资源环境科技发展态势分析平台

Conventional information processors convert information between different physical carriers for processing, storage and transmission. It seems plausible that quantum information will also be held by different physical carriers in applications such as tests of fundamental physics, quantum enhanced sensors and quantum information processing. Quantum controlled molecules, in particular, could transduce quantum information across a wide range of quantum bit (qubit) frequencies-from a few kilohertz for transitions within the same rotational manifold(1), a few gigahertz for hyperfine transitions, a few terahertz for rotational transitions, to hundreds of terahertz for fundamental and overtone vibrational and electronic transitions-possibly all within the same molecule. Here we demonstrate entanglement between the rotational states of a (CaH+)-Ca-40 molecular ion and the internal states of a Ca-40(+) atomic ion(2). We extend methods used in quantum logic spectroscopy(1,3) for pure-state initialization, laser manipulation and state readout of the molecular ion. The quantum coherence of the Coulomb coupled motion between the atomic and molecular ions enables subsequent entangling manipulations. The qubit addressed in the molecule has a frequency of either 13.4 kilohertz(1) or 855 gigahertz(3), highlighting the versatility of molecular qubits. Our work demonstrates how molecules can transduce quantum information between qubits with different frequencies to enable hybrid quantum systems. We anticipate that our method of quantum control and measurement of molecules will find applications in quantum information science, quantum sensors, fundamental and applied physics, and controlled quantum chemistry.

Quantum entanglement is realized between rotational levels of a molecular ion with energy differences spanning several orders of magnitude and long-lived internal states of a single atomic ion.

Memristor-enabled neuromorphic computing systems provide a fast and energy-efficient approach to training neural networks(1-4). However, convolutional neural networks (CNNs)-one of the most important models for image recognition(5)-have not yet been fully hardware-implemented using memristor crossbars, which are cross-point arrays with a memristor device at each intersection. Moreover, achieving software-comparable results is highly challenging owing to the poor yield, large variation and other non-ideal characteristics of devices(6-9). Here we report the fabrication of high-yield, high-performance and uniform memristor crossbar arrays for the implementation of CNNs, which integrate eight 2,048-cell memristor arrays to improve parallel-computing efficiency. In addition, we propose an effective hybrid-training method to adapt to device imperfections and improve the overall system performance. We built a five-layer memristor-based CNN to perform MNIST10 image recognition, and achieved a high accuracy of more than 96 per cent. In addition to parallel convolutions using different kernels with shared inputs, replication of multiple identical kernels in memristor arrays was demonstrated for processing different inputs in parallel. The memristor-based CNN neuromorphic system has an energy efficiency more than two orders of magnitude greater than that of state-of-the-art graphics-processing units, and is shown to be scalable to larger networks, such as residual neural networks. Our results are expected to enable a viable memristor-based non-von Neumann hardware solution for deep neural networks and edge computing.