All Forums  Origin Forum  Origin Forum  user-defined fitting function issue, dependency

Note: You must be registered in order to post a reply.
To register, click here. Registration is FREE!

Screensize: 640 x 480 800 x 600 1024 x 768 1280 x 1024 UserName: Password: Anti-Spam Code: Format Mode: Basic  Help  Prompt  Format: Font Andale Mono Arial Arial Black Book Antiqua Century Gothic Comic Sans MS Courier New Georgia Impact Lucida Console Script MT Bold Stencil Tahoma Times New Roman Trebuchet MS Verdana   Size 1 2 3 4 5 6   Color Black Red Yellow Pink Green Orange Purple Blue Beige Brown Teal Navy Maroon LimeGreen Message: * HTML is OFF * Forum Code is ON Smilies [quote][i]Originally posted by jrmontag[/i] [br]Origin Ver. and Service Release (Select Help-->About Origin): 8.1 SR3 Operating System: Win XP I have a user-defined non-linear fit that has worked in the past but will no longer fit data due to parameter dependencies. The Function Form is Origin C, and it's fitting a single column of data (w/ an associated weighting curve) representing a Lorentzian peak. The functional form has not changed since it was used , yet the fitting routine insists there are unity dependencies between the variable "amp" and "f0" (and occasionally other variables) and cannot reduce chi squared (though I can by manually adjusting paratmeters). I'm struggling to discover why the LevMar fitting routine won't work. Anyone had similar problems or have any suggestions? Thanks, -Josh The code for the fit is as follows: void _nlsfnewPSDnw( // Fit Parameter(s): double f0, double width, double amp, double theta, double phaseOff, double uncorrOff, double slope1, double slope2, // Independent Variable(s): double x, // Dependent Variable(s): double& fit) { double inPhase,outofPhase,modX,modY; inPhase = (x^2-f0^2)/((x^2-f0^2)^2+x^2*width^2); outofPhase = x*width/((x^2-f0^2)^2+x^2*width^2); modX = inPhase*cos(theta) - outofPhase*sin(theta) + slope1*(x-f0); modY = inPhase*sin(theta) + outofPhase*cos(theta) + slope2*(x-f0) + phaseOff; fit = amp*(modX^2 + modY^2) + uncorrOff; } [/quote]   Check here to include your profile signature. Check here to subscribe to this topic.