Search

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.

Quantum key distribution (QKD) exploits the quantum nature of light to share provably secure keys, allowing secure communication in the presence of an eavesdropper. The first QKD schemes used photons encoded in two states, such as polarization. Recently, much effort has turned to large-alphabet QKD schemes, which encode photons in high-dimensional basis states. Compared to binary-encoded QKD, large-alphabet schemes can encode more secure information per detected photon, boosting secure communication rates, and also provide increased resilience to noise and loss. High-dimensional encoding may also improve the efficiency of other quantum information processing tasks, such as performing Bell tests and implementing quantum gates. Here, we demonstrate a large-alphabet QKD protocol based on high-dimensional temporal encoding. We achieve record secret-key rates and perform the first field demonstration of large-alphabet QKD. This demonstrates a new, practical way to optimize secret-key rates and marks an important step towards transmission of high-dimensional quantum states in deployed networks., Comment: 6 pages, 4 figures

We report on a systematic experimental study of heralding efficiency and generation rate of telecom-band infrared photon pairs generated by spontaneous parametric down-conversion and coupled to single mode optical fibers. We define the correlated-mode coupling efficiency--an inherent source efficiency--and explain its relation to heralding efficiency. For our experiment, we developed a reconfigurable computer controlled pump-beam and collection-mode optical apparatus which we used to measure the generation rate and correlated-mode coupling efficiency. The use of low-noise, high-efficiency superconducting nanowire single-photon detectors in this setup allowed us to explore focus configurations with low overall photon flux. The measured data agree well with theory and we demonstrated a correlated-mode coupling efficiency of $97 \pm 2\%$, which is the highest efficiency yet achieved for this type of system. These results confirm theoretical treatments and demonstrate that very high overall heralding efficiencies can, in principle, be achieved in quantum optical systems. It is expected that these results and techniques will be widely incorporated into future systems that require, or benefit from, a high heralding efficiency., Comment: 10 pages, 4 figures

In this Letter, we derive an entropic Einstein-Podolsky-Rosen (EPR) steering inequality for continuous variable (CV) systems using only experimentally measured discrete probability distributions and details of the measurement apparatus. We use this inequality to witness EPR steering between the positions and momenta of photon pairs generated in spontaneous parametric downconversion (SPDC). We examine the asymmetry between parties in this inequality, and show that this asymmetry can be used to reduce the technical requirements of experimental setups intended to demonstrate the EPR paradox. Furthermore, we develop a more stringent steering inequality that is symmetric between parties, and use it to show that the downconverted photon pairs also exhibit symmetric EPR steering., Comment: 5 pages incl. references, 2 figures. See arXiv:1312.2604 for explicit derivation

High dimensional Hilbert spaces used for quantum communication channels offer the possibility of large data transmission capabilities. We propose a method of characterizing the channel capacity of an entangled photonic state in high dimensional position and momentum bases. We use this method to measure the channel capacity of a parametric downconversion state, achieving a channel capacity over 7 bits/photon in either the position or momentum basis, by measuring in up to 576 dimensions per detector. The channel violated an entropic separability bound, suggesting the performance cannot be replicated classically., Comment: 5 pages, 2 figures

We investigate the effect of turbulence on quantum ghost imaging. We use entangled photons and demonstrate that for a novel experimental configuration the effect of turbulence can be greatly diminished. By decoupling the entangled photon source from the ghost imaging central image plane, we are able to dramatically increase the ghost image quality. When imaging a test pattern through turbulence, this method increased the imaged pattern visibility from V = 0.14 +/- 0.04 to V = 0.29 +/- 0.04.

We demonstrate an experiment which utilizes a Sagnac interferometer to measure a change in optical frequency of 129 kHz per root Hz with only 2 mW of continuous wave, single mode input power. We describe the measurement of a weak value and show how even higher frequency sensitivities may be obtained over a bandwidth of several nanometers. This technique has many possible applications, such as precision relative frequency measurements and laser locking without the use of atomic lines., Comment: 4 pages, 3 figures, published in PRA

We describe a weak value inspired phase amplification technique in a Sagnac interferometer. We monitor the relative phase between two paths of a slightly misaligned interferometer by measuring the average position of a split-Gaussian mode in the dark port. Although we monitor only the dark port, we show that the signal varies linearly with phase and that we can obtain similar sensitivity to balanced homodyne detection. We derive the source of the amplification both with classical wave optics and as an inverse weak value., Comment: 5 pages, 4 figures, previously submitted for publication

The amplification obtained using weak values is quantified through a detailed investigation of the signal to noise ratio for an optical beam deflection measurement. We show that for a given deflection, input power and beam radius, the use of interferometric weak values allows one to obtain the optimum signal to noise ratio using a coherent beam. This method has the advantage of reduced technical noise and allows for the use of detectors with a low saturation intensity. We report on an experiment which improves the signal to noise ratio for a beam deflection measurement by a factor of 54 when compared to a measurement using the same beam size and a quantum limited detector.

We report on the use of an interferometric weak value technique to amplify very small transverse deflections of an optical beam. By entangling the beam's transverse degrees of freedom with the which-path states of a Sagnac interferometer, it is possible to realize an optical amplifier for polarization independent deflections. The theory for the interferometric weak value amplification method is presented along with the experimental results, which are in good agreement. Of particular interest, we measured the angular deflection of a mirror down to 560 femtoradians and the linear travel of a piezo actuator down to 20 femtometers.

We derive the weak value deflection given in a paper by Dixon et al. (Phys. Rev. Lett. 102, 173601 (2009)) both quantum mechanically and classically. This paper is meant to cover some of the mathematical details omitted in that paper owing to space constraints.