Integral Equation Matrix Approximation via Adaptive Cross
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for solving large-scale electromagnetic applications with integral equations are computationally expensive, requiring significant memory and processing resources, especially when dealing with complex objects and frequency domains.
Innovation Solution
The system matrix is represented as two partial matrices, Znear and Zfar, where Znear handles near interactions and Zfar is approximated using the Adaptive Cross Approximation (ACA) algorithm based on Gaussian points, reducing computational complexity and resource requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods are used to solve large-scale integral equations, then accuracy is maintained, but computational time and memory requirements increase significantly
Solution Approach 1:
The system matrix is divided into two distinct parts: Znear (near-field interactions) and Zfar (far-field interactions). This segmentation allows different computational methods to be applied to each part, with the far-field part being approximated using ACA to reduce computational time while maintaining accuracy for the near-field components.
Solution Approach 2:
The patent applies Adaptive Cross Approximation (ACA) to transform the far-field matrix Zfar into a low-rank approximation. This parameter change in matrix representation reduces the computational complexity from O(N²) to O(N), significantly decreasing computational time while preserving the essential electromagnetic interaction characteristics.
2Measurement precision
If conventional methods are used to solve large-scale integral equations, then accuracy is maintained, but memory and processing resources increase significantly
Solution Approach 1:
By separating the system matrix into Znear and Zfar components, the patent enables selective application of computational resources. The memory-intensive far-field calculations are reduced through ACA approximation, while near-field accuracy is preserved, thereby reducing overall memory and processing resource requirements.
Solution Approach 2:
The application of ACA transforms the far-field matrix into a compressed low-rank format, dramatically reducing the memory storage requirements and processing complexity. This parameter change in matrix representation allows large-scale problems to be solved with feasible computational resources.
3Productivity
If the system matrix is divided into near and far parts with ACA approximation, then computational time and resources are reduced, but implementation complexity increases
Solution Approach 1:
The clear segmentation into near-field and far-field parts provides a structured implementation framework. This segmentation simplifies the overall algorithm design by allowing independent optimization of each part, with well-defined interfaces between them, thereby managing implementation complexity.
Solution Approach 2:
The patent introduces an intermediate classification mechanism that automatically determines which matrix elements belong to Znear and which belong to Zfar based on distance criteria. This intermediary classification layer simplifies the implementation by automating the complex decision-making process of matrix partitioning.
Data Source
AI summary
Systems and methods are provided for approximating an electric current on a surface of an object. A mesh is received. A matrix comprising a system of equations is generated based on the mesh. An entry of the matrix is associated with (i) a source function that approximates an electric current on the surface of the object, and (ii) a test function that tests a field of the source function on the surface. A near part of the matrix and a far part of the matrix are determined. The far part of the matrix is represented using a submatrix that is based on Gaussian points on elements of the mesh. The electric current on the surface of the mesh is calculated by solving the system of equations of the matrix.