Physics307L:Schedule/Week 13 agenda/Weighted/Derivation

From OpenWetWare
Jump to navigationJump to search

See Taylor 2nd edition page 174 or Bevington 2nd edition page 58

Step 1: Probability of each measured mean, given parent distributions with same true value

(I.e., assume both data sets have the same "true" value, but with differing standard deviations)

Assume Guassian distributions

Probability is

[math]\displaystyle{ Prob(x_A) \propto \frac{1}{\sigma_A}e^{- \frac{(x_A-X)^2}{2 \sigma_A^2}} }[/math]

Step 2: What is joint probability of getting both means?

Simplify with chi-squared short-hand

[math]\displaystyle{ Prob(x_A, x_B) \propto \frac{1}{\sigma_A \sigma_B}e^{-\chi^2 / 2} }[/math]
[math]\displaystyle{ \chi^2 = \left( \frac{x_A-X}{\sigma_A}\right)^2 + \left( \frac{x_B-X}{\sigma_B}\right)^2 }[/math]

Step 3: Principle of maximum likelihood: minimize chi-squared with respect to X

Step 4: Solve for X

obtain result on previous page.

Step 5: Error propagation to obtain new sigma