Data Processing Device for Adaptive Constraint Weighting in MCMC Search

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The efficiency of Markov-Chain Monte Carlo (MCMC) search for solving combinatorial optimization problems is adversely affected by inappropriate weighting of constraint conditions, leading to either frequent constraint violations or reduced state transitions.

Innovation Solution

A data processing device and method that adjusts the coefficient values corresponding to constraint conditions during the search, increasing them when violations occur and determining their maintenance or decrease based on energy and evaluation function comparisons.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the coefficient representing the weight of the constraint condition is increased, then constraint violations are reduced, but state transitions become less likely to occur

Engineering Contradiction:
Improveconstraint satisfactionVSAvoidsearch efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent applies dynamics by making the coefficient values dynamic rather than static. The processing unit adjusts coefficient values during the MCMC search based on constraint violation detection and energy comparisons. When constraint violations occur, the corresponding coefficient is increased; when constraints are satisfied and energy improves, coefficients may be decreased. This dynamic adjustment allows the system to balance constraint satisfaction and search efficiency adaptively throughout the search process.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent implements parameter changes by modifying the coefficient values of constraint terms during the optimization search. The processing unit changes these parameters based on real-time feedback from constraint violation detection and energy evaluation. This allows the system to adapt the weighting of different constraint conditions dynamically, optimizing both constraint satisfaction and solution search efficiency throughout the MCMC process.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If the coefficient representing the weight of the constraint condition is decreased, then state transitions occur more frequently, but constraint violations increase

Engineering Contradiction:
Improvesearch efficiencyVSAvoidconstraint satisfaction
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent applies feedback by using constraint violation information and energy evaluation results to adjust coefficient values. The processing unit continuously monitors whether constraints are violated during state transitions and uses this feedback to increase the corresponding coefficient. Conversely, when constraints are satisfied and energy improves, the system provides feedback to potentially decrease coefficients. This closed-loop feedback mechanism ensures that constraint satisfaction and search efficiency are continuously optimized based on actual search performance.

Inventive Principle:
Principle #23Feedback

3Productivity

If the coefficient value is set appropriately to balance constraint satisfaction and state transitions, then search efficiency improves, but requires complex adjustment mechanisms

Engineering Contradiction:
Improvesearch efficiencyVSAvoidcoefficient adjustment mechanism
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent implements self-service by enabling the system to automatically adjust its own coefficient values without external intervention. The processing unit autonomously monitors constraint violations, evaluates energy changes, and modifies coefficient values based on predefined adjustment rules. This self-adjusting mechanism eliminates the need for manual coefficient tuning or complex external control systems, achieving adaptive optimization through the system's own operational feedback.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4592906A1Data processing device, processing program, and data processing method
Publication Date: 2025.07.30 FUJITSU LTD
  • EP4592906A1 patent drawingFigure 1
  • EP4592906A1 patent drawingFigure 2
  • EP4592906A1 patent drawingFigure 3

AI summary

A data processing device includes a storage unit that stores evaluation function information of an evaluation function of a combinatorial optimization problem represented by a sum of a quadratic cost term and a linear cost term that is a sum of a plurality of constraint terms weighted by a coefficient that represents a weight of each of a plurality of constraint conditions, and a processing unit that acquires the evaluation function information from the storage unit, searches for a solution to the combinatorial optimization problem based on the evaluation function information, increases a value of the coefficient that corresponds to a first constraint condition in a case where there is the first constraint condition in which constraint violation occurs among the plurality of constraint conditions at a first time point during the search for the solution, and determines whether to decrease or maintain the value of the coefficient that corresponds to the plurality of constraint conditions based on a result of comparison between a value of the quadratic cost term at the first time point and a value of the evaluation function obtained before the first time point in a case where the plurality of constraint conditions is satisfied at the first time point.