Least Squares Computations in Science and Engineering
Abstract
Least squares computations constitute a fundamental tool in science and engineering. The reason is that they play a critical role in fitting numerical models to real world observations. This AFOSR supported research effort has been concerned with the design and testing of new algorithms for least squares computations and optimization in science and engineering. The objectives were to mathematically develop, test, and analyze fast numerical algorithms for the efficient solution to problems on modem high performance computers. The focus of this project was the application of scientific computing technology in the area of signal and image processing. Very many problems lead to over determined systems of linear or nonlinear equations that are often solved by least squares or related optimization methods. Generally, the problems are accompanied by constraints, such as bound constraints, and the observations are corrupted by noise. The project has involved the application of scientific computing in the area of computational linear and nonlinear least squares methods with particular applications in image and signal processing, where recovering images is often an ill-posed inverse problem. Additional work included control computations associated with adaptive optics. Constrained least squares, Adaptive filtering, Adaptive optics, Deconvolution, Image restoration, Parallel algorithms, Trace maximization, Inverse problems, FFT.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 1994
- Accession Number
- ADA281213
Entities
People
- Robert J. Plemmons
Organizations
- Wake Forest University