes them crucial for future technological advancements.

When studying OAM beams, researcher often face challenges because these beams are very sensitive to how they are aligned, which makes them hard to analyze. Speckle patterns offer a solution. These are random patterns created when light hits rough surfaces or passes through inhomogeneous materials. Even though these patterns look random, they carry important details about the light. Directly analyzing OAM beams requires perfect alignment, which can be challenging. In contrast, speckle patterns can still provide useful information even when alignment is not perfect, making them a more reliable and cost-effective option for studying OAM beams.

So far, most research has focused on using machine learning to classify OAM beams into different categories based on their characteristics. However, classifying OAM beams is not sufficient for more advanced tasks. Determining the exact details of superposed OAM beams, such as their weights and relative phases, is crucial for improving the efficiency of technologies like high-capacity communication systems.

This study takes things further by using machine learning not just to classify but to predict these precise details through regression. By training a neural network on speckle pattern data, the model can predict both the weights and relative phases of superposed OAM beams more accurately. Unlike classification, this regression approach provides continuous, detailed predictions. The results show that this approach significantly improves how well OAM beams can be understood and controlled. This has major implications for technologies that rely on these beams, such as high-speed data transmission and quantum computing, where having precise control over the beams is essential.