Annealing Unit Speculative Inversion for Combinatorial Optimization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for solving combinatorial optimization problems, such as the Ising calculation device, face difficulties in efficiently transitioning from high energy states to lower energy states, especially when constraints like even number of spins are involved, leading to inefficient searches for minimum energy solutions.

Innovation Solution

The proposed solution includes an optimization apparatus with a temperature control unit, annealing unit, speculative inversion control unit, adoption determination unit, and energy calculation unit, which speculatively inverts state variables to reach a predetermined number of changes, determining adoption based on energy changes and temperature, thereby improving the efficiency of state transitions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If sequential state transition with single spin inversion is used, then the method is simple and easy to implement, but it is difficult to efficiently transition to lower energy states when constraints like even number of spins are involved

Engineering Contradiction:
ImproveEase of implementationVSAvoidSearch efficiency
Core Design Contradiction:
Ease of manufactureVSProductivity

Solution Approach 1:

The invention segments the spin inversion process into multiple independent inversion operations that can be performed in parallel. Instead of inverting one spin at a time sequentially, the system identifies multiple candidate spins and inverts them simultaneously, dividing the search space exploration into parallel paths that can be evaluated independently and then combined.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The invention performs preliminary identification of candidate spins that satisfy inversion conditions before actually executing the inversions. The system pre-calculates which spins should be inverted based on current state and constraints, then executes these pre-determined inversions in parallel, avoiding the need for sequential decision-making during the inversion process itself.

Inventive Principle:
Principle #10Preliminary action

2Productivity

If speculative inversion of multiple state variables is performed, then the transition efficiency to lower energy states improves, but the complexity of the apparatus increases

Engineering Contradiction:
ImproveSearch efficiencyVSAvoidApparatus complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The apparatus is segmented into distinct functional units: a condition determination unit that identifies candidate spins, a speculative inversion unit that performs parallel inversions, and an energy evaluation unit that assesses results. This modular segmentation allows complex speculative inversion operations to be broken down into manageable, independently implementable components.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The invention introduces an intermediary selection mechanism that chooses which spins to invert speculatively based on predetermined criteria. This intermediary layer filters the search space to identify promising candidate spins for speculative inversion, reducing the overall complexity by focusing computational resources on the most likely beneficial inversions rather than evaluating all possible spin combinations.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Stability of the object's composition

If deterministic adoption of state transitions is used, then the energy change is monotonically decreasing, but the system gets stuck in local solutions and cannot reach the global optimum

Engineering Contradiction:
ImproveEnergy monotonicityVSAvoidSolution optimality
Core Design Contradiction:
Stability of the object's compositionVSMeasurement precision

Solution Approach 1:

The system performs preliminary evaluation of multiple potential state transitions before committing to any single transition. By pre-calculating the energy changes associated with multiple speculative inversions and comparing them in advance, the system can select the most beneficial transition while maintaining deterministic adoption, thus avoiding local optima without sacrificing the stability of monotonic energy decrease.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The invention performs more inversion operations than traditionally necessary by speculatively inverting multiple spins simultaneously. This excessive action in the form of parallel speculative inversions allows the system to explore a broader portion of the search space and identify transitions that lead to global optima, while still maintaining deterministic adoption by selecting only those transitions that improve the overall energy state.

Inventive Principle:
Principle #16Partial or excessive action

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach enhances the processing performance for combinatorial optimization problems by allowing more efficient transitions to lower energy states, improving the overall efficiency of the search process.

Implementation Method 1

an annealing unit, configured to change a state of any one of a plurality of state variables included in an evaluation function that represents energy

Methodology Applied
Scientific EffectAnnealing: Annealing

Data Source

PatentEP3975032A1Optimization apparatus, optimization method, and optimization program
Publication Date: 2022.03.30 FUJITSU LTD
  • EP3975032A1 patent drawingFigure 1
  • EP3975032A1 patent drawingFigure 2
  • EP3975032A1 patent drawingFigure 3A

AI summary

An optimization apparatus includes an annealing unit changes a state of any one of a plurality of state variables included in an evaluation function, calculates a change amount of energy, and obtains a first total change amount by adding a second total change amount and the calculated change amount, a speculative inversion control unit repeats a process for speculatively selecting the state variable to be changed and making the annealing unit obtain the first total change amount, an adoption determination unit stochastically determines whether or not to adopt a state transition in which a predetermined number of the state variables are changed by the annealing unit, an energy calculation unit calculates transited energy when the state transition being determined to be adopted, and a search unit specifies the transited energy as a minimum energy when the transited energy is less than a previously specified minimum energy.