**Definition A**: For any *x* > 0 the **gamma function **is defined by

(Note: actually the gamma function can be defined as above for any complex number with non-negative real part.)

**Definition B**: For any *x* > 0 the **lower** **incomplete gamma function** is defined by

For any *x* > 0 the **upper** **incomplete gamma function** is defined by

**Property A:**

Proof: Follows from Definitions A and B.

**Property B**:

Proof: Follows by integrating by parts.

**Property C**:

**Property 1**

- Γ(1) = 1
- Γ(
*x*+ 1) =*x*Γ(*x*) - Γ(
*n*) = (*n*– 1)! For all natural numbers*n*= 0, 1, 2, 3, … - Γ(½) = √π =

Proof:

(1) By Definition 1

(2) Follows from Property C

(3) Follows from (1) and (2) by induction

(4) The proof of the fourth assertion results from the fact (Gaussian integral) that

We won’t prove this here, but note that by using the substitution *x* = , we have by Definition 1

**Observation**: Note that the gamma function Γ(*x*) is only defined for *x* > 0. Negative values can be defined via Property 1.2, namely via Γ(*x*) = Γ(*x+*1)/*x*. Thus, by Property 1.4, we see that Γ(*–*.5) = Γ(-.5*+*1)/(-.5) *= –*2. This approach only works for non-integer values since Γ(0) = Γ(1)/0, Γ(-1) = Γ(0)/(-1), etc. are undefined.

The following formula can be used to calculate the gamma function for non-integer negative values.

**Real Statistics Functions**: The Real Statistic Resource Pack provides the following formulas.

**XGAMMA**(*x*) = gamma function at *x* even when *x* is negative

**LowerGamma**(*x, a*) = lower incomplete gamma function γ(*x, a*)

**UpperGamma**(*x, a*) = upper incomplete gamma function Γ(*x, a*)

Are the definitions for the upper and lower incomplete gamma functions reversed? Isn’t the lower the integral from 0 to a and the upper the integral from a to infinity?

Thanks Will,

Yes, the two definitions were reversed. I have now corrected this. Thanks for catching it.

Charles