Optimized qudit-based computation schemes using optical neural networks
Since the onset of the digital age, the need to process tremendous amounts of data has become of critical importance for healthcare, financial, and government sectors. This has made information processing tasks increasingly complex, and thus far, there exists no realistic platform that can meet the stringent requirements of the computation industry, e.g., in terms of power and speed. A potential solution are systems that can enable parallel processing of data. This can be realized by using photonic bits that exist in multiple states simultaneously, thus demanding the use of only a few bits due to the increased information density stored within them. Qudits are excellent resources in this regard, since they can store and process large volumes of data at high speeds, and with high noise tolerances, while keeping the circuit complexity low. The challenge here, is the efficient generation and characterization of qudit states, and the implementation of universal computation gates..