# A contradiction involving derivative [duplicate]

Possible Duplicate:
Where is the flaw in this argument of a proof that 1=2?

I am unable to find where the error is occurring in the following (I guess I can't take derivative, but why?):

$\underbrace{x+x+\dotsc}_{x \text{ times}}=x^2$

Differentiating both sides w.r.t. $x$, $\underbrace{1+1+\dotsc~~}_{x \text{ times}}=2x \implies x=2x$ !

Could anyone please explain the error.

• Honestly, I asked this question being influenced by this questionlink – Tapu Oct 12 '11 at 18:35
• "$x$ times" only makes sense when refering to the number of times a term appears if $x$ is an integer. Therefore your first equation only makes sense for integers $x$, so you cannot differentiate it. – Mariano Suárez-Álvarez Oct 12 '11 at 18:40
• If $x$ isn't an integer, what does it mean to add "$x$ times"? – J. M. isn't a mathematician Oct 12 '11 at 18:41
• +1 to Mariano and J. M.; "Well, this is so simple then." :) – Tapu Oct 12 '11 at 18:47
• @Nate Eldredge, you are correct...its duplicate! I did not know, neither the problem pop up when I was writing. – Tapu Oct 15 '11 at 4:59

This is ancient. I saw it in the '70s. You're saying $x$ is the number of terms. The number of terms can be $0,1,2,3,4,\ldots$ but it cannot be $3.14159\ldots$. The rules of differentiation apply to functions of a continuous variable, not to functions of a variable that varies discretely like this.

• Sure, I have found it somewhere, but did not know the answer. – Tapu Oct 12 '11 at 18:58
• Why did someone downvote this? If there's some message you're trying to get across, you failed, since I have no idea what it is. An articulate comment here would work. – Michael Hardy Oct 12 '11 at 20:52
• It seems to me that the main problem is treating the x in "x times" as a constant rather than treating it as an integer, as explained in the answer by Dave L. Renfro and the last sentence of the answer by Bill Cook. While the usual definition of $a+a+\dotsc~~b\text{ times}$ does not work for non-natural b it would seem reasonable to extend it by defining it as $a\times b$. – aaa Oct 23 '11 at 19:13

Regarding this apparent paradox, it is often said that $$x + x + ... + x = x^{2}$$ is only meaningful when $$x$$ is a positive integer, and that differentiation rules for $$x^2$$ assume $$x$$ is a continuous variable. However, I don't think this gets to the heart of the matter. Consider $$f(x) = x^2$$ with domain equal to the set of positive integers. Then we can't take the derivative in the usual way, but we can still consider $$\Delta = 1$$ difference quotients:

$$\frac{f(x+1) - f(x)}{1} \;\;= \;\;(x+1)^{2} - x^{2} \;\;=\;\; 2x + 1$$

Note that for large values of $$x$$, $$2x + 1$$ is asymptotically equal to the derivative of $$x^2$$, but $$2x + 1$$ is not asymptotically equal to $$1$$ added to itself $$x$$-times, which is $$1 + 1 + ... + 1 = x.$$

The factor of $$2$$ error comes about not because we're looking at functions defined only for the integers, but because we're incorrectly applying the sum rule for differentiating $$x + x + ... + x$$ ($$x$$-times). Specifically, this isn't a fixed-length sum. In order to correctly increment $$x$$ in the expression $$x + x + ... + x$$ ($$x$$-times), we have to increment each of the summands of $$x$$ AND increment the number of the $$x$$'s being added. Or, another way to look at this, in getting $$1 + 1 + ... + 1 = x,$$ we're incorrectly applying the product rule to $$x^{2}=x \cdot x$$ to get $$1 \cdot x=x.$$

Back in 2003 I posted (in the ap-calculus listserv) an ASCII diagram of what's going on, but I haven't been able to get a version to correctly display here. However, that diagram (along with some more explanation) can be found at the following URL. (The correct display of my original post at Math Forum has not survived their last two or three site modifications.)

Incidentally, there is a Letter to the Editor by David W. Erbach in The (Two-Year) College Mathematics Journal [Volume 6, Number 4, December 1975, pp. 2-3] that uses the greatest integer function in some way to get things to work out.

• Dave, +1 and thank you for these nice references. – Tapu Oct 12 '11 at 21:53

You have $\sum_{k=1}^x x = x^2$ so your summation depends on $x$. Thus as "$x$ changes" so does the length of the sum. Thus you'd need to figure out: What do you mean by summing $x=2.763$ number of $x=2.763$'s?

This is really the same issue that appears when taking the derivative of $x^x$ (which is not $x \cdot x^{x-1}$ since the exponent is changing).

First recall the definition of a derivative: Let $X$ be a subset of the real numbers and $x_0\in X$ be such that $x_0$ is also a limit point of $X$. Let $f:X\rightarrow R$ be a function. If the limit:

$\lim_{x\to x_0;x\in X-\{x_0\}}\frac{f(x)-f(x_0)}{x-x_0}$

converges to some real number $L$ then we say that $f$ is differentiable at $x_0$ with derivative $L$.

Now in our case the moment we say $x+x+\cdots x$ times $=x^2$ we are implicitly assuming that $x$ is an element of the set of natural numbers. The function under consideration is given by $f(x)=x^2$ with domain the set of natural numbers. This function is not differentiable at any point $x_0$ as the set of natural numbers has no limit points. Hence we cannot take the derivative at all.