Combinatorial Optimization Program with Dynamic Search Ranges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Local search methods for combinatorial optimization problems often get stuck in local solutions, leading to prolonged solution finding times.
Innovation Solution
A program that facilitates escape from local solutions by performing local searches in a first search range, determining if a current state is stuck, and then fixing equal values of state variables in a second search range based on elite solution candidates, using modified evaluation functions to guide state transitions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If local search is performed in a limited search range to reduce calculation time, then productivity is improved, but the system gets stuck in local solutions and cannot escape, worsening reliability
Solution Approach 1:
The patent dynamically adjusts the search range based on the current search state. When a local solution is detected, the system expands the search range by releasing fixed state variables, allowing the search to escape local optima. This dynamic adaptation resolves the contradiction by making the search range flexible rather than static, enabling both efficient local search and escape from local solutions when necessary
Solution Approach 2:
The patent changes the parameter of search range by fixing or releasing state variables. By controlling which state variables are fixed, the system can switch between narrow search ranges (for efficiency) and wide search ranges (for escaping local solutions). This parameter change mechanism allows the system to optimize between calculation speed and solution quality based on the current search situation
2Reliability
If the search range is expanded to escape local solutions, then reliability is improved, but calculation time increases, worsening productivity
Solution Approach 1:
The patent segments the search process into multiple phases with different search ranges. The search is divided into initial local search phases (with fixed state variables for efficiency) and escape phases (with released state variables for broader exploration). This segmentation allows the system to spend most time in efficient local search while periodically expanding the range to escape local solutions, balancing reliability and productivity
Solution Approach 2:
The patent implements periodic expansion of the search range by monitoring for local solution conditions and triggering escape mechanisms at appropriate intervals. This periodic action between narrow and wide search ranges allows the system to maintain high productivity during normal operation while periodically ensuring escape from local optima, resolving the contradiction through rhythmic alternation between the two modes
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A program including instructions causing a computer to execute: searching for a solution to a combinatorial optimization problem through a local search (LS), in a first search range (SR) that is based on state variables (SVs); determining whether a current state (CS) represented by a combination of values of the SVs has fallen into a local solution in a search in the first SR; in a case where it is determined that the CS has fallen into the local solution, determining a first solution candidate (SC), based on SCs; and performing, in a state in which a value of a first SV that has a mutually equal value between the CS and the first SC among the SVs is fixed, the LS in a second SR that is based on a second SV that has mutually different values between the CS and the first SC among the SVs.