home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!haven.umd.edu!darwin.sura.net!zaphod.mps.ohio-state.edu!wupost!waikato.ac.nz!comp.vuw.ac.nz!cc-server4.massey.ac.nz!TMoore@massey.ac.nz
- Newsgroups: sci.math
- Subject: Re: Need Help on Matrix Differentiation
- Message-ID: <1992Sep8.202957.5617@massey.ac.nz>
- From: news@massey.ac.nz (USENET News System)
- Date: Tue, 8 Sep 92 20:29:57 GMT
- References: <18h8pnINNdp2@matt.ksu.ksu.edu> <TZq-0-=@engin.umich.edu>
- Organization: Massey University
- Lines: 37
-
- In article <TZq-0-=@engin.umich.edu>, takriti@engin.umich.edu (samer Takriti) writes:
- >
- > In article <18h8pnINNdp2@matt.ksu.ksu.edu> bubai@matt.ksu.ksu.edu (P.Chatterjee) writes:
- > >Hi,
- > >
- > >I was just wondering if somebody could help out regarding matrix differentiationThe problem is to minimize:
- > >
- > >y= x'Ax + 2x1 + 3x2 - 10
- > >
- > >where A is a 2x2 matrix : 25 7
- > > 7 13
- > >
- > >
- > >and x is the (x1 x2) column vector and x' denotes the transpose.
- > >
- > >It's easy if one expands the x'Ax term but I was wondering if there was some way doing it using matrix differentiation.
- > >
- > >Thanks for any help in this regard.
- > >
- >
- > Write the problem as:
- > y = x'.A.x + c'.x
- > (The constant is not important),
- > c = (2, 3)'.
- > You need the derivative to be zero, i.e.,
- > 2.A.x + c = 0
- > x = -0.5 A^(-1) c
- > x is the solution for your problem. Make sure that A is
- > positive definite, this will guarantee that x is a minimum
- > otherwise it may be a maximum.
- > -Samer
- >
- Sorry, ignore my last posting. A _was_ symmetric.
-
- Unfortunately, my newsreader doesn't have the ability to cancel.
-
- Terry Moore
-