arXiv subjects

Michael Hochberg

Publications and source records attributed to Michael Hochberg.

At least 19 recordsLinked to original sources

arXiv2024

Roadmapping the Next Generation of Silicon Photonics

Silicon photonics has developed into a mainstream technology driven by advances in optical communications. The current generation has led to a proliferation of integrated photonic devices from thousands to millions - mainly in the form of communication transceivers for data centers. Products in many exciting applications, such as sensing and computing, are around the corner. What will it take to increase the proliferation of silicon photonics from millions to billions of units shipped? What will the next generation of silicon photonics look like? What are the common threads in the integration and fabrication bottlenecks that silicon photonic applications face, and which emerging technologies can solve them? This perspective article is an attempt to answer such questions. We chart the generational trends in silicon photonics technology, drawing parallels from the generational definitions of CMOS technology. We identify the crucial challenges that must be solved to make giant strides in CMOS-foundry-compatible devices, circuits, integration, and packaging. We identify challenges critical to the next generation of systems and applications - in communication, signal processing, and sensing. By identifying and summarizing such challenges and opportunities, we aim to stimulate further research on devices, circuits, and systems for the silicon photonics ecosystem.

physics.optics↗

arXiv2022

Single chip photonic deep neural network with accelerated training

As deep neural networks (DNNs) revolutionize machine learning, energy consumption and throughput are emerging as fundamental limitations of CMOS electronics. This has motivated a search for new hardware architectures optimized for artificial intelligence, such as electronic systolic arrays, memristor crossbar arrays, and optical accelerators. Optical systems can perform linear matrix operations at exceptionally high rate and efficiency, motivating recent demonstrations of low latency linear algebra and optical energy consumption below a photon per multiply-accumulate operation. However, demonstrating systems that co-integrate both linear and nonlinear processing units in a single chip remains a central challenge. Here we introduce such a system in a scalable photonic integrated circuit (PIC), enabled by several key advances: (i) high-bandwidth and low-power programmable nonlinear optical function units (NOFUs); (ii) coherent matrix multiplication units (CMXUs); and (iii) in situ training with optical acceleration. We experimentally demonstrate this fully-integrated coherent optical neural network (FICONN) architecture for a 3-layer DNN comprising 12 NOFUs and three CMXUs operating in the telecom C-band. Using in situ training on a vowel classification task, the FICONN achieves 92.7% accuracy on a test set, which is identical to the accuracy obtained on a digital computer with the same number of weights. This work lends experimental evidence to theoretical proposals for in situ training, unlocking orders of magnitude improvements in the throughput of training data. Moreover, the FICONN opens the path to inference at nanosecond latency and femtojoule per operation energy efficiency.

cs.ET↗

arXiv2022

Delocalized Photonic Deep Learning on the Internet's Edge

Advances in deep neural networks (DNNs) are transforming science and technology. However, the increasing computational demands of the most powerful DNNs limit deployment on low-power devices, such as smartphones and sensors -- and this trend is accelerated by the simultaneous move towards Internet-of-Things (IoT) devices. Numerous efforts are underway to lower power consumption, but a fundamental bottleneck remains due to energy consumption in matrix algebra, even for analog approaches including neuromorphic, analog memory and photonic meshes. Here we introduce and demonstrate a new approach that sharply reduces energy required for matrix algebra by doing away with weight memory access on edge devices, enabling orders of magnitude energy and latency reduction. At the core of our approach is a new concept that decentralizes the DNN for delocalized, optically accelerated matrix algebra on edge devices. Using a silicon photonic smart transceiver, we demonstrate experimentally that this scheme, termed Netcast, dramatically reduces energy consumption. We demonstrate operation in a photon-starved environment with 40 aJ/multiply of optical energy for 98.8% accurate image recognition and <1 photon/multiply using single photon detectors. Furthermore, we show realistic deployment of our system, classifying images with 3 THz of bandwidth over 86 km of deployed optical fiber in a Boston-area fiber network. Our approach enables computing on a new generation of edge devices with speeds comparable to modern digital electronics and power consumption that is orders of magnitude lower.

cs.ET↗

arXiv2021

Experimental quantum speed-up in reinforcement learning agents

Increasing demand for algorithms that can learn quickly and efficiently has led to a surge of development within the field of artificial intelligence (AI). An important paradigm within AI is reinforcement learning (RL), where agents interact with environments by exchanging signals via a communication channel. Agents can learn by updating their behaviour based on obtained feedback. The crucial question for practical applications is how fast agents can learn to respond correctly. An essential figure of merit is therefore the learning time. While various works have made use of quantum mechanics to speed up the agent's decision-making process, a reduction in learning time has not been demonstrated yet. Here we present a RL experiment where the learning of an agent is boosted by utilizing a quantum communication channel with the environment. We further show that the combination with classical communication enables the evaluation of such an improvement, and additionally allows for optimal control of the learning progress. This novel scenario is therefore demonstrated by considering hybrid agents, that alternate between rounds of quantum and classical communication. We implement this learning protocol on a compact and fully tunable integrated nanophotonic processor. The device interfaces with telecom-wavelength photons and features a fast active feedback mechanism, allowing us to demonstrate the agent's systematic quantum advantage in a setup that could be readily integrated within future large-scale quantum communication networks.

quant-ph↗

arXiv2020

Quantum transport simulations in a programmable nanophotonic processor

Environmental noise and disorder play critical roles in quantum particle and wave transport in complex media, including solid-state and biological systems. Recent work has predicted that coupling between noisy environments and disordered systems, in which coherent transport has been arrested due to localization effects, could actually enhance transport. Photonic integrated circuits are promising platforms for studying such effects, with a central goal being the development of large systems providing low-loss, high-fidelity control over all parameters of the transport problem. Here, we fully map the role of disorder in quantum transport using a nanophotonic processor consisting of a mesh of 88 generalized beamsplitters programmable on microsecond timescales. Over 64,400 transport experiments, we observe several distinct transport regimes, including environment-assisted quantum transport and the ''quantum Goldilocks'' regime in strong, statically disordered discrete-time systems. Low loss and high-fidelity programmable transformations make this nanophotonic processor a promising platform for many-boson quantum simulation experiments.

quant-ph↗

arXiv2019

A Recurrent Ising Machine in a Photonic Integrated Circuit

Conventional computing architectures have no known efficient algorithms for combinatorial optimization tasks, which are encountered in fundamental areas and real-world practical problems including logistics, social networks, and cryptography. Physical machines have recently been proposed and implemented as an alternative to conventional exact and heuristic solvers for the Ising problem, one such optimization task that requires finding the ground state spin configuration of an arbitrary Ising graph. However, these physical approaches usually suffer from decreased ground state convergence probability or universality for high edge-density graphs or arbitrary graph weights, respectively. We experimentally demonstrate a proof-of-principle integrated nanophotonic recurrent Ising sampler (INPRIS) capable of converging to the ground state of various 4-spin graphs with high probability. The INPRIS exploits experimental physical noise as a resource to speed up the ground state search. By injecting additional extrinsic noise during the algorithm iterations, the INPRIS explores larger regions of the phase space, thus allowing one to probe noise-dependent physical observables. Since the recurrent photonic transformation that our machine imparts is a fixed function of the graph problem, and could thus be implemented with optoelectronic architectures that enable GHz clock rates (such as passive or non-volatile photonic circuits that do not require reprogramming at each iteration), our work paves a way for orders-of-magnitude speedups in exploring the solution space of combinatorially hard problems.

physics.optics↗

arXiv2019

Variational Quantum Unsampling on a Quantum Photonic Processor

Quantum algorithms for Noisy Intermediate-Scale Quantum (NISQ) machines have recently emerged as new promising routes towards demonstrating near-term quantum advantage (or supremacy) over classical systems. In these systems samples are typically drawn from probability distributions which --- under plausible complexity-theoretic conjectures --- cannot be efficiently generated classically. Rather than first define a physical system and then determine computational features of the output state, we ask the converse question: given direct access to the quantum state, what features of the generating system can we efficiently learn? In this work we introduce the Variational Quantum Unsampling (VQU) protocol, a nonlinear quantum neural network approach for verification and inference of near-term quantum circuits outputs. In our approach one can variationally train a quantum operation to unravel the action of an unknown unitary on a known input state; essentially learning the inverse of the black-box quantum dynamics. While the principle of our approach is platform independent, its implementation will depend on the unique architecture of a specific quantum processor. Here, we experimentally demonstrate the VQU protocol on a quantum photonic processor. Alongside quantum verification, our protocol has broad applications; including optimal quantum measurement and tomography, quantum sensing and imaging, and ansatz validation.

quant-ph↗

arXiv2018

Scalable feedback control of single photon sources for photonic quantum technologies

Large-scale quantum technologies require exquisite control over many individual quantum systems. Typically, such systems are very sensitive to environmental fluctuations, and diagnosing errors via measurements causes unavoidable perturbations. In this work we present an in situ frequency locking technique that monitors and corrects frequency variations in single photon sources based on microring resonators. By using the same classical laser fields required for photon generation as a probe to diagnose variations in the resonator frequency, our protocol applies feedback control to correct photon frequency errors in parallel to the optical quantum computation without disturbing the physical qubit. We implement our technique on a silicon photonic device and demonstrate sub 1 pm frequency stabilization in the presence of applied environmental noise, corresponding to a fractional frequency drift of <1 % of a photon linewidth. Using these methods we demonstrate feedback controlled quantum state engineering. By distributing a single local oscillator across a single chip or network of chips, our approach enables frequency locking of many single photon sources for large-scale photonic quantum technologies.

quant-ph↗

arXiv2018

Nonlinear characterisation of a silicon integrated Bragg waveguide filter

Bragg waveguides are promising optical filters for pump suppression in spontaneous Four-Wave Mixing (FWM) photon sources. In this work, we investigate the generation of unwanted photon pairs in the filter itself. We do this by taking advantage of the relation between spontaneous and classical FWM, which allows for the precise characterisation of the nonlinear response of the device. The pair generation rate estimated from the classical measurement is compared with the theoretical value calculated by means of a full quantum model of the filter, which also allows to investigate the spectral properties of the generated pairs. We find a good agreement between theory and experiment, confirming that stimulated FWM is a valuable approach to characterise the nonlinear response of an integrated filter, and that the pairs generated in a Bragg waveguide are not a serious issue for the operation of a fully integrated nonclassical source.

quant-ph↗

arXiv2016

Deep Learning with Coherent Nanophotonic Circuits

Artificial Neural Networks are computational network models inspired by signal processing in the brain. These models have dramatically improved the performance of many learning tasks, including speech and object recognition. However, today's computing hardware is inefficient at implementing neural networks, in large part because much of it was designed for von Neumann computing schemes. Significant effort has been made to develop electronic architectures tuned to implement artificial neural networks that improve upon both computational speed and energy efficiency. Here, we propose a new architecture for a fully-optical neural network that, using unique advantages of optics, promises a computational speed enhancement of at least two orders of magnitude over the state-of-the-art and three orders of magnitude in power efficiency for conventional learning tasks. We experimentally demonstrate essential parts of our architecture using a programmable nanophotonic processor.

physics.optics↗

arXiv2016

Energy correlations of photon pairs generated by a silicon microring resonator probed by Stimulated Four Wave Mixing

Compact silicon integrated devices, such as micro-ring resonators, have recently been demonstrated as efficient sources of quantum correlated photon pairs. The mass production of integrated devices demands the implementation of fast and reliable techniques to monitor the device performances. In the case of time-energy correlations, this is particularly challenging, as it requires high spectral resolution that is not currently achievable in coincidence measurements. Here we reconstruct the joint spectral density of photons pairs generated by spontaneous four-wave mixing in a silicon ring resonator by studying the corresponding stimulated process, namely stimulated four wave mixing. We show that this approach, featuring high spectral resolution and short measurement times, allows one to discriminate between nearly-uncorrelated and highly-correlated photon pairs.

quant-ph↗

arXiv2015

A highly scalable fully non-blocking silicon photonic switch fabric

Large port count spatial optical switches will facilitate flexible and energy efficient data movement in future data communications systems, especially if they are capable of nanosecond-order reconfiguration times. In this work, we demonstrate an 8x8 microring-based silicon photonic switch with software controlled switching. The proposed switch architecture is modular as it assembles multiple identical components with multiplexing/demultiplexing functionalities. The switch is fully non-blocking, has path independent insertion loss, low crosstalk and is straightforward to control. A scalability analysis shows that this architecture can scale to very large port counts. This work represents the first demonstration of real-time firmware controlled switching with silicon photonics devices integrated at the chip scale.

physics.optics↗

arXiv2014

Efficient, Compact and Low Loss Thermo-Optic Phase Shifter in Silicon

We design a resistive heater optimized for efficient and low-loss optical phase modulation in a silicon-on-insulator (SOI) waveguide and characterize the fabricated devices. Modulation is achieved by flowing current perpendicular to a new ridge waveguide geometry. The resistance profile is engineered using different dopant concentrations to obtain localized heat generation and maximize the overlap between the optical mode and the high temperature regions, while simultaneously minimizing optical loss due to free-carrier absorption. A 61.6 micrometer-long phase shifter was fabricated in a CMOS process with oxide cladding and two metal layers. The device features a phase-shifting efficiency of 24.77 +/- 0.43 mW/pi and a -3 dB modulation bandwidth of 130.0 +/- 5.59 kHz. The insertion loss measured for 21 devices across an 8-inch wafer was only 0.23 +/- 0.13 dB. Considering the prospect of densely integrated photonic circuits, we also quantify the separation necessary to isolate thermo-optic devices in the standard 220 nm SOI platform.

physics.optics↗

arXiv2014

An integrated source of spectrally filtered correlated photons for large scale quantum photonic systems

We demonstrate the generation of quantum-correlated photon-pairs combined with the spectral filtering of the pump field by more than 95dB using Bragg reflectors and electrically tunable ring resonators. Moreover, we perform demultiplexing and routing of signal and idler photons after transferring them via a fiber to a second identical chip. Non-classical two-photon temporal correlations with a coincidence-to-accidental ratio of 50 are measured without further off-chip filtering. Our system, fabricated with high yield and reproducibility in a CMOS process, paves the way toward truly large-scale quantum photonic circuits by allowing sources and detectors of single photons to be integrated on the same chip.

quant-ph↗

arXiv2014

Experimental demonstration of broadband Lorentz non-reciprocity in an integrable photonic architecture based on Mach-Zehnder modulators

We demonstrate the first active optical isolator and circulator implemented in a linear and reciprocal material platform using commercial Mach-Zehnder modulators. In a proof-of-principle experiment based on single-mode polarization-maintaining fibers, we achieve more than 12.5 dB isolation over an unprecedented 8.7 THz bandwidth at telecommunication wavelengths, with only 9.1 dB total insertion loss. Our architecture provides a practical answer to the challenge of non-reciprocal light routing in photonic integrated circuits.

physics.optics↗

arXiv2014

Broadband on-chip optical non-reciprocity using phase modulators

Breaking the reciprocity of light propagation in photonic integrated circuits (PIC) - especially in the CMOS-compatible silicon-on-insulator platform - is a topic of intense research. However, a practical solution for monolithic integrating of optical isolators and circulators remains elusive. Here, we propose and analyze a new non-reciprocal photonic architecture operating with standard single-mode waveguides (or optical fibers). Our design exploits cascaded phase modulators separated by optical delay lines and suitably driven by time shifted waveforms. Because it is based on fully balanced interferometers and does not involve resonant structures, our scheme is also intrinsically broadband. Using realistic parameters we calculate an extinction ratio superior to 20 dB and insertion loss below -3 dB.

physics.optics↗

arXiv2012

A 25 Gb/s Silicon Photonics Platform

Silicon has attracted attention as an inexpensive and scalable material system for photonic-electronic, system-on-chip development. For this, a platform with both photodetectors and modulators working at high speeds, with excellent cross-wafer uniformity, is needed. We demonstrate an optical-lithography, wafer-scale photonics platform with 25 Gb/s operation. We also demonstrate modulation with an ultra-low drive voltage of 1 Vpp at 25 Gb/s. We demonstrate attractive cross-wafer uniformity, and provide detailed information about the device geometry. Our platform is available to the community as part of a photonics shuttle service.

physics.optics↗

arXiv2010

Sub-Volt Silicon-Organic Electrooptic Modulator

Lowering the operating voltage of electrooptic modulators is desirable for a variety of applications, most notably in analog photonics , and digital data communications . In particular for digital systems such as CPUs, it is desirable to develop modulators that are both temperature-insensitive and compatible with typically sub-2V CMOS electronics ; however, drive voltages in silicon-based MZIs currently exceed 6.5V . Here we show an MZI modulator based on an electrooptic polymer-clad silicon slot waveguide, with a halfwave voltage of only 0.69V, and a bandwidth of 500 MHz. We also show that there are also paths to significantly improve both the bandwidth and drive voltage . Our silicon-organic modulator has an intrinsic power consumption less than 0.66 pJ/bit, nearly an order of magnitude improvement over the previous lowest energy silicon MZI .

physics.optics↗