Partial Derivative of Incomplete Gamma Function

155 Views Asked by At

I'm insterested in calculating

$$ \frac{\partial}{\partial \alpha} \gamma \left(p, \left( \frac{x-\mu}{\sigma} \right)^2/\alpha^2 \right) $$

where $\gamma(p,u)=\int_{0}^{u}w^{p-1}e^{-w}dw$ is the incomplete Gamma function. Using this definition, you'd get

$$ \frac{\partial}{\partial \alpha} \int_{0}^{\left( \frac{x-\mu}{\sigma} \right)^2/\alpha^2} w^{p-1}e^{-w}dw. $$

I thought, well, this looks just like the Fundamental Theorem of Calculus, except for the slight complication that there's a partial derivative and the upper limit of integration isn't just the variable $\alpha$. Anyway, after thinking for a while, I thought that the answer might simply be

$$ \frac{\partial}{\partial \alpha} \int_{0}^{\left( \frac{x-\mu}{\sigma} \right)^2/\alpha^2} w^{p-1}e^{-w}dw = w^{p-1}e^{-w} \cdot \frac{\partial}{\partial \alpha} \left[\left( \frac{x-\mu}{\sigma} \right)^2/\alpha^2\right]. $$

Is this reasoning correct or am I missing something?

1

There are 1 best solutions below

2
On BEST ANSWER

You also need to evaluate $w$ at $\left(\frac{x-\mu}{\sigma}\right)^2\frac{1}{\alpha^2},$ then you will obtain the correct answer. Note that after integrating, and evaluating $w$ at the limits, there will be no $w$ terms left. This is the Leibniz integral rule.