Abstract: A laser diode including a double heterostructure comprising a top layer, a buffer layer formed on a substrate, and an intrinsic active layer formed between the top layer and the buffer layer. The top layer and the buffer layer have opposite types of conductivity. The active layer has a bandgap smaller than that of the buffer layer or the top layer. The double heterostructure includes Ge, SiGe, GeSn, and/or SiGeSn materials.

Abstract: In some implementations, a device may generate a data set including at least modal gain values and modal loss values for a semiconductor optical amplifier (SOA) slice. The device may determine, based on the data set, respective widths for a plurality of slices of an SOA using an autoregressive model. A width, of the respective widths, for a slice, of the plurality of slices, may be associated with a maximum conversion efficiency achievable for the slice at a given current density. The device may generate information indicating a geometry for the SOA based on the respective widths for the plurality of slices.

Abstract: A laser diode including a double heterostructure comprising a top layer, a buffer layer formed on a substrate, and an intrinsic active layer formed between the top layer and the buffer layer. The top layer and the buffer layer have opposite types of conductivity. The active layer has a bandgap smaller than that of the buffer layer or the top layer. The double heterostructure includes Ge, SiGe, GeSn, and/or SiGeSn materials.

Abstract: Systems and methods for the detection of motion in a multi-dimensional signal are provided. According to aspects of the disclosure, data representing a time sequence of frames of a video stream is received. A multi-dimensional Gaussian window in is then constructed. Each frame of the received data is divided into a plurality of blocks and, for each block in the plurality of blocks, each pixel in the block is multiplied with a corresponding pixel of the Gaussian window to obtain a windowed video block. The FFT of the windowed video block is computed to obtain an amplitude and phase thereof. The occurrence of motion in the block is then determined based on the computed phase of the windowed video block.

Abstract: Systems and methods for the detection of motion in a multi-dimensional signal are provided. According to aspects of the disclosure, data representing a time sequence of frames of a video stream is received. A multi-dimensional Gaussian window in is then constructed. Each frame of the received data is divided into a plurality of blocks and, for each block in the plurality of blocks, each pixel in the block is multiplied with a corresponding pixel of the Gaussian window to obtain a windowed video block. The FFT of the windowed video block is computed to obtain an amplitude and phase thereof. The occurrence of motion in the block is then determined based on the computed phase of the windowed video block.

Abstract: Techniques for reconstructing a signal encoded with a time encoding machine (TEM) using a recurrent neural network including receiving a TEM-encoded signal, processing the TEM-encoded signal, and reconstructing the TEM-encoded signal with a recurrent neural network.

Abstract: Systems and methods for system identification, encoding and decoding multi-component signals are disclosed. An exemplary method can include receiving the one or more multi-component signals and separating the multi-component signals into channels. The method can also filter each of the channels using receptive field components. The values for the receptive field components can be aggregated and provided to neurons for encoding.

Abstract: Techniques for reconstructing a signal encoded with a time encoding machine (TEM) using a recurrent neural network including receiving a TEM-encoded signal, processing the TEM-encoded signal, and reconstructing the TEM-encoded signal with a recurrent neural network.