I know that:

$\Gamma(n) \quad=\quad (n-1)!\quad\forall n\in\mathbb N$

And that:

$\Gamma(t) \quad=\quad \int_{0}^\infty {x^{t-1}\over e^x}dx,\quad\Re(t)>0$

During some further research, (please correct me if I'm wrong) I learned that the Gamma Function can be defined for a negative value in such manner:

$\Gamma(s)\quad=\quad \Gamma(s+1)/s\quad\forall s\in\mathbb R^*$

Therefore (e.g.):

$\Gamma(-{1\over 2})\quad= \quad\Gamma({1\over 2})/(-{1\over 2})\quad=\quad -2\sqrt\pi$

Does this method also work for complex values of s with Real part less than $0$?

Or is there another method?


$\Gamma(z)$ is not going to be defined for all $z$ in $\mathbb{R}^*$. We can show this by doing the following manipulations, and in the process we can also show what values it will be defined for.

For your first question:

We start with the integral definition of the gamma function:

$$\begin{align*} \Gamma(z)&=\int_0^{\infty}e^{-t}t^{z-1}dt \\ &= \int_0^{1}e^{-t}t^{z-1}dt+\int_1^{\infty}e^{-t}t^{z-1}dt\\ &= \int_0^{1}\sum_{n=0}^{\infty}\frac{(-t)^n}{n!} t^{z-1}dt+\int_1^{\infty}e^{-t}t^{z-1}dt\\ &= \sum_{n=0}^{\infty}\int_0^{1}\frac{(-t)^n}{n!} t^{z-1}dt+\int_1^{\infty}e^{-t}t^{z-1}dt\\ &= \sum_{n=0}^{\infty}\frac{(-1)^n}{n!(z+n)}+\int_1^{\infty}e^{-t}t^{z-1}dt\;\;\;\;\;\;(*) \end{align*}$$

Note that the integral in line $(*)$ is defined $\forall z\in\mathbb{C}$, but the series runs over $n\in\left \{ \;0,\;1,\;2,\;3,\;\cdots\right \}$ and contains $(z+n)$ in the denominator. From this we can see that $\Gamma(z)$ is defined $\forall z\in\mathbb{C}\setminus\left \{ \;0,\;-1,\;-2,\;-3,\;\cdots\right \}$.

So we conclude that the statement $\quad\forall s\in\mathbb {R}^*,\;\;\Gamma(s)= \Gamma(s+1)/s$ is false, as $\Gamma(s)$ is undefined for some $s$ in $\mathbb {R}^*$.

For your second question:

The relation


will hold $\forall z\in\mathbb{C}\setminus\left \{ \;0,\;-1,\;-2,\;-3,\;\cdots\right \}$. We see this from the above section.

I am also assuming you are looking for a general way to create closed form representations of $\Gamma(z)$ for $z=iy-x$ where $x\in\mathbb{R}^+\setminus\mathbb{N}$ and $y\in\mathbb{R}$.

As far as I am aware there is no way to further simplify the following any more than shown:

$$\begin{align*} \Gamma(iy-x)&=\frac{\Gamma(iy-x+1)}{iy-x} \\ &= \frac{\Gamma(iy-x+2)}{(iy-x)(iy-x+1)}\\ &= \frac{\Gamma(iy-\left \{ x\right \})}{(iy-x)(iy-x+1)\cdots(iy-\left \{ x \right \}-1)} \end{align*}$$

as there is no other 'nicer' way to represent $\Gamma(iy-\left \{ x\right \})$ where $\left \{ x\right \}$ is the fractional part of $x$. One could also argue that what is shown directly above isn't much of a simplification, but this does show that the relation holds for $z$ such that $\Re(z)<0$.

  • $\begingroup$ Thank you so much. This helped me more than you can imagine. $\endgroup$ – Max Echendu Nov 18 '15 at 0:11
  • $\begingroup$ You are very much welcome. $\endgroup$ – dxdydz Nov 18 '15 at 1:29

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.