-2

So I was doing problems from Michael Spivak's calculus book & I've faced this question $$\lim_{x \to y}{\frac{x^n- y^n}{x - y}}$$

The answer guide isn't clear/detailed. It's saying that the answer is $ny^{n-1}$ can someone tell me why is that the correct answer? :(

Anne Bauval
  • 34,650
Aueriga
  • 306
  • 3
    it's a $\frac{0}{0}$ indeterminate form, so you might wanna use L'Hospital. Other than that you can factor the numerator – Fix Feb 16 '23 at 15:45
  • What is the context? Are you supposed to calculate the derivative of $x\mapsto x^n$? (if you still don't know it, you cannot use L'Hospital). – Anne Bauval Feb 16 '23 at 15:55
  • 5
    Hint: The key observation is that $x^n-y^n=(x-y)(x^{n-1}+x^{n-2}y+\dots+xy^{n-2}+y^{n-1})$. Do you know how to proceed from there? – Don Feb 16 '23 at 16:00
  • See: https://math.stackexchange.com/questions/117660/proving-xn-yn-x-yxn-1-xn-2-y-x-yn-2-yn-1 – NoChance Feb 16 '23 at 16:05
  • @Don, NoChance: There is nowhere specified that n is a natural number. Please don’t make that particular assumption. – WindSoul Feb 16 '23 at 18:08
  • @AnneBauval there is no particular context, that's a famous law that the limit of the equation above is equal to ny^n-1 (those are just symbols ofc, it'll work for any x, y, and n) but I wasn't understanding why, anyway now I can understand it I found someone who's explained it – Aueriga Feb 16 '23 at 20:43
  • @Don yes! I've found the answer, thank you anyway for helping – Aueriga Feb 16 '23 at 20:44
  • 1
    Duplicate https://math.stackexchange.com/a/2501665/386889 – Anne Bauval Feb 16 '23 at 23:57

2 Answers2

4

That limit is precisely the limit definition of the derivative of $x^n$ at $x = y$. Recall $$ f'(a) = \lim_{h \to 0} \frac{f(a+h)-f(a)}{h} = \lim_{x \to a} \frac{f(x)-f(a)}{x-a}. $$ Consider this when $f(x) = x^n$ and $a = y$.

But maybe this question is being asked in Spivak's book before derivatives have been introduced. (EDIT: It is; see below.) You didn't tell us where it appears in that book. If it appears before the chapter on derivatives, use the standard factorization of $x^n - y^n$ with a factor of $x-y$: $$ x^2-y^2 = (x-y)(x+y), $$ $$ x^3 - y^3 = (x-y)(x^2+xy+y^2), $$ $$ x^4 - y^4 = (x-y)(x^3+x^2y+xy^2+y^3) $$ and so on (the factorization for general $n$ has already been mentioned in a comment above).

EDIT: I found the problem in Spivak's book: part (iv) of problem 1 in Chapter 5. Here is the whole exercise.

Find the following limits. (These limits all follows, after some algebraic manipulations, from the various parts of Theorem 2; be sure you know which ones are used in each case, but don't bother listing them.)

(i) $\lim_{x \to 1} (x^2-1)/(x+1)$.

(ii) $\lim_{x \to 2} (x^3-8)/(x-2)$.

(iii) $\lim_{x \to 3} (x^3-8)/(x-2)$.

(iv) $\lim_{x \to y} (x^n-y^n)/(x-y)$.

(v) $\lim_{h \to 0} (\sqrt{a+h}-\sqrt{a})/h$.

That phrase "some algebraic manipulations" in the instructions show that $n$ is meant to be an integer, and (EDIT) Theorem $2$ in Chapter $5$ of the book is about limits being preserved under adding, multiplying, and reciprocating (for a nonzero limit). In fact I'd say $n$ is meant to be a positive integer, although the limit formula is true if $n$ is an arbitrary integer, or even an arbitrary real number, but to handle real exponents goes beyond mere algebraic manipulations. The reader of Spivak is supposed to figure out how to factor $x^n - y^n$ by having already done this in earlier parts for $x^2-1$ and $x^3-8$. OP, were you able to do parts (i), (ii), and (iii) already?

KCd
  • 46,062
  • I believe that the question is not on how do you solve the limit using derivatives, because derivatives are limits and applying the derivative formula is taking advantage of someone else figuring out the limit. The question is neither how do you solve the limit for n a natural number because the rational expression simplifies to a finite sum. The question is how do you prove the derivative formula by evaluating the limit in the most general circumstance when n is not a natural number. – WindSoul Feb 16 '23 at 17:52
  • 1
    @WindSoul are you saying that in Spivak’s book the $n$ is not specified to be an integer? That it isn’t mentioned by the OP could just be an oversight. Moreover, using the notation $n$ strongly suggests the intention was for it to be an integer. Would you write a function of a real or complex variable as $f(n)$? – KCd Feb 16 '23 at 18:16
  • The information you take for granted might as well be mentioned in the referenced book but is not mentioned here- in this question. If you want to be rigorous in your answer then please do edit it to reflect that while the questioner doesn’t mention specifically that n was a natural number, you assumed in your answer that it was. Further on, I find your solution fitting with the tags offered by OP, however using derivatives seems a straightforward solution which wouldn’t require help for someone acknowledged with derivatives. As for f(n): you’re right to be wrong, no such mention by OP. – WindSoul Feb 16 '23 at 18:37
  • I didn't really want to solve it with derivatives, I just wanted to knew from where was the law generated, but thanks anyway – Aueriga Feb 16 '23 at 20:46
  • @WindSoul see my update to my answer (two edits): the question in the book really was asking how to do this when $n$ is a natural number. It is not about arbitrary real exponents. – KCd Feb 16 '23 at 22:35
  • @KCd, Nowhere in the exercise there is any reference to n being a natural number. You may want to refer to ”Theorem 2”, as pointed in the exercise. – WindSoul Feb 16 '23 at 23:09
  • @Ryu, if you don’t want to solve the limit by using derivatives then you need to solve the limit otherwise. Don’t assume n a natural number. – WindSoul Feb 16 '23 at 23:13
  • @WindSoul can you explain more in why would it affect the answer if n is natural or not? – Aueriga Feb 16 '23 at 23:19
  • 1
    @WindSoul the instructions to use algebraic manipulations and the concrete cases preceding part (iv) make it clear to me that $n$ in part (iv) is meant to be an integer (and more likely a positive integer). I looked up Theorem 2 in Chapter 5 of the book: it says limits of functions behave well under addition, multiplication and reciprocals (for a nonzero limit in the last part). That just reinforces that Spivak meant for $n$ to be a positive integer, not an arbitrary real number. I am not going to spend any more time discussing this. – KCd Feb 16 '23 at 23:39
  • @Ryu, I have already provided the argument in the first comment of this answer:”… because the rational expression simplifies to a finite sum”. – WindSoul Feb 16 '23 at 23:39
  • @KCd, the derivative formula stands for n any real number. The problem is to prove it neither for a natural n because that is elementary, nor using the result of the limit because that is not a proof. The problem is to prove the derivative formula by solving the limit-what algebraic manipulations means, and not solve the limit by using the derivative formula-like you did. – WindSoul Feb 16 '23 at 23:45
  • @WindSoul, well, you may be correct and maybe using derivatives is the best way -maybe the only- to prove that formula, but I mean, after all differentiation is built on limits, I don't feel it's right to use it to prove a limit law, you know what I mean? – Aueriga Feb 16 '23 at 23:47
  • Is not the only way. You realize that the derivative formula was obtained by algebraically solving he limit. This is what the exercise asks for. – WindSoul Feb 16 '23 at 23:48
  • @KCd I gave this answer a down vote by accident, and am unable to correct it, unless the answer is edited. If you make make an edit,(maybe change something, save, and then change it back) I should be able to change my down vote to an up vote. – Ben Feb 20 '23 at 19:43
  • @Ben okay, I made a minor edit. – KCd Feb 20 '23 at 23:53
  • Vote corrected, ty – Ben Feb 21 '23 at 03:04
1

I've just came out with the solution guys some friend of mine explained it to me and sent me a video it's actually easy but it uses the Binomial Theorem in the steps (I mean, if you wish to do it without derivatives, which what I wanted in the first place) So first of all, you assume that you wanna approach y from the positive side, and the limit of y from the positive side is just the same as (y + a) where a is a small number that tends to zero a → 0 The limit of the function should be the same from both sides, and we wish to rewrite it as the following $$\lim_{x \to y^+}{\frac{x^n- y^n}{x - y}} = \lim_{x \to y}{\frac{x^n- y^n}{x - y}}$$

And limx→y+ X is just y+a, where: a → 0

then

$$\lim_{x \to y^+}{\frac{x^n- y^n}{x - y}} = \lim_{a \to 0}{\frac{(y + a)^n- y^n}{(y + a) - y}}$$

then we'll take y as a common factor from the first term in the numerator

$$\lim_{a \to 0}{\frac{[y(1 + {\frac{a}{y}})]^n- y^n}{(y + a) - y}}$$ and the 2 terms in denominator: -y and y, will cancel out each other, and we can also distribute the power "n" for the first term $$\lim_{a \to 0}{\frac{[y^n(1 + {\frac{a}{y}})^n- y^n}{a}}$$

and now we're having 2 terms subtracted from each other in the numerator, they both are having the variable yn, so we can take yn as a common factor

$$\lim_{a \to 0}{\frac{y^n[(1 + {\frac{a}{y}})^n- 1]}{a}}$$

Now I KNOW that may seem hard or complex or even meaningless but trust me after those steps everything becomes easier Right now we need to use thee Binomial Theorem, this might be a little bit complicated to understand in text but if you take a paper and a pen & give it a try you'll find that the next step isn't hard the term (1 + a/y)n can be expanded using the binomial theorem, the first term will be +1, the second will be n × a/y, the third will be nC2 × (a/y)2, and then you repeat the process n times, and (a/y)n will be our last term, we will have a -1 at the end from our original expression though, so after the expansion it will look like 1+(a/y)+nC2 × (a/y)n+...-1

nC2 itself could be simplified

$$^nC_2 = {\frac{n!}{(n-2)!2!}} = {\frac{n(n-1)(n-2)!}{(n-2)!2!}}$$ the term (n-2)! in the denominator will cancel out the one in the numerator $${\frac{n(n-1)}{2!}}$$

Now, our main expression will be:

$$\lim_{a \to 0}{\frac{y^n[1 + n{\frac{a}{y}}+{\frac{n(n-1)}{2!}}({\frac{a}{y}})^2+...+({\frac{a}{y}})^n- 1]}{a}}$$

And now, the 1 and the -1 will cancel out each other, wo we will have :

$$\lim_{a \to 0}{\frac{y^n[n{\frac{a}{y}}+{\frac{n(n-1)}{2!}}({\frac{a^2}{y^2}})+...+({\frac{a^n}{y^n}})]}{a}}$$

Now if you pay more attention, you'll notice that a is a common factor in all the terms of the expansion, swap it out!

$$\lim_{a \to 0}{\frac{y^na[{\frac{n}{y}}+{\frac{n(n-1)}{2!}}({\frac{a}{y^2}})+...+({\frac{a^{n-1}}{y^{n-1}}})]}{a}}$$

Now we have an a on the numerator and on the denominator, they'll cancel out each other we'll have $$\lim_{a \to 0}{y^n[{\frac{n}{y}}+{\frac{n(n-1)}{2!}}({\frac{a}{y^2}})+...+({\frac{a^{n-1}}{y^n}})]}$$

Now, substitute for all the a's in the expression with 0, and you'll have all the terms containing a canceled out you'll have the following

$$y^n * {\frac{n}{y}} = ny^{n-1}$$

so, here we came out with the famous rule: $$\lim_{x \to y^+}{\frac{x^n- y^n}{x - y}} = ny^{n-1} $$

and that's all :)

Aueriga
  • 306
  • 3
    You have greatly complicated the solution by switching the limit to $a \to 0$ where $x = y+a$, since it leads you to use the binomial theorem. Stick to the original version of the problem and use the factorization of $x^n - y^n$ as $(x-y)(x^{n-1} + x^{n-2}y + x^{n-3}y^2 + \cdots + xy^{n-2} + y^{n-1})$, with the second factor having $n$ terms $x^iy^{n-1-i}$ where $0 \leq i \leq n-1$. That involves nothing like binomial coefficients and is how you should have solved the earlier parts of the problem in Spivak's book where $n$ is $2$ and $3$. – KCd Feb 16 '23 at 22:37
  • @KCd well, it's not a matter of "complexity", it's all about what solution you're more comfortable with, I see no problem using any of the 2 methods it's all algebra at the end, & there isn't much difference between factoring the expression to n terms and expanding it I guess, but thanks for your help – Aueriga Feb 16 '23 at 22:47