what is (1/2)!
What is the factorial of 1/2?
If your answer is not the gamma function, then I’ll ask you again after you have read this article.
Your answer might have changed…
Introduction
In the late 1720s, Leonhard Euler was thinking about how to extend the factorial to non-integer values.
This was the start of a rich theory used all over the scientific world. A theory of one of the most important functions in mathematics.
Leonhard Euler is, without doubt, one of the greatest mathematicians in history. To give you an idea of Euler’s powers here follows some examples that show his brilliance.
First of all, Euler had an outstanding memory! He was able to recite Virgil’s Aeneid from beginning to end, detailing in what line every page of the edition he owned began and ended.
To give you some context, the Aeneid comprises a total of 9,896 lines!
Euler was also extremely productive. He produced about 30,000 pages in his lifetime and it is estimated that he accounted for about a third of all published scientific papers in the 18’th century!!!
He even published papers after he died!
Many of those pages were written while he was blind, and for that reason, Euler has been called the Beethoven of mathematics.
Beethoven could not hear his music. Likewise, Euler could not see his calculations.
Actually, Euler was quite optimistic about the loss of his vision. He is known to have said something like:
“In this way I will have fewer distractions”.
One should think that this would slow him down, but in fact, when he became blind, his production went up!
Euler also had phenomenal computational powers. On one occasion, two students disagreed over the result of the sum of 17 terms in a series because their results differed in the fifth decimal place.
Euler computed the correct result in his mind in a few seconds.
This anecdote was referred to by his colleague Nicolas de Condorcet who at Euler’s death wrote a lengthy eulogy in which he declares that Euler is
“one of the greatest and most extraordinary men that nature has ever produced".
So Euler was a great mathematician, to say the least, and he was thinking about how to extend the factorial function.
I will show you what he came up with and the surprising properties that followed. Later in the article, I’ll reveal how we would give meaning to 1/2! and what the value of this symbol is.
The Factorial
Before moving on with the story, let’s pause for a bit and recall what the factorial is.
It is simply a product of the first n natural numbers.
n! = n ⋅ (n - 1) ⋅ (n - 2)⋅⋅⋅ 3 ⋅ 2 ⋅ 1
For example:
5! = 5 ⋅ 4 ⋅ 3 ⋅ 2 ⋅ 1 = 120.
One of the reasons that the factorial is important in mathematics is that it represents the number of ways we can arrange things.
Imagine that you have 12 books on a book shelf. In how many ways do you think you can arrange those?
The answer to this question is 12! which is about 479 million ways of arranging the 12 books.
As you can see from this example, the factorial function grows very fast! In fact, it grows super-exponential. That is, it grows faster than exponentially.
The Gamma Function
When Euler was thinking about a problem, it usually ended up getting solved. However, what truly made him great, was the way he solved problems.
As we will see soon, it was usually a very creative line of thought and some alien ideas that turned out to be very clever.
In 1738, Euler came up with a generalization of the factorial in the form of a function defined by a certain integral, namely
Where log is the natural logarithm (sometimes denoted ln).
By the substitution s = exp(-t), where exp is the exponential function with base e (a number also discovered by Euler), we get
Thus we arrive at the amazing fact that
which is equivalent to Euler’s definition. To provethat this integral is in fact the factorial, let’s call the integral on the right-hand side Π(n) and let’s do some partial integration:
That is a nice functional equation that makes us capable of proving the formula by induction.
We want to prove that Π(n) = n! for all natural numbers n.
First of all, note that
That is, Π(1) = 1 = 1!.
Next, assume that Π(n - 1) = (n - 1)!. Then we have
Π(n) = n Π(n - 1) = n (n - 1)! = n!
where we have used the functional equation above.
By induction, the proof is done.
Note that in the definition of Π(n) above, n doesn’t actually have to be a natural number. The expression makes sense for all complex numbers with non-negative real part.
The modern way of working with these generalized factorials is through the gamma function. The gamma function is very similar to the function that we called Π and it is defined by the following.
Note that Γ(n) = Π(n - 1) = (n - 1) ! for all natural numbers n.
Thus, the gamma function also satisfies a similar functional equation i.e.
Γ(z+1) = z Γ(z).
So the gamma function is a generalized factorial function in the sense that Γ(n+1) = n! for all non-negative whole numbers n.
But is it a unique generalization?
Unfortunately, the answer is no. However, if we give it a certain constraint, then it turns out that it is.
The constraint has to do with a concept called logarithmic convexity, but I will not write about it in detail here, since it is a bit off-topic in terms of what I want to show you.
The specific requirement is that the function log Γ is convex.
A twice-differentiable function f is logarithmically convex if and only if
f´´(x) f(x) ≥ f´(x)².
The important thing is that the gamma function is in a specific mathematical sense the natural choice if you want to generalize the factorial.
The Weierstrass Product
Countless definitions and forms of the gamma function have been found. A particularly nice one is a certain infinite product.
Before getting there, let’s try to derive some interesting results from our definition.
The first thing we will do might look weird at first but sometimes in mathematics, you should just try stuff out and follow the logical consequences while using your intuition.
We will write the exponential function as a limit and plug it into our definition of the gamma function.
First, recall that
This can be proved in many ways.
The obvious way is to set it up so we can use L’Hôpital’s rule, we will however take another approach.
Recall that the geometric series has a closed-form:
which holds if |x| < 1.
Note that by substituting x by -x, we get:
Now we can do some further manipulations of the two sides.
Assume n > x, then we can substitute z = x/n.
Now, if we take the limit as n goes to infinity, it is pretty clear that
With this result in the toolbox, it is now a matter of straightforward calculations to derive the wanted result.
By a substitution, this is equivalent to the statement
Let us now use this result in the definition of Γ(z).
Let’s agree to call the integral inside the limit I(n, z).
Using partial integration a couple of times gives us:
This pattern continues, and when we finally get rid of the exponent of the 1 - t/n term, we can integrate to get
and to get Γ(z), we take the limit:
which is a nice result in its own right and is also a very famous result, but we don’t want to stop here.
Let’s proceed with some simple manipulations of this limit.
Here we have added and subtracted by ∑ z/i in the exponent of e. Note that log is again the natural logarithm.
We can now split the exponent and use the fact that a sum in the exponent is a product.
Recall that Euler’s constant is defined by
If we now take the limit in the expression above for the gamma function, we get a beautiful result known as the Weierstrass product of the gamma function.
This is a mathematical pearl! Look at that…
In some ways, this is a better representation of Γ, but we will get back to that a little later.
Euler’s Reflection Formula
One of the most beautiful relations in mathematics is due to Leonhard Euler. I am not talking about his famous identity this time, but rather the formula known as the reflection formula.
Euler discovered the following amazing result, linking the gamma function to the trigonometric functions.
The proof of this fact is as follows.
Recall the sine function’s infinite product also discovered by Euler!
If you want to know how Euler derived this product, you can take a look at the article: Infinity in Numbers linked at the bottom of this article.
Recall the Weierstrass product for Γ that can be written
Now it is pretty straightforward to calculate the following by comparing the products of Γ(z) and Γ(-z).
Now we can use the functional equation for the gamma function
Γ(1 - z) = - z Γ(-z)
to derive
It is clear that z cannot be an integer, because the denominator above would then be 0.
Applications of the Gamma Function
The gamma function pops up all over the place throughout mathematics.
From statistics, number theory, and complex analysis in mathematics, to string theory in physics. It seems to be a mathematical glue that ties different fields together and there’s a good reason why as we will see a little later.
One of the reasons it is important for number theory is that it has a special relationship with the Riemann zeta function.
Let’s take a look at the definition once again, but this time playing around with a substitution. Let n be a natural number. Then by the substitution t = nx, we get
and since this holds for all natural numbers n, we can sum on both sides to get
Thus we have arrived at a beautiful relation between the zeta function and the gamma function:
This however, is only valid for Re(s) > 1.
This is the first hint that the two functions share an intimate relationship.
A deeper and more interesting result, which I consider to be one of the most beautiful functional equations in the world, is the following, which I will state without proof:
Bernhard Riemann found this bad boy in 1859 and it gives a lot of knowledge of the zeta function via the gamma function.
For instance, one can clearly see the trivial zeros of ζ at the negative even integers. This is because, by analytically continuing Γ(s) to the whole complex plane, we see that it has poles at the non-positive integers. And since the gamma factor on the left-hand side blows up at the negative even integers but the right-hand side is finite, then ζ(s) has to be zero at those points.
In theoretical physics, the beta function that Euler (also) discovered, was used by Italian theoretical physicist Gabriele Veneziano in 1968 to describe strongly interacting mesons.
The Euler beta function can be defined by Β(x, y) = Γ(x)Γ(y) / Γ(x + y).
The reason for this is that it turns out to describe the first known scattering amplitude in string theory, and is in a sense a unique solution to this problem. It also has to do with the poles at the negative integers of Γ.
Another strikingly beautiful result concerns the growth of the gamma function. This is called Stirling’s formula and it states
This says that the order of growth of the two sides above is the same i.e. the limit of their ratio tends to 1 as z tends to infinity.
Euler’s Amazing Integral Formula
In the derivation of the integral formula for Γ(s) ζ(s) we summed on both sides and created some series.
Instead of doing that, Euler did something brilliant. He made a more general substitution and then his mind exploded with creativity, ending up with an amazing formula that holds all kinds of interesting stuff.
Let’s see how he did it and what these formulas are.
At Euler’s time, not much was known about complex analysis, but he had a fantastic intuition and since he knew that this relation holds when w is a positive real number, he considered what happens when w is a complex number with Re(w) > 0.
Let w ∈ ℂ with Re(w) > 0. Then by complex conjugating w on both sides of the equation above, we get
Now comes a brilliant thought!
Euler says let w = a + bi, let the argument of w be θ and let |w| = r such that w = r ⋅ exp(θi). Then we can write the above in an interesting way.
This is the super formula that contains so many beautiful relations as we will soon realize.
The final step is to write this out into its corresponding real and imaginary parts (using Euler’s world-famous identity) and consider both formulas hiding in the notation.
These formulas are indescribably beautiful!
Notice that they are a generalization of the gamma function, since if we let w=1, then we get the definition of the gamma function back from the cosine integral equation.
In the following section, we will use Euler’s integrals to solve the Dirichlet Integral.
A Generalization of the Dirichlet Integral
This is a fun problem.
Find the value:
This is a very famous problem and there are many ways to solve it. Most of them revolve around the Laplace transform, double integrals, or even the Feynman trick!
We will try to derive it from Euler’s beautiful formulas above. Actually, we will generalize this problem into a more general result that has this integral as a special case.
To do that, we will start by using Euler’s reflection formula to rewrite the left-hand side of the sine equation. However, before we do that recall from calculus that we can use L’Hôpital’s rule to show that
Let’s take the left-hand side of Euler’s sine integral formula and transform it a bit.
By our above calculation, we get
Since -π < θ < π, we have
Thus, by taking the limit on the right-hand side, we get
This is a quite nice formula.
Note that if we take the limit as a goes to 0 then the left-hand side will tend towards π/2 for all real b ≠ 0.
That is, the following holds:
The specific case w=i will solve the Dirichlet integral since then a=0 and b=1.
So the Dirichlet integral I = π/2.
Analytic Continuation of the Gamma Function
There is one quite important thing that we haven’t discussed yet.
Recall the definition of the gamma function:
One can prove that this integral only converges for Re(z) > 0.
However, in complex analysis holomorphic and meromorphic functions have a nice property that given a function f with domain D, if there is another meromorphic function g with a domain containing D as a subset, and if f = g on an open subset of D (if you don’t know what this is, then you can think of a small disk in the complex plane), then f = g on all of D. Moreover, g is the unique function extending f such that f = g on D.
That is, the function f can be extended to a larger domain in only one way. It simply has different representations.
So even though the definition above is fine when the real part of z is a positive real number, we need to remember that this is just one representation of the gamma function.
We look at the function from only one angle so to speak.
If we take a look at the Weierstrass product
one can show that this converges for all complex numbers z except for the non-positive integers. So this representation is in a sense a better one. A global one!
This also shows that the gamma function doesn’t have any zeros and it has poles at the negative integers and at 0.
There is another way of analytically continue the gamma function. Recall that
Γ(z+1) = z Γ(z)
Rearanging this, reveals
Γ(z) = Γ(z+1)/z
In the same way we get
Showing that we can make an analytic continuation displaying meromorphic representations of Γ where we also see the poles at the non-positive integers.
What is 1/2!?
Since Γ(n+1) = n! for all non-negative whole numbers n, we can give meaning to 1/2! by calculating Γ(1/2 + 1) = Γ(3/2).
But how do we do that?
Well, first note that by the functional equation Γ(z+1) = z Γ(z), we can simplify the problem a bit.
Γ(3/2) = 1/2 Γ(1/2).
Thus, it suffices to find Γ(1/2).
To do that, we use Euler’s reflection formula once again on the special case z = 1/2:
Therefore, we can now interpret
Let me ask you one more time.
What is your favorite function?
Comment down
Comments
Post a Comment