Linear Algebra 1: Projection and Least Square Problem

最新推荐文章于 2024-11-11 22:14:35 发布

frankchenfu

最新推荐文章于 2024-11-11 22:14:35 发布

阅读量135

点赞数

分类专栏：数学文章标签：线性代数线性回归

本文链接：https://blog.csdn.net/frankchenfu/article/details/128511067

版权

在解决线性方程组Ax=b时，如果A是瘦高矩阵，可能存在无解的情况。此时，我们寻求使Ax-b的模长最小的x^。最小二乘问题要求找到使Ax-b的模长最小的x^。当b不在A的列空间内时，x^是b到由Ax定义的平面的投影。通过解方程ATAx=ATb，即正常方程，可以找到最小二乘解。若ATA可逆，则x^=(ATA)^{-1}ATb。

摘要由CSDN通过智能技术生成

Summarized from CUHK(SZ) MAT2041(2022Fall) Lecture 16.

Definition

Using Gauss-Jordan Elimination, we can solve the linear system $A x = b$ , where $A$ is a $n\times n$ matrix and $x, b$ are $n\times 1$ vectors. However, in real world, matrix $A$ is not always a square matrix. There are thousands of people rating for a few aspects of one thing. Thus $A$ is usually a tall matrix. If $A$ is a tall matrix, the solution set to $A x = b$ may be empty. Then we want to find a best approximation of $x$ that minimizes