These authors contributed equally to this work.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.

These authors contributed equally to this work.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.Modern computers based on the von Neumann architecture are far more power-hungry and less effective than their biological counterparts -central nervous systemsfor a wide range of tasks including perception, communication, learning, and decision making. With the increasing data volume associated with processing big data, developing computers that learn, combine, and analyze vast amounts of information quickly and efficiently is becoming increasingly important. For example, speech recognition software (e.g., Apple's Siri) is typically executed in the cloud since these computations are too taxing for mobile hardware; realtime image processing is an even more demanding task [1]. To address the shortcomings of von Neumann computing architectures for neural networks, much recent work has focused on increasing artificial neural network computing speed and power efficiency by developing electronic architectures (such as ASIC and FPGA chips) specifically tailored to a task [2][3][4][5]. Recent demonstrations of electronic neuromorphic hardware architectures have reported improved computational performance [6]. Hybrid optical-electronic systems that implement spike processing [7][8][9] and reservoir computing [10,11] have also been investigated recently. However, the computational speed and power efficiency achieved with these hardware architectures are still limited by electronic clock rates and ohmic losses.Fully-optical neural networks offer a promising alternative approach to microelectronic and hybrid optical-electronic implementations. Linear transformations (and certain nonlinear transformations) can be performed at the speed of light and detected at rates exceeding 100 GHz [12] in photonic networks, and in some cases, with minimal power consumption [13]. For example, it is well known that a common lens performs Fourier transform without any power consumption, and that certain matrix operations can also be performed optically without consuming power. However, implementing such transformations with bulk opti...

Environmental noise and disorder play a critical role 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 out the role of static and dynamic disorder in quantum transport using a low-loss, phase-stable, nanophotonic processor consisting of a mesh of 56 generalized beamsplitters programmable on microsecond timescales. Over 85,600 transport experiments, we observe several distinct transport regimes, including environment-enhanced transport in strong, statically disordered systems. Low loss and programmability make this nanophotonic processor a promising platform for many-boson quantum simulation experiments.Quantum walks (QWs), the coherent analogy to classical random walks, have emerged as a useful model for experimental simulations of quantum transport (QT) phenomena in physical systems. QWs have been implemented in platforms including trapped ions 1,2 , ultra-cold atoms 3 , bulk optics 4-8 and integrated photonics 4,9-16 . Integrated photonic implementations are particularly attractive for relatively large coherence lengths, high interferometric visibilities, integration with single-photon sources 17,18 and detectors 19 , and the promise of scaling to many active and reconfigurable components. The role of static and dynamic disorder in the transport of quantum walkers has been of particular interest in the field of quantum simulation 20,21 .Control over static (time-invariant) and dynamic (timevarying) disorder enables studies of fundamentally interesting and potentially useful QT phenomena in discrete-time (DT) QWs. In systems with strong dynamic disorder, illustrated in Fig. 1(a), a quantum walker evolving over T time steps travels a distance proportional to √ T ; the coherent nature of the quantum walker is effectively erased, resulting in classical, diffusive transport characteristics 22,23 . In contrast, a quantum walker (or coherent wave) traversing an ordered system travels a distance proportional to T as a result of coherent interference between superposition amplitudes -a regime known as ballistic transport (see Fig. 1(b)). Perhaps most notably, a quantum walker propagating through a system with strong, static disorder becomes exponentially localized in space and time, inhibiting transport, as illustrated in Fig. 1(c). This QT phenomena is known as Anderson localization 24 and it has been observed in several systems, including optical media [9][10][11]25,26 . For systems in which transport has been arrested due to Anderson localization, it has recently been predicted that adding environmental noise (dynamic disord...

Photonic-integrated circuits have emerged as a scalable platform for complex quantum systems. A central goal is to integrate single-photon detectors to reduce optical losses, latency and wiring complexity associated with off-chip detectors. Superconducting nanowire single-photon detectors (SNSPDs) are particularly attractive because of high detection efficiency, sub-50-ps jitter and nanosecond-scale reset time. However, while single detectors have been incorporated into individual waveguides, the system detection efficiency of multiple SNSPDs in one photonic circuit—required for scalable quantum photonic circuits—has been limited to <0.2%. Here we introduce a micrometer-scale flip-chip process that enables scalable integration of SNSPDs on a range of photonic circuits. Ten low-jitter detectors are integrated on one circuit with 100% device yield. With an average system detection efficiency beyond 10%, and estimated on-chip detection efficiency of 14–52% for four detectors operated simultaneously, we demonstrate, to the best of our knowledge, the first on-chip photon correlation measurements of non-classical light.

No abstract

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 of the structure, while simultaneously minimizing optical loss due to free-carrier absorption. A 61.6 µm 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/π 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.

We demonstrate the generation of quantum-correlated photon pairs combined with the spectral filtering of the pump field by more than 95 dB on a single silicon chip using electrically tunable ring resonators and passive Bragg reflectors. Moreover, we perform the demultiplexing and routing of signal and idler photons after transferring them via an optical fiber to a second identical chip. Nonclassical 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-compatible process, paves the way toward large-scale quantum photonic circuits by allowing sources and detectors of single photons to be integrated on the same chip.

Quantum information science offers inherently more powerful methods for communication, computation, and precision measurement that take advantage of quantum superposition and entanglement. In recent years, theoretical and experimental advances in quantum computing and simulation with photons have spurred great interest in developing large photonic entangled states that challenge today's classical computers. As experiments have increased in complexity, there has been an increasing need to transition bulk optics experiments to integrated photonics platforms to control more spatial modes with higher fidelity and phase stability. The silicon-on-insulator (SOI) nanophotonics platform offers new possibilities for quantum optics, including the integration of bright, nonclassical light sources, based on the large third-order nonlinearity (χ (3) ) of silicon, alongside quantum state manipulation circuits with thousands of optical elements, all on a single phase-stable chip. How large do these photonic systems need to be? Recent theoretical work on Boson Sampling suggests that even the problem of sampling from ~30 identical photons, having passed through an interferometer of hundreds of modes, becomes challenging for classical computers. While experiments of this size are still challenging, the SOI platform has the required component density to enable low-loss and programmable interferometers for manipulating hundreds of spatial modes.Here, we discuss the SOI nanophotonics platform for quantum photonic circuits with hundreds-to-thousands of optical elements and the associated challenges. We compare SOI to competing technologies in terms of requirements for quantum optical systems. We review recent results on large-scale quantum state evolution circuits and strategies for realizing high-fidelity heralded gates with imperfect, practical systems. Next, we review recent results on silicon photonics-based photonpair sources and device architectures, and we discuss a path towards large-scale source integration. Finally, we review monolithic integration strategies for single-photon detectors and their essential role in on-chip feed forward operations.