Problem with curve fit

Share icon

Fri, 05/11/2012 - 04:49 pm

I would like to reproduce this result, i e have the Covariance matrix parameters printed after completion of curvefit

//¥CurveFit/M=2 Power Temperature[3,] /X=T_Us /W=Temperature_Err /I=1 /D /R /F={0.683000, 1}
// Fit converged properly
// Curve fit with data subrange:
// Row: 3, 30
// fit_Temperature= W_coef[0]+W_coef[1]*x^W_coef[2]
// Res_Temperature= Temperature - (W_coef[0]+W_coef[1]*T_Us^W_coef[2])
// W_coef={1421.9,4.3185,2.9768}
// V_chisq= 31.0292; V_npnts= 28; V_numNaNs= 0; V_numINFs= 0;
// V_startRow= 3; V_endRow= 30; V_startCol= 0; V_endCol= 0;
// W_sigma={394,0.909,0.0656}
// Coefficient values ± one standard deviation
// y0 = 1421.9 ± 394
// A = 4.3185 ± 0.909
// pow = 2.9768 ± 0.0656
// M_Covar[0][0]= {155471,-340.136,24.3494}
// M_Covar[0][1]= {-340.136,0.825423,-0.059566}
// M_Covar[0][2]= {24.3494,-0.059566,0.00430317}

however
even using /M=2 i do not the M_Covar results, even though a can edit M_Covar. Is there a way to specify we want them print?

I tried
CurveFit/M=2 HillEquation FS_Refl (,35)/X=Us (,35)/W=FS_Refl_Err(,35) /I=1 /D
but i got

Fit converged properly
Curve fit with data subrange:
FS_Refl[0,35]
fit_FS_Refl= W_coef[0]+(W_coef[1]-W_coef[0])*(x^W_coef[2]/(x^W_coef[2]+W_coef[3]^W_coef[2]))
W_coef={0.0064212,0.34443,8.033,13.789}
V_chisq= 1.2633;V_npnts= 36;V_numNaNs= 0;V_numINFs= 0;
V_startRow= 0;V_endRow= 35;
W_sigma={0.00379,0.0131,1.12,0.324}
Coefficient values ± one standard deviation
base =0.0064212 ± 0.00379
max =0.34443 ± 0.0131
rate =8.033 ± 1.12
xhalf =13.789 ± 0.324

Many thanks
Quote:
even using /M=2 i do not the M_Covar results, even though a can edit M_Covar. Is there a way to specify we want them print?


I'm not the curve fitting expert, but . . .

CurveFit will not print the covariance matrix, but you can print it like this:
Print M_Covar


Or you can view it in a table like this:
Edit M_Covar

May 11, 2012 at 06:02 pm - Permalink