Semiconductor Design Optimization via Dimensionality Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional optimization methods struggle to find optimal input parameters in simulations with a large number of parameters within a practical time frame, as the complexity increases.
Innovation Solution
An optimization apparatus that includes an output data acquirer, input/output data storage, evaluation value calculator, input parameter converter, and next-input parameter decider, which reduces the dimension number of input parameters through conversion and repeats operations until a predetermined condition is met, using techniques like principal component analysis and Bayesian optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional optimization algorithms are used to search for optimum input parameters, then optimal parameters can be found with high precision, but the optimization time becomes impractically long when the number of input parameters increases
Solution Approach 1:
The patent segments the high-dimensional parameter space into multiple lower-dimensional subspaces by dividing the set of input parameters into multiple groups. Each subspace is then optimized independently using conventional algorithms, which reduces the computational complexity and optimization time while maintaining acceptable precision in each segment.
Solution Approach 2:
The patent transforms the original high-dimensional optimization problem into multiple lower-dimensional problems by changing the dimensionality through parameter grouping. This dimensional reduction allows conventional optimization algorithms to operate efficiently in each lower-dimensional subspace, thereby reducing overall optimization time while preserving essential optimization accuracy.
2Productivity
If the number of input parameters is reduced through dimensionality reduction, then optimization time decreases, but the ability to represent the full parameter space is compromised
Solution Approach 1:
The patent segments the complete parameter space into multiple lower-dimensional subspaces that collectively cover the entire original space. By optimizing each segment independently and combining results, the method maintains comprehensive parameter space coverage while achieving faster optimization speeds in each individual segment.
Solution Approach 2:
The patent merges multiple lower-dimensional optimization results to reconstruct the solution for the original high-dimensional problem. This combining approach ensures that the full parameter space is represented in the final solution, compensating for the dimensional reduction in individual segments and maintaining overall adaptability.
Data Source
AI summary
An optimization apparatus includes an output data acquirer to acquire output data expressing a result of experiment or simulation based input parameters, input/output data storage to store the input parameters and the output data corresponding to the input parameters, as a pair, an evaluation value calculator to calculate evaluation values of the output data, an input parameter converter to generate conversion parameters of a dimension number changed from the dimension number of the input parameters, a next-input parameter decider to decide next input parameters based on pairs of the conversion parameters and the evaluation values corresponding to the conversion parameters, and a repetition determiner to repeat operations of the output data acquirer, the input/output data storage, the evaluation value calculator, the input parameter converter, and the next-input parameter decider, until satisfying a predetermined condition.