From help-octave-request at bevo dot che dot wisc dot edu Thu Jan 8 12:39:48 2004 Subject: Re: Gnuplot From: "Henry F. Mollet" To: Joe Koski CC: Octave_post Date: Thu, 08 Jan 2004 10:38:10 -0800 > The gnuplot manual looks like it was > glued together in the 80s, and it reads like a treatise on ancient graphics > systems (DEC Regis, Tektronix 401x, etc.) and operating systems(Vax VMS, > Amiga, etc.). Just one of the challenges with use of free software, I guess. > I'm just happy things work as well as they do. > > Joe Koski Gnuplot manual is very terse indeed. However, it may be beneficial in some cases to read it and even use gnuplot instead of octave? It appears that gnuplot has an "easy" to use fitting procedure for fitting data to a non-linear fitting function f(x). It uses the Marquardt-Levenberg algorithm to minimize the least squares (="loss-function". I don't know if there is a way to use my own loss-function (e.g. max likelihood) and I also don't know how to create the loss-function as a function of the parameters. If I had the latter, then I could splot the loss function in parameter space to get confidence levels instead of having to use asymptotic standard errors which are said to be useful for only qualitative purposes. I have searched Octave for the equivalent or better but no luck so far. The optimization chapter (24) in the manual has 24.1 Linear Programming (empty) 24.2 Quadratic Programming (empty) 24.3 Nonlinear Programming (empty) 24.4 Linear Least squared (providing functions gls and ols). Am I looking in the correct place? Henry ------------------------------------------------------------- Octave is freely available under the terms of the GNU GPL. Octave's home on the web: http://www.octave.org How to fund new projects: http://www.octave.org/funding.html Subscription information: http://www.octave.org/archive.html -------------------------------------------------------------