Nonlinear least squares
in [Matlab]
|
From: Hu Qing on 11 May 2010 19:04 Dear MATLAB User Community! I have a problem with nonlinear least squares fit. I have a series data (xi,yi) i = 1,2,3,...n plot (x, y) xlabel ('distance[m]') ylabel ('level [m]') legend ('profile') and I want to fit it to this equation: F1(x) = -A*(x+xs)^p Three parameters A, p, xs are defined by fitting values. how can i do this? Thank you very much!
From: Marcus M. Edvall on 11 May 2010 21:00 You can solve it with tomSym using the following code: n = 300; x = rand(n,1); y = rand(n,1); toms A xs p sol = ezsolve(y+A*(x+xs).^p); There are probably quite a bit of options. Best wishes, Marcus http://tomopt.com/ http://tomsym.com/
From: Alan Weiss on 12 May 2010 07:36 On 5/11/2010 7:04 PM, Hu Qing wrote: > Dear MATLAB User Community! > > I have a problem with nonlinear least squares fit. > I have a series data (xi,yi) i = 1,2,3,...n > plot (x, y) > xlabel ('distance[m]') > ylabel ('level [m]') > legend ('profile') > > and I want to fit it to this equation: > > F1(x) = -A*(x+xs)^p > Three parameters A, p, xs are defined by fitting values. > > how can i do this? > > Thank you very much! Take a look at lsqcurvefit. Example: http://www.mathworks.com/access/helpdesk/help/toolbox/optim/ug/brn4noo.html#brp3l6v-1 Function reference page: http://www.mathworks.com/access/helpdesk/help/toolbox/optim/ug/lsqcurvefit.html Alan Weiss MATLAB mathematical toolbox documentation
From: Frederic Moisy on 16 May 2010 08:09 "Hu Qing" <bluerivers(a)live.com> wrote in message <hscnp6$kt3$1(a)fred.mathworks.com>... > Dear MATLAB User Community! > > I have a problem with nonlinear least squares fit. > I have a series data (xi,yi) i = 1,2,3,...n > plot (x, y) > xlabel ('distance[m]') > ylabel ('level [m]') > legend ('profile') > > and I want to fit it to this equation: > > F1(x) = -A*(x+xs)^p > > Three parameters A, p, xs are defined by fitting values. > > how can i do this? > > Thank you very much! You can also have a look to the Ezyfit toolbox: http://www.mathworks.com/matlabcentral/fileexchange/10176-ezyfit-2-30 plot(x,y) showfit('F1(x) = -A*(x+xs)^p ');
From: Hu Qing on 29 May 2010 04:33
"Marcus M. Edvall" <edvall(a)gmail.com> wrote in message <2b5bdccc-99d5-4bc3-aae9-3378192252a6(a)37g2000yqm.googlegroups.com>... > You can solve it with tomSym using the following code: > > n = 300; > x = rand(n,1); > y = rand(n,1); > toms A xs p > sol = ezsolve(y+A*(x+xs).^p); > > There are probably quite a bit of options. > > Best wishes, Marcus > http://tomopt.com/ > http://tomsym.com/ Dear Marcus That's great! Thanks for help. Best wishes, HuQing |