Search

Integrated neuromorphic photonic circuits aim to power complex artificial neural networks (ANNs) in an energy and time efficient way by exploiting the large bandwidth and the low loss of photonic structures. However, scaling photonic circuits to match the requirements of modern ANNs still remains challenging. In this perspective, we give an overview over the usual sizes of matrices processed in ANNs and compare them with the capability of existing photonic matrix processors. To address shortcomings of existing architectures, we propose a time multiplexed matrix processing scheme which virtually increases the size of a physical photonic crossbar array without requiring any additional electrical post-processing. We investigate the underlying process of time multiplexed incoherent optical accumulation and achieve accumulation accuracy of 98.9% with 1 ns pulses. Assuming state of the art active components and a reasonable crossbar array size, this processor architecture would enable matrix vector multiplications with 16,000 × 64 matrices all optically on an estimated area of 51.2 mm2, while performing more than 110 trillion multiply and accumulate operations per second.

The integration of artificial intelligence (AI) systems in the daily life greatly increases the amount of data generated and processed. In addition to the large computational power required, the hardware needs to be compact and energy efficient. One promising approach to fulfill those requirements is phase-change material based photonic neuromorphic computing that enables in-memory computation and a high degree of parallelization. In the following, we present an optimized layout of a photonic tensor core (PTC) which is designed to perform real valued matrix vector multiplications and operates at telecommunication wavelengths. We deploy the well-studied phase-change material Ge2Sb2Te5 (GST) as an optical attenuator to perform single positive valued multiplications. In order to generalize the multiplication to arbitrary real factors, we develop a novel symmetric multiplication unit which directly includes a reference-computation branch. The variable GST attenuator enables a modulation depth of 5 dB over a wavelength range of 100 nm with a wavelength dependency below 0.8 dB. The passive photonic circuit itself ensures equal coupling to the main-computation and reference-computation branch over the complete wavelength range. For the first time, we integrate wavelength multiplexers (MUX) together with a photonic crossbar array on-chip, paving the way towards fully integrated systems. The MUX are crucial for the PTC since they enable multiple computational channels in a single photonic crossbar array. We minimize the crosstalk between the channels by designing Bragg scattering based MUX. By cascading, we achieve an extinction ratio larger than 61 dB while the insertion loss is below 1 dB.

Ice rises or ice rumples act as ice-shelf pinning points that can play an important role in regulating the ice discharge of marine outlet glaciers. As an example, the observed recent gradual ungrounding of the ice shelf of West Antarctica's Thwaites Glacier from its last pinning points likely diminished the buttressing effect of the ice shelf and thus contributed to the destabilization of the outlet. Here we use an idealized experimental setting to simulate the response of a marine outlet glacier resting on a landward down-sloping (retrograde) bed to a stepwise ungrounding of its ice shelf from a topographic high and a subsequent re-grounding. We show that the glacier retreat down the retrograde bed, induced by the loss in pinning-point buttressing, can be unstable and irreversible given a relatively deep subglacial bed depression. In this case, glacier retreat and re-advance show a hysteretic behavior, and if the bed depression is sufficiently deep, the glacier does not recover but remains locked in its retreated state. Conversely, reversibility requires a sufficiently shallow bed depression. Based on a simple flux balance analysis, we argue that the combination of a deep bed depression and limited ice-shelf buttressing hampers grounding-line re-advance due to the dominant and highly non-linear influence of the bed depth on the ice discharge across the grounding line. We conclude that outlets that rest on a deep bed depression and are weakly buttressed, such as Thwaites Glacier, are more susceptible to abrupt and irreversible retreat than more strongly buttressed glaciers on more moderate retrograde slopes, such as Pine Island Glacier. Our findings further suggest that the (ir)reversibility of large-scale grounding-line retreat may be strongly affected by calving-front migration and associated changes in ice-shelf buttressing. [ABSTRACT FROM AUTHOR]

Ice rises or ice rumples act as ice-shelf pinning points that can have an important role in regulating the ice discharge of marine outlet glaciers. As an example, the observed recent gradual ungrounding of the ice shelf of West Antarctica's Thwaites Glacier from its last pinning points likely diminished the buttressing effect of the ice shelf and thus contributed to the destabilization of the outlet. Here we use an idealized experimental setting to simulate the response of an Antarctic-type, instability-prone marine outlet glacier to a successive ungrounding of its ice shelf from a topographic high and a subsequent re-grounding. We show that the glacier retreat down the landward down-sloping (retrograde) bed, induced by the loss in pinning-point buttressing, can be unstable and irreversible given a relatively deep subglacial bed depression. In this case, glacier retreat and re-advance show a hysteretic behavior and if the bed depression is sufficiently deep, the glacier does not recover from but remains locked in its collapsed state. Conversely, reversibility requires a sufficiently shallow bed depression. Based on a simple flux balance analysis, we argue that the combination of a deep bed depression and limited ice-shelf buttressing hampers grounding-line re-advance due to the dominant and highly non-linear influence of the bed depth on the ice discharge across the grounding line. We conclude that outlets that rest on a deep bed depression and are weakly buttressed, such as Thwaites Glacier, are more susceptible to abrupt and irreversible retreat than stronger buttressed glaciers on more moderate retrograde slope, such as Pine Island Glacier. In particular, our results suggest that the wide and deep marine bed depression in the interior of Thwaites Glacier's drainage basin might promote potential future unstable retreat and also represent a strong limitation for a possible re-advance of the glacier in case it would collapse. [ABSTRACT FROM AUTHOR]

The integration of artificial intelligence systems into daily applications like speech recognition and autonomous driving rapidly increases the amount of data generated and processed. However, satisfying the hardware requirements with the conventional von Neumann architecture remains challenging due to the von Neumann bottleneck. Therefore, new architectures inspired by the working principles of the human brain are developed, and they are called neuromorphic computing. The key principles of neuromorphic computing are in-memory computing to reduce data shuffling and parallelization to decrease computation time. One promising framework for neuromorphic computing is phase-change photonics. By switching to the optical domain, parallelization is inherently possible by wavelength division multiplexing, and high modulation speeds can be deployed. Non-volatile phase-change materials are used to perform multiplications and non-linear operations in an energetically efficient manner. Here, we present two prototypes of neuromorphic photonic computation units based on chalcogenide phase-change materials. First is a neuromorphic hardware accelerator designed to carry out matrix vector multiplication in convolutional neural networks. Due to the neuromorphic architecture, this prototype can already operate at tera-multiply-accumulate per second speeds. Second is an all-optical spiking neuron, which can serve as a building block for large-scale artificial neural networks. Here, the whole computation is carried out in the optical domain, and the device only needs an electrical interface for data input and readout. [ABSTRACT FROM AUTHOR]

Integrated neuromorphic photonic circuits aim to power complex artificial neural networks (ANNs) in an energy and time efficient way by exploiting the large bandwidth and the low loss of photonic structures. However, scaling photonic circuits to match the requirements of modern ANNs still remains challenging. In this perspective, we give an overview over the usual sizes of matrices processed in ANNs and compare them with the capability of existing photonic matrix processors. To address shortcomings of existing architectures, we propose a time multiplexed matrix processing scheme which virtually increases the size of a physical photonic crossbar array without requiring any additional electrical post-processing. We investigate the underlying process of time multiplexed incoherent optical accumulation and achieve accumulation accuracy of 98.9% with 1 ns pulses. Assuming state of the art active components and a reasonable crossbar array size, this processor architecture would enable matrix vector multiplications with 16,000 × 64 matrices all optically on an estimated area of 51.2 mm2, while performing more than 110 trillion multiply and accumulate operations per second. [ABSTRACT FROM AUTHOR]