Change of Variables
For strictly monotone with differentiable inverse, has density .
Proof of the monotone case
Suppose first that is strictly increasing. Then
and differentiating gives the density , which is non-negative because .
If is strictly decreasing, then
and differentiating gives the density , non-negative because . Both cases combine into .
Jacobian formula for several random variables
Theorem 4.28
Let be a random variable with values in and density . Let be a bijection with a continuous derivative on , and
Then, the random variable has density
where is the Jacobian determinant of .
Example: polar coordinates of two standard normals
Let be independent and set , . The map has Jacobian determinant
so the density of is
for and . The joint density factorises into a function of times a function of , so and has density for , independently of each other.
Related
Stated in
- Theorem 4.16§4.5.2 Linear Transformations of Normal Distributions
- Theorem 4.28§4.6.5 Transformation of Random Variables
