Trouble with Calculating Gamma Function Integral Convergence on Wolfram Alpha?

[Mandelbroth]

This is mostly calculus, but the question is computer based, I think.

The antiderivative of the gamma function is, fairly trivially, $\displaystyle \int_{0}^{+\infty}\frac{t^{z-1}}{e^{t}\ln{t}}-C$, where C is an arbitrary constant.

Why does Wolfram Alpha have trouble calculating the convergence of that integral at any given point?

 

[mathman]

Definite integrals do not have a constant of integration. What values of z are you having trouble with? The integrand blows up at t = 0 for Re(z) <1 and the integral diverges for z = 0, -1, -2, etc.

 

[Mandelbroth]

Definite integrals do not have a constant of integration. What values of z are you having trouble with? The integrand blows up at t = 0 for Re(z) <1 and the integral diverges for z = 0, -1, -2, etc.

That's not what I'm saying. :tongue:

I'm saying that $\displaystyle \int \Gamma(z) dz = \int\int_{0}^{+\infty}\frac{t^{z-1}}{e^{t}}\, dt \ dz = \int_{0}^{+\infty}\frac{t^{z-1}}{e^{t}\ln{t}}-C$. I'm evaluating the antiderivative, or "indefinite" integral, of the Gamma function. It's fairly evident that I used Fubini's theorem, because it's the only sensible way to obtain that result.

I'm asking why the antiderivative of the Gamma function can't be evaluated by Wolfram Alpha.

 

[mathman]

I now understand what you did. However I have never worked with Wolfram Alpha, so I can't help you there.