home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!centerline!noc.near.net!news.Brown.EDU!qt.cs.utexas.edu!cs.utexas.edu!swrinde!mips!darwin.sura.net!haven.umd.edu!mimsy!afterlife!adm!aos!celms
- From: celms@vim.brl.mil (Dr. Aivars Celmins )
- Newsgroups: sci.math.stat
- Subject: Re: Parameter Covariance Matrix in Nonlinear Regression
- Message-ID: <984@aos.brl.mil>
- Date: 19 Aug 92 14:15:42 GMT
- References: <muzic.713536981@ds2.uh.cwru.edu>
- Sender: news@aos.brl.mil
- Organization: U.S. Army Ballistic Research Laboratory, APG, MD.
- Lines: 23
- Nntp-Posting-Host: vim.brl.mil
-
- In article <muzic.713536981@ds2.uh.cwru.edu> muzic@ds2.uh.cwru.edu (Ray Muzic) writes:
- >I have 2 references that give different answers. I believe one of the
- >answers is restricted to the case of a model that is linear in the
- >parameters. However, I seem to recall that for a nonlinear model, an
- >approximation that the model is linear about the optimum point is often
- >employed. Does this mean the covariance of the parameter estimates may be
- >estimated assuming the model is linear about the operating point?
- >
-
- The suggested linearization produces an approximation to the parameter
- covariance matrix that is less than first order accurate. A true
- first order approximation contains second order derivatives of the
- model function. For a derivation and discussion of the corresponding
- formula see
-
- A. Celmins, Least-Squares Optimization with Implicit Model Functions, in
- "Mathematical Programming with Data Perturbations II", Anthony V. Fiacco,
- Ed., Marcel Dekker, 1983, pp. 131-152.
-
- The covariance computation is discussed on pp. 136-138.
-
- Aivars Celmins
-
-