home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!ogicse!pdxgate!caiq@eecs.cs.pdx.edu
- From: caiq@eecs.cs.pdx.edu (Qin Cai)
- Newsgroups: sci.math.num-analysis
- Subject: Least Square Estimation
- Message-ID: <6770@pdxgate.UUCP>
- Date: 28 Jan 93 03:46:31 GMT
- Article-I.D.: pdxgate.6770
- Sender: news@pdxgate.UUCP
- Organization: Portland State University, Portland, OR
- Lines: 18
-
- I have a Least Square Estimation problem which took me a long time but
- still couldn't figure out the the method to solve it. I'm sure that
- there is no expilict solution, what I want to know is the computer
- searching algorithm. Could somebody help me?
-
- The problem is to minimize the next function:
-
- \sum_{s} ||b_{s} - A_{s} *B *c_{s} ||^{2}
-
- where b_{s}: vector of T by 1,(T is very large, could be thousands), known
- A_{s}: Matrix of T by m, known
- B: Matrix of m by n,(m > n, m:n could be 100:5), unkown.
- c_{s}: vector of n by 1, unknown.
-
- \sum_{s} means sum over s, s is the foot note.
-
- Any helpful suggestion is appreciated.
-
-