Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

In another question of mine at mse I had the detail of the two assumed identities $$ C_{b,p} = \int_0^\infty b^{x^p} dx $$ and $$ C_{b,p}={n! \over (- \beta)^n} $$
for some $ \small n \in \mathbb N $ and $ \small p= \frac 1n $ which I'd checked with some $0 \lt b \lt 1$, $ \small \beta=\ln(b)$ .

I played a bit with my formulae and got aware, that if we rewrite the second identity $$ C_{b,p}={ \Gamma(1+n) \over (- \beta)^n} $$

it looks as if we can extend this even to noninteger $n$ and we'll actually get by this an integral formula for the $\Gamma$ in the form of: $$ \Gamma(n) = {(-\ln(b))^n \over n }\int_0^\infty \exp \left( \ln (b) x^{1/n} \right) dx \tag 1 $$

I've looked at wikipedia and HOMF (E.Weissstein seems to be inaccessible???) but could not find a similar formula which would allow me to check more whereabouts (range of convergences in ranges of the parameters etc.), so I ask here, whether

  1. this formula is basically correct/applicable for convergent cases

  2. and what the allowed ranges for $b$ and $n$ are

  3. and (since I'm still terribly weak in the manipulation with integrals)
    how that formula (1) could be related to the standard integral definition of the $\Gamma$ if this is possible...

[update:] Example Pari/GP-code gives equality up to machine precision (200 dec digits selected)

[b=0.5,p=0.5+Pi/1000]    \\ set parameters for convergent cases
                         \\ here n =1/p shall be noninteger
 %1516 = [0.500000000000, 0.503141592654]

[w=log(b),    n=1/p ]
      %1517 = [-0.693147180560, 1.98751209322]

v1  = (-w)^n/n * intnum(x=0,9999999, b^x^p)  \\ int_0^oo 
 %1518 = 0.994784377172

v2 =  gamma(n)
 %1519 = 0.994784377172

 %1520 = -2.077003953 E-200
share|cite|improve this question
up vote 3 down vote accepted

This is most easily handled by setting $a=-\ln b$. Assuming $n>0$, you must have $a>0$ (i.e., $b<1$), else the integral in (1) won't converge. You can now substitute $x=(u/a)^n$ in the integral, to get $$\int_0^\infty \exp(-ax^{1/n})\,dx=\frac{n}{a^n}\int_0^\infty u^{n-1}e^{-u}\,du=\frac{n}{a^n}\Gamma(n).$$ This should work for any $n>0$ and $a>0$. The integral in (1) will also diverge if $n<0$, since then the integrand goes to $1$ as $x\to\infty$.

share|cite|improve this answer
@Did Oh, duh. Yeah, I should have spotted that, but I got interrupted and had to run just before posting. Thanks for pointing it out, I have edited the answer accordingly. – Harald Hanche-Olsen Jan 20 '13 at 19:54

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.