6

How many different proofs are there that $a^n-b^n =(a-b)\sum_{i=0}^{n-1} a^i b^{n-1-i} $ for positive integer $n$ and real $a, b$?

You can use any techniques you want. My proof just uses algebra, summation, and induction, but if you want to use invariant sheaves over covalent topologies, that is fine.

I decided that I would try to produce a proof by induction. I find it interesting that my proof shows that if it is true for $n-1$, then it is true for $n+1$. This means that two base cases have to be proven: $n=1$ and $n=2$. Fortunately, those are easy.

I am sure that my proof is known, but I do not recall having seen it before.

Here is the induction step:

$\begin{array}\\ a^{n+1}-b^{n+1} &=a^{n+1}-a^nb+a^nb-b^{n+1}\\ &=a^{n+1}-a^nb +a^nb-ab^n +ab^n-b^{n+1}\\ &=(a-b)a^n +ab(a^{n-1}-b^{n-1}) +(a-b)b^n\\ &=(a-b)(a^n+b^n) +ab(a^{n-1}-b^{n-1}) \\ &=(a-b)(a^n+b^n) +ab((a-b)\sum_{i=0}^{n-2} a^i b^{n-2-i}) \ \ \text{(The induction hypothesis)} \\ &=(a-b)(a^n+b^n+ab\sum_{i=0}^{n-2} a^i b^{n-2-i}) \\ &=(a-b)(a^n+b^n+\sum_{i=0}^{n-2} a^{i+1} b^{n-1-i}) \\ &=(a-b)(a^n+b^n+\sum_{i=1}^{n-1} a^{i} b^{n-i}) \\ &=(a-b)\sum_{i=0}^{n} a^{i} b^{n-i} \\ \end{array} $

Bill Dubuque
  • 272,048
marty cohen
  • 107,799
  • 1
    Just expand the right hand side and cancel everything in sight. Induction is overkill. (Also, in general induction will only work if you already know one side of some equation. What happens when you really want to find some expression?) – Martin Brandenburg Aug 17 '14 at 06:05
  • 2
    There's no reason to force $a,b$ to be real; might as well make them indeterminates in the polynomial ring $\Bbb Z[x,y]$. Then you can just tensor and evaluate as needed. – anon Aug 17 '14 at 06:07
  • Alternative proof: Take the usual geometric series $a^n-1=(a-1) \sum_{i<n} a^i$, now homogenize. – Martin Brandenburg Aug 17 '14 at 06:08
  • I certainly know about the usual multiply and cancel proof. I just wanted to see if I could prove it by induction. Also, I wonder if there is a proof by induction that goes from n to n+1, rather than n-1 to n+1. – marty cohen Aug 17 '14 at 16:37
  • @Martin The (inductive!) cancellation is a prototypical application of telescopy - see my answer. – Bill Dubuque Aug 17 '14 at 17:58
  • Here for proofs using combinatorial arguments – Petite Etincelle Aug 18 '14 at 22:03

3 Answers3

3

By telescopy $\ f_n = x^n\,\Rightarrow \displaystyle \overbrace{f_n-f_0 =\sum_{k\,=\,0}^{n-1}\left[f_{k+1}\:\!-\:\!f_k\right]}^{\textstyle\ \ x^{n} - 1 \:\!=\:\! \sum\ {\overbrace{[x^{k+1}-{x^k}^{\phantom |}\!}^{\Large \color{#0a0}{(x\,-\,1)}\,\color{#c00}{x^k}\!\!}]}_{\phantom{|_|}}} =\, \color{#0a0}{(x\!-\!1)}\sum_{k\,=\,0}^{n-1}\, \color{#c00}{x^k} $

The sought result (Factor Theorem) follows by homogenization, i.e. $\, x\to a/b\,$ then scale by $\,b^n.$

Remark $\ $ The simple theorem employed to evaluate the above telescopic sum may be viewed as a discrete analog of the Fundamental Theorem of Integral Calculus

$$\begin{eqnarray} f &=& \sum \Delta f,\quad \Delta f(n) = f(n+1) -f(n)\\ f & =& \ \int D f,\quad D\, f(x) = f'(x)\end{eqnarray}$$

However, the proof of the discrete analog (the Fundamental Theorem of Difference Calculus) is just a trivial one-line induction exploiting telescopic cancellation.

Bill Dubuque
  • 272,048
2

Using $$\sum_{k=0}^{n-1}x^k={1-x^n\over1-x}$$ we get \begin{align} a^n-b^n &=a^n\left[1-\left({b \over a}\right)^n\right]\\ &=a^n\left[1-{b \over a}\right]\sum^{n-1}_{k=0}{b^k\over a^k}\\ &=\sum^{n-1}_{k=0}a^{n-k}b^k-\sum^{n-1}_{k=0}a^{n-k-1}b^{k+1}\\ &=(a-b)\sum^{n-1}_{k=0}a^{n-k-1}b^k \end{align}

SuperAbound
  • 5,554
0

$P(x)=x^n-1$ has the root $x=1$ so it factors as $P(x) = (x-1)Q(x)$. Differentiating the product $k \lt n$ times with the Leibniz rule and using that $(x-1)^{(j)} = 0$ for $j \ge 2\,$:

$$ n(n-1)\dots(n-k+1)\,x^{n-k} = (x-1)Q^{(k)}(x) + \binom{k}{1} Q^{(k-1)}(x) $$

Setting $\,x=0\,$ gives $\,Q^{k}(0)=k\,Q^{(k-1)}(0)\,$, so by induction $\,Q^{(k)}(0) = k!\,$, then by Taylor's theorem:

$$ Q(x) = \sum_{k=0}^{n-1} \frac{Q^{(k)}(0)}{k!}\,x^k = \sum_{k=0}^{n-1} \,x^k \quad\iff\quad x^n-1 = (x-1)\left(1 + x + \dots+x^{n-1}\right) $$

Homogenizing the latter equality produces the title identity.

dxiv
  • 76,497