skip to main content


Title: High-Speed Multidimensional Optical Computing

A coherent multi-dimensional photonic tensor accelerator performing high-speed matrix-matrix multiplication is proposed and demonstrated. A pattern recognition experiment is demonstrated at a 25Gbps modulation speed exploiting orthogonal dimensions of light including time, wavelength, and spatial mode.

 
more » « less
Award ID(s):
1932858
PAR ID:
10562765
Author(s) / Creator(s):
; ; ; ; ;
Publisher / Repository:
Optica Publishing Group
Date Published:
ISBN:
978-1-957171-29-6
Page Range / eLocation ID:
JW4A.74
Format(s):
Medium: X
Location:
Tacoma, Washington
Sponsoring Org:
National Science Foundation
More Like this
  1. We propose a coherent multi-dimensional (wavelength, spatial mode, polarization, etc.) photonic tensor accelerator capable of performing high-speed artificial neural network computation. High-speed matrix-vector and matrix-matrix multiplication were experimentally demonstrated. 
    more » « less
  2. Abstract

    Sum‐accelerated pseudospectral methods herein are used to describe the kinetic energy matrix within density functional theory descriptions of few‐atom systems. The resulting energies are compared to the conventional finite difference results. It is demonstrated that the sum‐acceleration allows for significantly improved scaling with respect to the grid step size and matrix bandwidth while maintaining a computational speed equivalent with that of the finite difference description. In addition, analytical expressions for the pseudopotential matrix elements are introduced to alleviate the eggbox effect—the dependence of the energy on the relative position of the atoms and grid points.

     
    more » « less
  3. Matrix completion problems arise in many applications including recommendation systems, computer vision, and genomics. Increasingly larger neural networks have been successful in many of these applications but at considerable computational costs. Remarkably, taking the width of a neural network to infinity allows for improved computational performance. In this work, we develop an infinite width neural network framework for matrix completion that is simple, fast, and flexible. Simplicity and speed come from the connection between the infinite width limit of neural networks and kernels known as neural tangent kernels (NTK). In particular, we derive the NTK for fully connected and convolutional neural networks for matrix completion. The flexibility stems from a feature prior, which allows encoding relationships between coordinates of the target matrix, akin to semisupervised learning. The effectiveness of our framework is demonstrated through competitive results for virtual drug screening and image inpainting/reconstruction. We also provide an implementation in Python to make our framework accessible on standard hardware to a broad audience. 
    more » « less
  4. Matrix multiplication is a fundamental building block for large scale computations arising in various applications, including machine learning. There has been significant recent interest in using coding to speed up distributed matrix multiplication, that are robust to stragglers (i.e., machines that may perform slower computations). In many scenarios, instead of exact computation, approximate matrix multiplication, i.e., allowing for a tolerable error is also sufficient. Such approximate schemes make use of randomization techniques to speed up the computation process. In this paper, we initiate the study of approximate coded matrix multiplication, and investigate the joint synergies offered by randomization and coding. Specifically, we propose two coded randomized sampling schemes that use (a) codes to achieve a desired recovery threshold and (b) random sampling to obtain approximation of the matrix multiplication. Tradeoffs between the recovery threshold and approximation error obtained through random sampling are investigated for a class of coded matrix multiplication schemes. 
    more » « less
  5. Abstract

    We study quantum decoherence numerically in a system consisting of a relativistic quantum field theory coupled to a measuring device that is itself coupled to an environment. The measuring device and environment are treated as quantum, non-relativistic particles. We solve the Schrödinger equation for the wave function of this tripartite system using exact diagonalization. Although computational limitations on the size of the Hilbert space prevent us from exploring the regime where the device and environment consist of a truly macroscopic number of degrees of freedom, we nevertheless see clear evidence of decoherence: after tracing out the environment, the density matrix describing the system and measuring device evolves quickly towards a matrix that is close to diagonal in a subspace of pointer states. We measure the speed with which decoherence spreads in the relativistic quantum field theory for a range of parameters. We find that it is less than the speed of light but faster than the speed of the massive charges in the initial state.

     
    more » « less