Multi-Valued Variable Optimization via Parallel Transition Candidates

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In the context of using Ising machines for solving assignment problems with multi-valued variables, the increase in the number of instances of x_i complicates data movement and parallel trials, making it difficult to compute combinatorial optimization problems efficiently.

Innovation Solution

A data processing device that includes a memory to store evaluation function information for multi-valued variables and a processor to generate transition destination candidates, calculate change amounts, and specify transitions, allowing for parallel processing of multi-valued variables.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If multi-valued variables are transformed into binary state variables (0 or 1) to solve assignment problems using an Ising machine, then the problem can be solved using standard Ising model techniques, but the number of instances of x_i increases and data movement becomes complicated

Engineering Contradiction:
Improveability to solve assignment problemsVSAvoiddata movement complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent segments the multi-valued variable update process into independent parallel trials. Instead of transforming multi-valued variables into binary variables, it divides the search space into multiple candidate states that can be evaluated independently and in parallel, reducing data movement complexity while maintaining problem-solving capability

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a new dimension of parallel trials where multiple candidate states are evaluated simultaneously. This allows the system to handle multi-valued variables directly without binary transformation, as each trial represents a different possible value transition in the multi-valued variable space

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Adaptability or versatility

If the number of instances of x_i is increased to represent multi-valued variables, then the assignment problem can be modeled, but parallel trials become difficult and computational efficiency decreases

Engineering Contradiction:
Improvemulti-valued variable representationVSAvoidcomputational efficiency
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent performs preliminary generation of transition destination candidates before executing the update process. By pre-computing multiple candidate states and their associated evaluation function changes, the system enables parallel trials to proceed efficiently without the overhead of binary transformation and complex data movement during the optimization process

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent changes the parameter representation from binary state variables to multi-valued variables with direct transition capabilities. This parameter change allows the system to maintain computational efficiency by enabling parallel evaluation of multiple candidate states without requiring binary encoding and decoding operations

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4468208A1Data processing device, program, and data processing method
Publication Date: 2024.11.27 FUJITSU LTD
  • EP4468208A1 patent drawingFigure 1
  • EP4468208A1 patent drawingFigure 2
  • EP4468208A1 patent drawingFigure 3

AI summary

A data processing device including: a memory configured to store evaluation function information indicating an evaluation function of a combinatorial optimization problem represented by using a plurality of multi-valued variables; and a processor coupled to the memory, the processor being configured to perform processing including: generating, based on the evaluation function information, transition destination candidates that are a part of respective transitionable ranges of values of the plurality of multi-valued variables; calculating, for each of the plurality of multi-valued variables, a change amount of the values of the evaluation function associated with a transition to the transition destination candidates, based on the evaluation function information; specifying the multi-valued variables for which the transition is accepted, from among the plurality of multi-valued variables, based on the change amount; and causing the values of the specified multi-valued variables to transition to the values of the transition destination candidates.