数值分析 6.Direct Methods for Solving Linear Systems-CSDN博客

本文链接：https://blog.csdn.net/HGGshiwo/article/details/112758492

求解线性方程组Ax=b

1. Gaussian Elimination

高斯消除法，首先将矩阵化为上三角阵，然后从后解未知数

Step 1:如果a11不为0，则

,

(第2到n行减去mi1*row1)

Step k:如果aii不为0，则

从第k行到n行减去上一行乘n

得到上三角矩阵：

Backward-substitution：

次数计算：

消元法：共执行n-1步，

乘法:a：(n-k)(n-k)次，

b：(n-k)次，

除法：m：(n-k)次

乘除法：

向后替代：

除法：xn：1次

除法：xi：n-i+1次

2.Pivoting Strategies

较小的aii会导致系数变大,需要预先进行处理

Partial Pivoting (or maximal column pivoting)

1.在第k行准备往下减前，首先找到下方k列绝对值最大的元素的列p

2.将p和k行进行交换

O(n^2)的比较

Scaled Partial Pivoting

1.算法开始前将每行进行归一（除每行最大的元素）

2.和上方的算法一样

O(n^2)的比较和O(n^2)的除法

Complete Pivoting (or maximal pivoting)

1.对于第i行，寻找其下方矩阵中绝对值最大的元素

2.将该元素通过行列交换到aii位置

O(n^3/3)的比较

3.LU分解

如果不需要行互换就可以完成高斯分解，则可以LU分解

LU分解的具体步骤：

https://blog.csdn.net/qq_40688707/article/details/89256737

4.Special Types of Matrices

Strictly Diagonally Dominant Matrix：

主对角元的绝对值严格大于该行的其他元素值

1. A strictly diagonally dominant matrix A is nonsingular.

2.Gaussian elimination can be performed without row or column interchanges, and the computations will be stable with respect to the growth of roundoff errors.

Positive Definite Matrix:

正定阵：对称且任意x, x^TAx>0

1.(=>)逆矩阵正定，主对角线大于0.

2.(=>)max | aij | <=max | akk |; ( aij )^2 < aiiajj for each i 不等于 j.

3.(<=>)Each of A’s leading principal submatrices Ak has a positive determinant.

4.<=>可以分解为 LDL^t ,L为unitary lower-triangular matrix，D为diagonal matrix with positive diagonal entries.

5.<=>A can be factored in the form LL^T , where L is lower-triangular with nonzero diagonal entries.

Algorithm: Choleski’s Method(正定阵进行LLT分解)

To factor the symmetric positive definite n*n matrix A into LLt, where L is lower-triangular.

Input: the dimension n; entries aij for 1<=i, j <=n of A.

Output: the entries lij for 1<=j <=i and 1<=i <=n of L.

Step 1 Set

Step 2 For j = 2, …, n, set

Step 3 For i = 2, …, n-1do steps 4 and 5

Step 4 Set

Step 5 For j = i+1, …, n, set

Step 6 Set

Step 7 Output ( lij for j = 1, …, i and i = 1, …, n );

STOP.