# B Proof that (x^n)/n! has a limit of 0 at infinity

Tags:
1. Jul 8, 2017

### Derek Hart

I understand that the standard proof is a bit different from my own, but I want to know if my reasoning is valid. PROOF:
Firstly, I assume that x is positive.

I then consider p = inf{n∈ℕ : n>x} . In other words, I choose "p" to be the smallest natural number greater than x. If we choose n>p, then we can rewrite the original statement as ((x/1)(x/2)...(x/(p-1)))*((x/p)...(x/n)). If we let α=((x/1)(x/2)...(x/(p-1))), then xn/n! = α((x/p)...(x/n)) where each term x/k is less than one. Thus α((x/p)...(x/n)) < α(x/n). So if we want to require that (xn/n!) < ε for some ε>0, then we can choose n satisfying α(x/n) < ε ⇒ n > (αx/ε). Since α = (xp-1/(p-1)!), we must have n > (xp/ε(p-1)!). Thus, any n satisfying n > max[p, (xp/ε(p-1)!)] will ensure that xn/n! < ε.

Is this subsequent method for finding sufficiently large "n" viable in practice, and are my findings valid?

2. Jul 8, 2017

### Staff: Mentor

That works, sure.

This could be misinterpreted. What exactly is part of the denominator?
Better write it as xp/(ε(p-1)!) to avoid ambiguity.
Or, even better, use TeX: $\displaystyle n > \frac{x^p}{\epsilon (p-1)!}$.

3. Jul 8, 2017

### Staff: Mentor

There's a function that does this, the least integer function, also known as the ceiling function. See http://www.mathwords.com/c/ceiling_function.htm. The notation is $\lceil x \rceil$. Unrendered, this looks like \lceil x \rceil.

For example, $\lceil 4.52 \rceil = 5$.

There's a related function, the greatest integer function (or floor function), denoted as $\lfloor x \rfloor$.

4. Jul 8, 2017

### Derek Hart

Yeah sorry I meant for both epsilon and the factorial to be in the denominator. Didn't even notice my mistake at first

Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?
Draft saved Draft deleted