No abstract

Using memristor crossbar arrays to accelerate computations is a promising approach to efficiently implement algorithms in deep neural networks. Early demonstrations, however, are limited to simulations or small-scale problems primarily due to materials and device challenges that limit the size of the memristor crossbar arrays that can be reliably programmed to stable and analog values, which is the focus of the current work. High-precision analog tuning and control of memristor cells across a 128 × 64 array is demonstrated, and the resulting vector matrix multiplication (VMM) computing precision is evaluated. Single-layer neural network inference is performed in these arrays, and the performance compared to a digital approach is assessed. Memristor computing system used here reaches a VMM accuracy equivalent of 6 bits, and an 89.9% recognition accuracy is achieved for the 10k MNIST handwritten digit test set. Forecasts show that with integrated (on chip) and scaled memristors, a computational efficiency greater than 100 trillion operations per second per Watt is possible.

Matter with a high energy density (>10(5) joules per cm(3)) is prevalent throughout the Universe, being present in all types of stars and towards the centre of the giant planets; it is also relevant for inertial confinement fusion. Its thermodynamic and transport properties are challenging to measure, requiring the creation of sufficiently long-lived samples at homogeneous temperatures and densities. With the advent of the Linac Coherent Light Source (LCLS) X-ray laser, high-intensity radiation (>10(17) watts per cm(2), previously the domain of optical lasers) can be produced at X-ray wavelengths. The interaction of single atoms with such intense X-rays has recently been investigated. An understanding of the contrasting case of intense X-ray interaction with dense systems is important from a fundamental viewpoint and for applications. Here we report the experimental creation of a solid-density plasma at temperatures in excess of 10(6) kelvin on inertial-confinement timescales using an X-ray free-electron laser. We discuss the pertinent physics of the intense X-ray-matter interactions, and illustrate the importance of electron-ion collisions. Detailed simulations of the interaction process conducted with a radiative-collisional code show good qualitative agreement with the experimental results. We obtain insights into the evolution of the charge state distribution of the system, the electron density and temperature, and the timescales of collisional processes. Our results should inform future high-intensity X-ray experiments involving dense samples, such as X-ray diffractive imaging of biological systems, material science investigations, and the study of matter in extreme conditions.

We have used the Linac Coherent Light Source to generate solid-density aluminum plasmas at temperatures of up to 180 eV. By varying the photon energy of the x rays that both create and probe the plasma, and observing the K-α fluorescence, we can directly measure the position of the K edge of the highly charged ions within the system. The results are found to disagree with the predictions of the extensively used Stewart-Pyatt model, but are consistent with the earlier model of Ecker and Kröll, which predicts significantly greater depression of the ionization potential.