1

Please tell if the problem can be solved using telescoping technique or not.

If yes, how to prove $a^n − b^n = (a − b) \sum_{i=1}^{n}a^{n-i} b^{i-1}\le (a − b)na^{n−1}$ using that. It is given that $a,b \in \mathbb{R}{+},\, a\gt b,\, n \in \mathbb{N}.$

I tried as follows, but was unsuccessful to pursue:

$a^n − b^n = a^n+\sum_{i=1}^{n-1}(a^ib^{n-i}-a^ib^{n-i})-b^n=a^n+\sum_{i=1}^{n-1}a^ib^{n-i}-\sum_{i=1}^{n-1}a^ib^{n-i}-b^n$


Edit : based on the selected answer's comment.

Writing a few terms of the series, $\sum_{i=1}^n (a^{n+1-i}b^{i-1}-a^{n-i}b^i)$ get:

For $n =5$, get the terms as:
$i=1, \,\, a^{5+1-1}b^{1-1}-a^{5-1}b^1 = a^5-a^4b.$
$i=2, \,\, a^{5-1}b^{2-1}-a^{5-2}b^2 = a^4b-a^3b^2.$
$i=3, \,\, a^{5-2}b^{3-1}-a^{5-3}b^3 = a^3b^2-a^2b^3.$
$i=4, \,\, a^{5-3}b^{4-1}-a^{5-4}b^4 = a^2b^3-a^1b^4.$
$i=5, \,\, a^{5-4}b^{3-1}-a^{5-3}b^5 = a^1b^4-b^5.$

Adding all the terms, get:

$a^5-a^4b+ a^4b-a^3b^2+a^3b^2-a^2b^3+a^2b^3-a^1b^4+a^1b^4-b^5 = a^5 - b^5$

jiten
  • 4,524
  • 1
    Can you simply assume $a>b$? – rtybase Jun 16 '19 at 12:22
  • 1
    @rtybase Thanks for reminding. It is a condition. But, how it helps is not clear. – jiten Jun 16 '19 at 12:26
  • 1
    Then, you are done, right? $b^{i-1}< a^{i-1}$ ... – rtybase Jun 16 '19 at 12:27
  • @rtybase But, the first part is not solved still. I mean the unsuccessful attempt. – jiten Jun 16 '19 at 12:32
  • 1
    You can find some proofs of the equality here: How to simplify $a^n - b^n$? – Minus One-Twelfth Jun 16 '19 at 12:56
  • @MinusOne-Twelfth the response by Mawhrin-skel feels the best as details the easier approach without resort to induction. I request something extra: is there any analytical significance of the formula.It is a novice's question, but some response or discussion will help a lot. Also, it seems like telescoping technique will not work here, as the second answer (there) by Tony Stark shows a binomial expansion, so the inside terms do not cancel out each other – jiten Jun 16 '19 at 13:13

4 Answers4

2

\begin{align} (a-b)\sum_{i=1}^n (a^{n-i}b^{i-1}) &=\sum_{i=1}^n (a^{n+1-i}b^{i-1}-a^{n-i}b^i )\\ &=a^n+\sum_{i=2}^n a^{n+1-i}b^{i-1}-\sum_{i=1}^{n-1}a^{n-i}b^i - b^n \\ &= a^n+\sum_{i=1}^{n-1}a^{n-i}b^i-\sum_{i=1}^{n-1}a^{n-i}b^i-b^n\\ &=a^n-b^n \end{align}

You might like to read the working backward to be similar to what you attempted.

$b<a$, then

$$b^{i-1}\le a^{i-1}$$

$$a^{n-i}b^{i-1}\le a^{n-1}$$

$$\sum_{i=1}^na^{n-i}b^{i-1}\le \sum_{i=1}^na^{n-1}=na^{n-1}$$

Siong Thye Goh
  • 149,520
  • 20
  • 88
  • 149
  • 1
    You have seemingly used the telescoping approach. Please comment on my last comment to post to @MinusOne-Twelfth, for impossibility of using the telescoping approach, as the binomial approach shows an expansion. I mean the answer at : https://math.stackexchange.com/a/409071/424260. – jiten Jun 16 '19 at 13:14
  • Write out a few terms and verify if it is a telescoping sum. A question can be approached from various angles sometimes, it doesn't mean that if an approach works, the other approach doesn't. – Siong Thye Goh Jun 16 '19 at 13:20
  • 1
    So, the binomial expansion is just a dumber way to prove the same. I mean that the binomial expansion will not show the cancelling terms, so not a correct way or as detailed way as the telescoping approach. So, is the telescoping approach a finer one. Also, any possible analytical significance of the formula proved in the first part will guide me a lot to search for 'real' significance of the formula, which goes beyond even the telescoping approach. – jiten Jun 16 '19 at 13:21
  • no comments about intelligence of various methods. – Siong Thye Goh Jun 16 '19 at 13:25
  • Please tell something about the 'real' or analytical significance of the formula in question. I mean if could grasp more than derivation that concerns the behavior of the formula in real plane. Also, find my edit for showing the telescoping nature of the derivation by first few terms. – jiten Jun 16 '19 at 14:13
  • Please see my new post at: https://math.stackexchange.com/q/3264938/424260. It concerns with a completeness (real analysis) based proof of existence of irrational roots of $\sqrt{2}$. It has attracted no comment (connected to solve the problem(s) posed) even after around 30 minutes, and having suitable tags to attract attention. It is mostly concerned with inequalities, and how change of sign turns $\lt$ to $\le$. – jiten Jun 17 '19 at 05:42
  • Please post something for my last post, as feel something is lacking in selected answer, but cannot tell clearly enough. It might be the comment by @HennoBrandsma is pointing to invalid use of IVT in the selected answer, but not explains his point clearly. It might be that there is something still lacking in there. – jiten Jun 17 '19 at 06:27
2

For the equalty,

By telescoping: We have,

$$(a-b)\sum_{i=1}^na^{n-i}b^{i-1}=a\sum_{i=1}^na^{n-i}b^{i-1}-b\sum_{i=1}^na^{n-i}b^{i-1},$$ which can be rewritten $$\sum_{i=1}^na^{n+1-i}b^{i-1}-\sum_{i=1}^na^{n-i}b^i=a^n+\sum_{i=2}^na^{n+1-i}b^{i-1}-b^n-\sum_{i=1}^{n-1}a^{n-i}b^i,$$ which on simplifying gives $$a^n-b^n+\sum_{i=1}^{n-1}a^{n-i}b^i-\sum_{i=1}^{n-1}a^{n-i}b^i=a^n-b^n.$$ By induction: You wish to prove $$a^n-b^n=(a-b)\sum_{i=1}^n a^{n-i}b^{i-1}.$$ Let's try induction. For the $n=1$ case we have

$$(a-b)a^0b^0=a^1-b^1,$$ so the base case holds.

Now suppose the general case is true and consider the $n+1$ case. We have

$$(a-b)\sum_{i=1}^{n+1}a^{n+1-i}b^{i-1}=(a-b)\sum_{i=1}^n a^{n+1-i}b^{i-1}+(a-b)a^0b^n,$$

which can be rewritten $$a(a^n-b^n)+(a-b)b^n=a^{n+1}-ab^n+ab^n-b^{n+1}=a^{n+1}-b^{n+1},$$ so in fact the general case holds.


For the inequality, you can use induction too:

$$a^n-b^n\leq (a-b)na^{n-1}.\tag{*}$$ The base case clearly holds since $a^1-b^1\leq (a-b)a^0$. Now suppose (*) holds and consider the $n+1$ case,

$$a^{n+1}-b^{n+1}=a^na-b^nb=(a^n-b^n)(a+b)-a^nb+b^na\tag{1}$$

But since $a>b$, then

$$(1)\leq (a^n-b^n)(a+b)-a^nb+a^nb=(a^n-b^n)(a+b)-b(a^n-b^n),$$ which we can write $$(a^n-b^n)a\leq a(a-b)na^{n-1}=(a-b)na^n\leq(a-b)(n+1)a^n,$$ as required.


pshmath0
  • 10,565
  • I request that the derivation be in a constructive manner, not by induction. Also, if possible the analytical significance of the same be at least hinted at. – jiten Jun 16 '19 at 12:57
  • 1
    Sorry I misread, I thought "... using telescoping technique or not." referred to any method. I have also worked out the inequality proof too. – pshmath0 Jun 16 '19 at 13:06
1

Geometric progression \begin{align} (a-b)\sum_{i=1}^{n}a^{n-i}b^{i-1}&=(a-b)a^{n-1}\left(\frac{1-\left(\frac{b}{a}\right)^n}{1-\frac{b}{a}}\right)\\ &=a^n-b^n. \end{align} Can you complete the answer now?

Explorer
  • 3,147
0

Hint: Use that $$\sum_{i=1}^na^{n-i}b^{i-1}=\frac{a^n-b^n}{a-b}$$ if $$a\ne b$$