Bei Postal Teachers posted on "Nature Photonics" to overcome the control problem of photon chip control

On July 7th, Xu Xingyuan, a teacher of the National Key Laboratory of the National Key Laboratory of the School of Electronic Engineering of our school, and the Australian ARTHUR LOWERY team, published in the top optical journal "Nature Photonics", entitled "Self-Calibration ProgrammMABLE Photonic Integrated CirCuits" " Research Papers. The work proposes and realizes self -learning programmable photon chips. Through the reference path on the chip, it can realize the phase recovery of any double -ended photonic device. Combined with the self -learning algorithm, it is expected to completely solve the stringing and control problems of photon chips.

The photon chip consists of a dense separate integrated optical element, which can provide an ultra -wide simulation bandwidth and Petabits/s order -level information throughput rate. It has broad application prospects in the fields of optical communication, signal processing, quantum computing, and artificial intelligence. Considering that the design, processing, and sealing cycle of photon chips is still long, new photon chips that can be programmed and supports multiple information processing functions have important research and application value. However, the control problem of programmable photon chips has not been fully resolved, which seriously limits its controllable dimensions and practical applications. On the one hand, chip processing errors and chip bruises, especially for large -scale and complicated structures of photon chips, are difficult to accurately compensate. On the other hand, the real -time test of chip parameters is still difficult -mainly limited by the phase jitter or the working bandwidth of the instrument to measure the instrument.

The self-learning photon chip proposed by Mr. Xu Xingyuan in the thesis is connected in parallel with a programmable nucleus and the reference path with the shortest delay to form a minimum phase filter. Essence By using the frequency sweeping laser and optical power measurement of the insertion loss (that is, the square of frequent response), and the Hilbert transformation, the chip can respond frequently. Through the chip's complete trusted interest, you can calculate the various device parameters of the chip, which is convenient for subsequent parameters to directly control the feedback control of each device, which greatly accelerates the convergence process. This scheme can accurately restore the frequency response of the chip, and to achieve the learning of a variety of signal processing functions through the automatic control algorithm. Without any chip's acendous knowledge, it can converge within 25 iteration. This work will greatly improve the controllable dimensions and performance of photon chips. As one of the core technologies of photon chips, it will promote the expansion and application of photonic technology in important areas such as signal processing, photoresm network, quantum computing, and artificial intelligence.

The working principle of self -programming of programming photon chips

- END -