Hewlett Packard Enterprise Development LP

Thomas Van Vaerenbergh

Jason Pelc

Raymond G Beausoleil

John Paul Strachan

Suhas Kumar

David Kielpinski

Nikolas Anton Tezak

Catherine Graves

Mir Ashkan Seyedi

Can Li

Ranojoy Bose

Peng Sun

Marco Fiorentino

Di Liang

Yuan Yuan

Bassem Tossoun

Antoine Descos

Gabriel Joel Mendoza

Sean Hooten

Gabriel Joe Mendoza

Shuren Hu

Systems and methods for an optical ternary content addressable memory (TCAM) are provided. The optical TCAM implements a time-division multiplexing (TDM) based encoding scheme to encode each bit position of a search word in the time domain. Each bit position is associated with at least two time slots. The encoded optical signal comprising the search word is routed through one or more modulators configured to represent a respective TCAM stored word. If a mismatch between at least one bit position of the search word and at least one TCAM stored word occurs, a photodetector or photodetector array will detect light.

Time division multiplexing (TDM) based optical ternary content addressable memory (TCAM)

Embodiments of the present disclosure provide etch-variation tolerant optical coupling components and processes for making the same. An etch-variation tolerant geometry is determined for at least one waveguide of an optical coupling component (e.g., a directional coupler). The geometry is optimized such that each fabricated instance of an optical component design with the etch-variation tolerant geometry has substantially the same coupling ratio at any etch depth between a shallow etch depth and a deep etch depth.

Etch variation tolerant directional couplers

A method for determining a solution to a constrained optimization problem includes programming a weights matrix of a Hopfield network with a first encoded matrix representation of an initial constrained optimization problem. The method also includes employing the Hopfield network to determine a solution to the initial constrained optimization problem. Additionally, the method includes encoding a plurality of constrained optimization problems associated with a target constrained optimization problem into a plurality of encoded matrix representations each of which are a combination of the first and the second encoded matrix representations. The plurality of encoded matrix representations increases in convergence to the second encoded matrix representation of the target constrained optimization problem sequentially. The method further includes re-programming the weights matrix of the Hopfield network in an iterative manner with the plurality of encoded matrix representations.

Adiabatic annealing scheme and system for edge computing

Systems and methods are provided for implementing a hardware accelerator. The hardware accelerator emulates a neural network, and includes a memristor crossbar array, and a non-linear filter. The memristor crossbar array can be programmed to calculate node values of the neural network. The nodes values can be calculated in accordance with rules to reduce an energy function associated with the neural network. The non-linear filter is coupled to the memristor crossbar array and programmed to harness noise signals that may be present in analog circuitry of the hardware accelerator. The noise signals can be harnessed such that the energy function associated with the neural network converges towards a global minimum and modifies the calculated node values. In some embodiments, the non-liner filter is implemented as a Schmidt trigger comparator.

Systems and methods for harnessing analog noise in efficient optimization problem accelerators

Recurrent neural networks, and methods therefor, are provided with diagonal and programming fluctuation to find energy global minima. The method may include storing the matrix of weights in memory cells of a crossbar array of a recursive neural network prior to operation of the recursive neural network; altering the weights according to a probability distribution; setting the weights to non-zero values in at least one of the memory cells in a diagonal of the memory cells in the crossbar array; and operating the recursive neural network.

Recurrent neural networks with diagonal and programming fluctuation to find energy global minima

Systems are provided for implementing a hardware accelerator. The hardware accelerator emulate a stochastic neural network, and includes a first memristor crossbar array, and a second memristor crossbar array. The first memristor crossbar array can be programmed to calculate node values of the neural network. The nodes values can be calculated in accordance with rules to reduce an energy function associated with the neural network. The second memristor crossbar array is coupled to the first memristor crossbar array and programmed to introduce noise signals into the neural network. The noise signals can be introduced such that the energy function associated with the neural network converges towards a global minimum and modifies the calculated node values.

Systems for introducing memristor random telegraph noise in Hopfield neural networks

Staged neural networks and methods are described herein. In some embodiments, the methods may identify a plurality of second NP hard/complete problems that are similar to the first NP hard/complete problem and identify solutions to the second NP hard/complete problems. The methods may train a deep neural network with the second NP hard/complete problems and the solutions. The methods may provide the first NP hard/complete problem to the trained deep neural network to generate a preliminary solution to the first NP hard/complete problem and provide the preliminary solution to a recursive neural network configured to execute an energy minimization search. The recursive neural network may generate a final solution to the problem based on the preliminary solution.

Staged neural networks for solving NP hard/complete problems

An optical grating coupler defining an axis and configured to couple light between a planar waveguide and an optical fiber, including first and second entry surfaces and a plurality of scattering regions symmetric to the axis and arranged such scattering strength presented to incoming light by the plurality of scattering regions changes from weak to strong along a beam path of the incoming light to match a Gaussian mode profile of the optical fiber.

Polarization diversity grating couplers with low loss and zero PDW/PDL

One embodiment provides an optical encoder. The optical encoder includes an optical comb source to generate a multi-wavelength optical signal; a number of optical filters sequentially coupled to the optical comb source, with a respective optical filter being tunable to pass or block a particular wavelength of the multi-wavelength optical signal based on a corresponding bit value of a multi-bit search word; and a common output for the optical filters to output the filtered multi-wavelength optical signal, which encodes the multi-bit search word and can be used as an optical search signal for searching an optical content-addressable memory (CAM).

Optical comb source for content-addressable memory encoders

An apparatus includes a photodetector and a memristor coupled to the photodetector. The photodetector is configured to receive and convert optical signals to electrical signals to program the memristor to an on or off state. The apparatus further includes a ring resonator coupled to the memristor and configured to modulate light based on the on or off state of the memristor.

Optical switchable system and device

In example implementations, an apparatus includes a plurality of nodes, a pump coupled to the plurality of nodes and a connection network. In one example, each one of the plurality of nodes may store a value. The pump provides energy to the each one of the plurality of nodes. The connection network may include a two dimensional array of elements, wherein each group of the two dimensional array of elements is in communication with a respective one of the plurality of nodes, wherein the connection network may be tuned with parameters associated with encoding of an Ising problem. The connection network may process the value stored in each one of the plurality of nodes. The Ising problem may be solved by the value stored in each one of the plurality of nodes at a minimum energy level.

Apparatus for solving Ising problems

Systems and methods for an optical ternary content addressable memory (TCAM) is provided. In various embodiments, one or more search words can be encoded in a multi-wavelength input signal. Each bit position associated with a set of wavelengths of the input signal, each wavelength corresponding to a logic value. A plurality of copies of the input signal can be coupled to an optical search engine comprising a plurality of rows of stored words. In various embodiments, the search word may be encoded in the amplitude of a single wavelength. Each bit position can be associated with a set input waveguides, and a logic value can be encoded based on whether amplitude of the associated wavelength is detected on a respective input waveguide of the set of waveguides. A mismatch of at least one bit is indicated if light is detected on an output of the optical TCAM.

Wavelength division multiplexing (WDM)-based and multipath interferometry based optical ternary content addressable memory (TCAM)

Examples herein relate to single-etch layer grating couplers. In particular, implementations herein relate to single-etch layer grating couplers comprising a planar waveguide wherein a first grating section and a second grating section are etched into. The first grating section is configured to scatter light at a given wavelength at a first angle relative to vertical and the second grating section is configured to scatter light at the given wavelength at a second angle relative to vertical. The stagger-tuning of the first grating section and the second grating section results in the light at the given wavelength being scattered at a desired angle the same or near an angle at which an optical fiber is offset from vertical.

Single-etch wide-bandwidth grating couplers with individually-tuned grating sections

A photonic circuit design system includes a photonic circuit design tool to facilitate user inputs to generate a photonic circuit netlist comprising a photonic design component of a photonic circuit design. The system includes a memory system to store the photonic circuit netlist and a component library comprising a plurality of predetermined photonic design components from which the photonic design component is selected. The component library further includes physical data associated with physical characteristics of the plurality of predetermined photonic design components. The system further includes a photonic circuit generator to access the photonic circuit netlist from the memory and the associated photonic design component from the component library based on the photonic circuit netlist and to implement hierarchical algorithm associated with a physical layout and an optical interconnection of the photonic design component in a photonic circuit package based on the physical characteristics to generate the photonic circuit design.

Growing community of inventors

Palo Alto, CA, United States of America

Thomas Van Vaerenbergh