1
$\begingroup$

We use the beta function $\text B(\alpha,\beta)=\int_0^1 t^{\alpha-1}(1-t)^{\beta-1}\text dt$ to prove the duplication formula for the gamma function:

$$\Gamma(2\alpha) = \frac {2^{2\alpha -1}\Gamma\left(\alpha+\frac 1 2\right)\Gamma(\alpha)}{\sqrt \pi}$$

Is there a similar formula for the duplication of the beta function? I.e., can I find $\text B(2\alpha,\beta)$ and/or $\text B(\alpha,2\beta)$ from $\text B(\alpha,\beta)$ and perhaps other functions? Can something like that be found for the incomplete beta function $\text B_x(\alpha,\beta)=\int_0^x t^{\alpha-1}(1-t)^{\beta-1}\text dt$?

And more specifically, can I find $\text B_x(2\alpha,\alpha)$ from $\text B_x(\alpha,\alpha)$?

$\endgroup$
4
$\begingroup$

Interesting question. We may start from the definition of the Beta function:

$$B(m, n) = \int_0^1 t^{m-1}(1-t)^{n-1}\ \text{d}t$$

and rewrite it when $m\to 2m$:

$$B(2m, n) = \int_0^1 t^{2m-1}(1-t)^{n-1}\ \text{d}t$$

$$B(2m, n) = \int_0^1 t^{2m-1} \frac{(1-t)^n}{1-t}\ \text{d}t$$

Now, since the range of integration is $[0, 1]$, we are allowed to make use of the geometric series

$$\frac{1}{1-t} = \sum_{k = 0}^{+\infty} t^k$$

Hence

$$B(2m, n) = \int_0^1 t^{2m-1} (1-t)^n \sum_{k = 0}^{+\infty} t^k\ \text{d}t = \sum_{k = 0}^{+\infty} \int_0^1 t^{2m-1} (1-t)^n t^k\ \text{d}t$$

And easily write:

$$\sum_{k = 0}^{+\infty} \int_0^1 t^{2m-1+k} (1-t)^n\ \text{d}t$$

Calling now

$$2m-1+k = a ~~~~~~~~~~~ n = b$$

We notice that the integral is well known:

$$ \int_0^1 t^a (1-t)^b\ \text{d}t \equiv B(a+1, b+1)$$

Then we end up with the partial result (re-expanding $a$ and $b$):

$$B(2m, n) = \sum_{k = 0}^{+\infty} B(2m+k, n+1)$$

That series does exist and it does converge to a known result:

$$\sum_{k = 0}^{+\infty} B(2m+k, n+1) = \frac{\Gamma (n) \Gamma (2 m+n+1) B(2 m,n+1)}{\Gamma (n+1) \Gamma (2 m+n)}$$

What you end up with is a sort of recursive relation for the Beta function:

$$B(2m, n) = \frac{\Gamma (n) \Gamma (2 m+n+1) B(2 m,n+1)}{\Gamma (n+1) \Gamma (2 m+n)}$$

BUT

The above expression can be simplified!

$$\frac{\Gamma (n) \Gamma (2 m+n+1) B(2 m,n+1)}{\Gamma (n+1) \Gamma (2 m+n)} \equiv \frac{\Gamma (2 m) \Gamma (n)}{\Gamma (2 m+n)}$$

What we obtained is actually nothing than what we would have obtained by simply substituting at the beginning $m\to 2m$ in the Gamma function / Beta function definition.

$$B(2m, n) = \frac{\Gamma (2 m) \Gamma (n)}{\Gamma (2 m+n)}$$

This really suggest that such a particular duplication formula for the Beta function may not exist at all, since all you need is the Gamma function and ITS duplication formula, through which you can evaluate $\Gamma(2m)$.

Seems like that this is the only "duplication formula" for the beta function.

(Also, I found nothing on reviews or literature).

$\endgroup$
2
$\begingroup$

Note that $$B(a,b) = \frac{\Gamma(a)\Gamma(b)}{\Gamma(a+b)},$$ so $$B(2a,b) = \frac{\Gamma(2a)\Gamma(b)}{\Gamma(2a+b)} = 2^{2a-1} \frac{\Gamma(a)\Gamma(b)}{\Gamma(a+b)} \cdot \frac{\Gamma(a+1/2)\Gamma(a+b)}{\Gamma(1/2)\Gamma(2a+b)},$$ or $$B(2a,b) = 2^{2a-1} \frac{B(a,b) B(a,a+b)}{B(a,1/2)}.$$ This is about as good as I could get.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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