Normalizing a probability density function

645 Views Asked by At

I need to find a normalization term $N(\alpha,\beta)$ for the probability density function:

$$PDF(\alpha,\beta)=(x-x_1)^{\alpha}e^{-\beta(x-x_1)}$$

In other words, solve the following equation:

$$N(\alpha,\beta)=\int_{x_1}^{x_2}(x-x_1)^{\alpha}e^{-\beta(x-x_1)}\mathrm{d}x$$

1

There are 1 best solutions below

0
On BEST ANSWER

I'm assuming you want some kind of closed form for $N(\alpha,\beta).$

For that, set $\beta(x-x_1)=t.$

Then, $\beta \,\mathrm{d}x=\mathrm{d}t.$

It follows that $$N(\alpha,\beta)=\beta^{-\alpha-1} \int_0^{\beta(x_2-x_1)} t^{\alpha} e^{-t}\, \mathrm{d}t$$

It is known that $$\gamma(s,x)= \int_0^{x} t^{s-1} e^{-t}\, \mathrm{d}t$$

where $\gamma(s,x)$ is the lower incomplete gamma function.

Thus, $$N(\alpha,\beta)=\beta^{-\alpha-1} \, \, \gamma\left[\alpha+1, \beta(x_2-x_1) \right] $$

and we are done.