Precomputed Embedding Library for Quantum Annealer Optimization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Determining optimal embeddings for quantum annealers is computationally difficult and time-consuming, as it requires significant computational effort on classical computers, which can be inefficient given the NP-hard nature of the problem.

Innovation Solution

A library of precomputed optimized embeddings for clique graphs of varying sizes is created and stored, allowing for quick retrieval and modification to generate problem-specific embeddings for quantum annealers, reducing the computational burden on classical computers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If optimal embeddings are computed using conventional algorithms on classical computers, then embedding accuracy is improved, but computational time increases significantly

Engineering Contradiction:
Improveembedding accuracyVSAvoidcomputational time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent precomputes and stores optimal embeddings for clique graphs of various sizes in a library before they are needed. When a quantum annealing problem requires an embedding, the system retrieves the precomputed embedding from the library rather than computing it in real-time. This preliminary action resolves the contradiction by providing accurate embeddings (maintaining embedding accuracy) while eliminating the computational time requirement at the moment of use.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent creates copies of precomputed clique graph embeddings and stores them in a library for reuse. Instead of recomputing embeddings each time they are needed, the system copies and retrieves stored embeddings that match the problem requirements. This copying approach maintains embedding accuracy through reuse of proven embeddings while significantly reducing computational time by avoiding redundant calculations.

Inventive Principle:
Principle #26Copying

2Speed

If precomputed embeddings are stored in a library for quick retrieval, then computational speed is improved, but memory storage requirements increase

Engineering Contradiction:
Improveembedding computation speedVSAvoidlibrary storage space
Core Design Contradiction:
SpeedVSVolume of stationary object

Solution Approach 1:

The patent organizes the embedding library by parameters such as clique graph size and topology type. Embeddings are stored with indexed parameters that allow efficient retrieval based on problem characteristics. This parameter-based organization enables the system to quickly locate and retrieve only the specific embeddings needed for particular problem sizes, improving computational speed while optimizing storage space by avoiding unnecessary embeddings.

Inventive Principle:
Principle #35Parameter changes

3Adaptability or versatility

If a comprehensive library of embeddings for all possible clique sizes is precomputed, then adaptability to different problem sizes is improved, but the complexity of library management increases

Engineering Contradiction:
Improveproblem size coverageVSAvoidlibrary management complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent segments the embedding library into distinct sections or categories based on clique graph parameters such as size, topology, and other characteristics. Each segment contains embeddings for specific problem types and sizes. This segmentation improves adaptability by making it easy to select the appropriate segment for a given problem while reducing library management complexity through organized, modular structure that simplifies maintenance and updates.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4372627A1Persistent and parallel embeddings for quantum annealers
Publication Date: 2024.05.22 TRIMBLE INC
  • EP4372627A1 patent drawingFigure 1
  • EP4372627A1 patent drawingFigure 2A~2B
  • EP4372627A1 patent drawingFigure 3A~3B

AI summary

Computation of optimal embeddings for an optimization problem to be solved using a quantum annealer can be accelerated by pre-computing optimal embeddings in a target graph corresponding to a quantum processor architecture of clique graphs (or fully-connected network graphs) having various numbers of nodes. Pre-computed embeddings can be stored in a library. To solve an optimization problem, a clique-graph embedding of appropriate size can be retrieved from the library and modified to match the problem to be solved. The modified embedding can be executed using an appropriate quantum annealer.