MHB Marginal probability density functions (pdf)

AI Thread Summary
The discussion focuses on finding the marginal probability density functions (pdfs) for two independent gamma distributions, specifically for the transformations Y1 = X1/(X1+X2) and Y2 = X1+X2. The user has derived the joint pdf and applied the Jacobian but is struggling to express the marginal pdf for Y1. Additionally, they seek guidance on finding the marginal pdf for Y2 given the transformation Y1 = X1/X2 and Y2 = X2. The conversation emphasizes the utility of the Gamma function in computing the necessary integrals for these transformations. Understanding these concepts is crucial for successfully deriving the marginal pdfs.
QuebecRb
Messages
1
Reaction score
0
I have a set of two related queries relating to marginal pdfs:
i.How to proceed finding the marginal pdfs of two independent gamma distributions (X1 and X2) with parameters (α1,β) and (α2,β) respectively, given the transformation: Y1=X1/(X1+X2) and Y2=X1+X2.
I am using the following gamma formula:

View attachment 3407

Having written the joint pdf and having applied the Jacobean, I have reached the final stage of writing the expression for the marginal (Y1):

View attachment 3408

but I cannot proceed further, obtaining the marginal pdf.

ii.Additionally, given the following transformations , Y1=X1/X2 and Y2=X2, I have written the expression for the marginal (Y2):

View attachment 3409

How do I find this marginal pdf?

Any enlightening answers would be appreciated.
 

Attachments

  • 1.JPG
    1.JPG
    2.9 KB · Views: 107
  • 2.JPG
    2.JPG
    5.6 KB · Views: 113
  • 3.JPG
    3.JPG
    6.6 KB · Views: 128
Physics news on Phys.org
I'm a little bit confused about the exercice. If I understand it correctly, you have given two independent gamma distributed random variables $X_1 \sim \Gamma(\alpha_1, \beta) $ and $X_2 \sim \Gamma(\alpha_2,\beta) $ and the goal is to compute the distribution of $Y_1 = \frac{X_1}{X_1+X_2}$ and $Y_2 = X_1+X_2$. Am I right here?

I did not check if your solution for $f_{Y_1}(y_1)$ and $f_{Y_2}(y_2)$ is correct, but to proceed the Gamma function can be useful. It is defined as
$$\Gamma(z) = \int_{0}^{\infty} t^{z-1}e^{-t}dt$$

For example to compute the following integral
$$\int_{0}^{\infty} y_2^{\alpha_1+\alpha_2-1}e^{-\beta y_2(y_1+1)}dy_2 $$

Let $\beta y_2(y_1+1) = t \Rightarrow dy_2 = \frac{dt}{\beta(y_1+1)}$. Thus the integral becomes
$$\frac{1}{\beta (y_1+1)} \int_{0}^{\infty} \left[\frac{t}{\beta(y_1+1)}\right]^{\alpha_1+\alpha_2-1} e^{-t}dt = \frac{1}{\beta^{\alpha_1+\alpha_2}(y_1+1)^{\alpha_1+\alpha_2}} \int_{0}^{\infty} t^{\alpha_1+\alpha_2-1}e^{-t}dt = \frac{1}{\beta^{\alpha_1+\alpha_2}(y_1+1)^{\alpha_1+\alpha_2}} \Gamma(\alpha_1+\alpha_2)$$
 
Back
Top