Of Notes
Moment Generating Function
Table of Contents
Instead of manually doing moments of a distribution we leverage a generating function which analytically defines for some moment
Taking a random variable
Expanding the Exponential
The exponential is the standard choice because its taylor series packs every power of
By linearity of the expected value:
so every raw moment already sits inside
Extracting Moments by Differentiating
Differentiating with respect to
which is the first moment. Differentiating again:
which is the second moment. Always evaluating at
whenever the MGF exists in a neighbourhood of
Why the Exponential
- Sums of independent random variables become products:
, hence when and are independent. - For continuous variables, the integral is a Laplace transform, and setting
gives a Fourier transform: the Characteristic Function.
We can extend this concept onto the next level: Cumulant Generating Functions