Non linear curve fitting by changing parameters

Question

Araz 2012-11-19

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/54145-non-linear-curve-fitting-by-changing-parameters

Hi

Im trying to fitt two non linear curves to each other the red curve to the blue . I have tried the basic fitting method but it doesn't give me a good match.

In the equation h=... of my code by changing some parameters which is 2.41, 0.389, 0.3, 0.406, 0.06, 0.406 i should get the red curve to be fitted on blue curve.

The equation h in my code is :

h=sqrt(1+(2.41./(1-(0.389e-6./lambda).^2))+(0.3./((1-(0.406e-6./lambda).^2)+((1i.*0.06).*(0.406e-6./lambda))))); I have changed much the parameters but the curves doesn't fit.

What can i do to fit this two curves to each other by changing only thus six parameters?

Best Regards

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Star Strider 2012-11-19

编辑：Star Strider 2012-11-19

The value of h is complex. Do you want to fit the real(h) and imag(h) components separately, or do you want to fit abs(h)?

I also see only five parameters: [2.41; 0.389E-006; 0.3; 0.406E-006; 0.06].

This post ‘fitting of two curves and 6 unknown parameters’ also seems to quote the same equation. How to R and w fit? Should they be in your equation also?

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Michael 2012-11-19

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/54145-non-linear-curve-fitting-by-changing-parameters#answer_65824

What is the blue curve? Usually a curve fitter algorithm requires a model (your h function) with specified parameter space (your six numbers), but also data to be fitted. Is the blue curve a set of (h,lambda) points?

In this format I'd suggest trying lsqcurvefit, you can specify your model h(lambda) in the argument, along with a starting point and your data.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Araz 2012-11-19

blue curve is transmission vs lambda. The red curve in my code is also transmission vs lambda. but the curve differ from each other. is it possible to send the code to u email? u can have a look!

请先登录，再进行评论。