Ising Machine Penalty Component for Duplicate Solution Avoidance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Ising machines struggle to efficiently calculate multiple solutions for 0-1 optimization problems without duplicates, leading to increased computational costs due to the need for repeated runs to find alternative and near-optimal solutions.
Innovation Solution
A calculation device equipped with an improved algorithm that adds a penalty component to the momentum of particles in the ballistic or discrete SB algorithm, preventing the calculation of duplicate solutions by shifting particle positions towards constrained solutions, thereby reducing computational expense.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the Ising machine is run multiple times to find multiple alternative solutions, then the quantity of solutions is improved, but the computational cost increases due to duplicate solutions
Solution Approach 1:
The patent applies preliminary anti-action by introducing a penalty term in the objective function that prevents the system from converging to previously found solutions. The penalty term is constructed using the already-found solutions, and it actively repels the particle trajectories away from these solutions during the optimization process. This allows the Ising machine to generate diverse solutions in a single run without requiring multiple executions, thereby reducing computational cost while maintaining solution quantity.
Solution Approach 2:
The patent changes the parameters of the optimization problem dynamically by modifying the objective function to include time-dependent penalty terms. The penalty coefficient is adjusted during the evolution process, and the set of constrained solutions is updated as new solutions are found. This parameter change strategy enables the system to adaptively explore different regions of the solution space, generating multiple distinct solutions without repeating the same computational paths.
2Adaptability or versatility
If the Ising machine is run multiple times to avoid duplicate solutions, then the diversity of solutions is improved, but the time consumption increases
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing the penalty terms based on previously found solutions before initiating the optimization process. The penalty landscape is prepared in advance, incorporating information about constrained solutions, which guides the particle trajectories away from known solutions from the start. This preliminary preparation enables the system to efficiently generate diverse solutions in a single run without requiring iterative executions, thereby reducing time consumption while maintaining solution diversity.
Solution Approach 2:
The patent maintains continuity of useful action by integrating the penalty term calculation and solution generation into a single continuous optimization process. Rather than stopping, analyzing duplicates, and restarting multiple times, the system continuously evolves the particle trajectories while the penalty term actively prevents convergence to duplicate solutions. This continuous process with built-in diversity enforcement eliminates idle time between runs and achieves solution diversity more efficiently.
3Measurement precision
If a penalty component is added to prevent duplicate solutions, then the accuracy of solution distinctness is improved, but the device complexity increases
Solution Approach 1:
The patent implements feedback by using the found solutions to construct penalty terms that are fed back into the objective function. The penalty term is calculated based on the current particle positions and the set of already-found solutions, providing continuous feedback that steers the particles away from duplicate configurations. This feedback mechanism ensures high accuracy in solution distinctness while maintaining relatively simple implementation, as the penalty calculation reuses existing solution data without requiring complex additional hardware or algorithms.
Data Source
AI summary
A calculation device includes an updating circuit. The updating circuit updates, for each of N particles, a first variable representing the position of a target particle and a second variable representing momentum of the target particle. M constrained solutions each include N constrained values. The first variable is updated to change to a first value when the first variable is smaller than the first value and change to a second value when the first variable is greater than the second value. The second variable is updated based on the first variable of each particle and a penalty component of the target particle. The penalty component represents momentum for shifting the position of the target particle toward an opposite polarity. The penalty component indicates a value being greater as the first variable corresponding to the target particle is closer to the M constrained solutions.