j********3
发帖数: 560 | 1 【 以下文字转载自 Computation 讨论区 】
发信人: johnlee123 (no), 信区: Computation
标 题: 请教Matlab中矩阵求逆问题
发信站: BBS 未名空间站 (Mon Feb 19 13:44:00 2007)
请问对于一个接近奇异的矩阵，应该如何去求它的逆矩阵？Matlab自己的inv函数不能
给出准确的结果，运行之后给出如下提示
Warning: Matrix is close to singular or badly scaled.
Results may be inaccurate. RCOND = 7.828442e-018.
将所得矩阵和原矩阵相乘，得到的也不是单位矩阵，并且差得也很大。请问有什么方法
来解决这个问题？多谢了！ |
T******r
发帖数: 2937 | 2 pinv?
【在 j********3 的大作中提到】
: 【 以下文字转载自 Computation 讨论区 】
: 发信人: johnlee123 (no), 信区: Computation
: 标 题: 请教Matlab中矩阵求逆问题
: 发信站: BBS 未名空间站 (Mon Feb 19 13:44:00 2007)
: 请问对于一个接近奇异的矩阵，应该如何去求它的逆矩阵？Matlab自己的inv函数不能
: 给出准确的结果，运行之后给出如下提示
: Warning: Matrix is close to singular or badly scaled.
: Results may be inaccurate. RCOND = 7.828442e-018.
: 将所得矩阵和原矩阵相乘，得到的也不是单位矩阵，并且差得也很大。请问有什么方法
: 来解决这个问题？多谢了！
|
j********3
发帖数: 560 | 3 谢谢！试了pinv，它能给出更好的结果，但是还是不够精确。应该怎么办呢？
【在 T******r 的大作中提到】
: pinv?
|
N**D
发帖数: 10322 | 4 what is your goal of computing the inverse?
most of time, you don't need the explicit form of the inverse.
【在 j********3 的大作中提到】
: 谢谢！试了pinv，它能给出更好的结果，但是还是不够精确。应该怎么办呢？
|
j********3
发帖数: 560 | 5 My goal actually is to solve a linear equation system. And at the beginning
I actually used "\", which is said to choose the optimized algorithm. But it
gave the warning information the same as when I used "inv" function. Thus I
wonder whether there is better approach to deal with a matrix close to
singular. Thank you.
【在 N**D 的大作中提到】
: what is your goal of computing the inverse?
: most of time, you don't need the explicit form of the inverse.
|
D*******a
发帖数: 3688 | 6 bad condition?那没啥办法
beginning
it
I
【在 j********3 的大作中提到】
: My goal actually is to solve a linear equation system. And at the beginning
: I actually used "\", which is said to choose the optimized algorithm. But it
: gave the warning information the same as when I used "inv" function. Thus I
: wonder whether there is better approach to deal with a matrix close to
: singular. Thank you.
|
