# Note the excruciatingly painful difference between

• Lesson Plan
• 66

This preview shows pages 25–27. Sign up to view the full content.

Note the excruciatingly painful difference between the intuitive equation 8.3c and the formally correct equation 8.4: on the right-hand side of the arrows, the intuitive one depends on s 2 M (the variance of the datapoint residuals ) , as you’d think it should, while the formal one depends on σ 2 meas M (the adopted intrinsic measurement variances of the data), which are chosen by the guy doing the fit. If you do an unweighted average, and derive a certain variance, and next do a weighted average in which you choose some values for σ meas that happen to be wrong, the two fits give different results for s 2 w,mean . This is crazy. To get around this difficulty, we follow the procedure in BR equations 4.20 to 4.26. This introduces an arbitrary multiplicative factor for the weights and goes through the ML calculation to derive, instead of equation 8.4, the far superior s 2 w,mean,BR = hatwider χ 2 w meas,m s 2 w M , (8.5) which is precisely the same as our intuitive guess, equation 8.3c. The difference between the formal equation 8.5 and the intuitive equations 8.3b and 8.4 is the numerator, which contains the reduced chi-square hatwider χ 2 ; for the case where all σ meas,m are identical, hatwider χ 2 = s 2 w σ 2 meas . Note that χ 2 and hatwider χ 2 are defined in equations 8.1. 8.2. The multivariate chi-square fit Here we generalize § 8.1, which dealt with the weighted average, to the multivariate case. In this case, chi-square fitting is just like least-squares fitting except for the following: 1. In the least-squares matrix X of equation 2.1a, each row m is a different measurement with a different intrinsic variance σ m . For chi-square fitting you generate a new matrix X χ , which

This preview has intentionally blurred sections. Sign up to view the full version.

– 26 – is identical to X except that each row m (which contains a particular equation of condition) is divided by σ m . This new matrix is the same as NR’s design matrix (Figure 15.4.1), which they denote by A . 2. For chi-square fitting, divide each datapoint y m in equation 2.1b by σ m . You are generating a new data vector Y χ , which is identical to Y except that each datapoint is divided by σ m . This new data vector is the same as NR’s vector b . 3. Note that the above two steps can be accomplished matrixwise by defining the M × M diagonal matrix [ σ ] in which the diagonal elements are σ m . [ σ ] = σ 0 0 . . . 0 0 σ 1 . . . 0 . . . . . . . . . 0 0 0 0 σ M - 1 (8.6) in which case we can write X χ = [ σ ] - 1 · X (8.7a) Y χ = [ σ ] - 1 · Y . (8.7b) 4. Carry through the matrix calculations in equations 8.8 below (using the matrices subscripted with χ ). You’ve divided each row, i.e. the equation of condition for each row m , by a common factor, so the solution of that particular equation of condition is unchanged. However, in the grand scheme of things—i.e. the normal equations—it receives a greater or lesser weight by a factor 1 σ 2 m .
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

### What students are saying

• As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

Kiran Temple University Fox School of Business ‘17, Course Hero Intern

• I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

Dana University of Pennsylvania ‘17, Course Hero Intern

• The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

Jill Tulane University ‘16, Course Hero Intern