Parallel Search Unit Feedback for Faster Combinatorial Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information processing systems face challenges in efficiently finding optimal solutions for combinatorial optimization problems, as they often require extensive time to obtain optimal solutions and have low solution finding performance due to independent ground state searches.
Innovation Solution
An information processing system with multiple search units that share and compare solutions using an Ising-type energy function, where each unit generates a new state variable string based on the best solution found, allowing for synchronized or asynchronous searches to improve the chances of finding optimal solutions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple search units independently perform ground state searches, then the search coverage is increased, but the time required to find the optimal solution increases and the possibility of obtaining the optimal solution decreases
Solution Approach 1:
The patent implements feedback by having search units share information about their found solutions. Each search unit transmits solution information to other search units, which then uses this information to update their own search states. This feedback mechanism allows search units to learn from each other's discoveries, increasing the overall probability of finding the optimal solution while maintaining efficient parallel search operations.
Solution Approach 2:
The patent merges the search efforts of multiple search units by combining their solution information. Instead of independent searches that waste time, the system combines the findings from multiple search units and uses this combined information to guide subsequent searches. This merging approach increases solution finding performance by leveraging collective search results.
2Productivity
If multiple search units independently perform ground state searches, then parallel processing capability is utilized, but the solution finding performance remains insufficient
Solution Approach 1:
The feedback mechanism enables search units to share successful search paths and solutions. When one search unit finds a good solution, this information is fed back to other search units, which can then use it to improve their own search efficiency. This increases both productivity by reducing redundant searches and reliability by propagating successful search strategies across all units.
Solution Approach 2:
The system performs preliminary actions by having search units exchange solution information before completing their full search cycles. This preliminary sharing of findings allows other search units to adjust their search strategies in advance, avoiding wasted computational effort on already-explored regions and improving overall solution finding performance.
3Loss of time
If the system uses traditional sequential search methods, then the search process is simple to implement, but the time required to obtain the optimal solution is excessive
Solution Approach 1:
The patent segments the search process into multiple independent search units that operate in parallel. Each search unit maintains its own search state and can independently explore the solution space. This segmentation reduces the time required to find optimal solutions by distributing the search workload, while the complexity is managed through standardized communication protocols between segments.
Solution Approach 2:
The system transitions from sequential search to parallel search by adding a temporal dimension to the search process. Multiple search units operate simultaneously in different time steps, exchanging information across this new dimension. This dimensional change dramatically reduces search time while the complexity increase is offset by the regular structure of inter-unit communication.
Data Source
AI summary
An information processing apparatus includes a memory and a processor. The processor configured to acquire, among a plurality of solutions, a first solution corresponding to a best value among values of a plurality of the energy functions corresponding to the plurality of solutions, each of the plurality of solutions being represented by a value of each of a plurality of state variables included in an energy function, generate a first state variable string based on the first solution, and search for a solution with the first state variable string as a start state.