In order to prove that $\int_{0}^{0.5}\frac{1}{x^2-1}$ converges I compared it to $\int_{0}^{0.5}\frac{1}{x}$ which converges, by checking that $\lim \frac{\frac{1}{x^2-1}}{\frac{1}{x}}=0$ when $x \to 0$. Then I went to Wolfram Alpha and tried to check $\int_{0}^{0.5}\frac{1}{x^2-0.1}$ (It is suppose to converges by the same test) but it said that it diverges, while $\int_{0}^{0.5}\frac{1}{x^2-0.3}$ doesn't. (with $0.2$ it just couldn't compute). What's really going on there between $0.1$ to $0.3$? Is W.A wrong?

Edit: Sorry! $\int_{0}^{0.5}\frac{1}{x}$ obviously doesn't converge. So, in addition to the rest of the question, how can I prove my original integral does converge?

Thank you very much.

  • 3
    $\begingroup$ Short answer to the edited question: Because $x^2$ ranges from $0$ to $0.25$, and $0.1$ is in this range (making the denominator $0$) whereas $0.3$ is outside of this range. $\endgroup$ – Eric Naslund Feb 2 '12 at 14:23
  • $\begingroup$ It should also be pointed out that there are missing $dx$ terms in most of your posts (at least when they originally are written). This is a bad habit to start! $\endgroup$ – JavaMan Feb 2 '12 at 17:58

Between $x=0$ and $x=0.5$, the function $\frac{1}{x^2-1}$ is perfectly respectable! Note that the denominator is never $0$ in our interval. The largest absolute value is reached at $x=0.5$. So your function has no issues, it is continuous on a closed interval. For the problem you were initially considering, we are finished.

But the question you were led to ask is more interesting, and shows a good effort to understand the situation.

Look first at $\frac{1}{x^2-0.3}$. The denominator is $0$ at $x=\pm\sqrt{0.3}$. The positive root is roughly $0.547722$, outside our interval, though not by much. Thus the function $\frac{1}{x^2-0.3}$ is well-behaved in the interval $[0,0.5]$.

Look now at $\frac{1}{x^2-0.1}$. The denominator is $0$ at $x=\pm\sqrt{0.1}$. The positive root is about $0.3162278$, and this is inside our interval. So our function blows up inside our interval, and there may be a problem. Indeed there is.

You know that a function can blow up, but despite that the integral converges. A standard example is $\int_0^1 \frac{dx}{\sqrt{x}}$. We will show that $\int_0^{0.5}\frac{dx}{x^2-0.1}$ diverges.

As mentioned above, there is potential trouble at $\sqrt{0.1}$. To make typing easier, let $a=\sqrt{0.1}$. Our function is not defined at $x=a$, and blows up near $x=a$. Recall that $a$ is in our interval. When we are dealing with a singularity inside our interval, it is useful to break up the interval into two integrals, in this case from $0$ to $a$ and from $a$ to $0.5$.

We will show that $\int_0^a \frac{dx}{x^2-0.1}$ diverges. (The integral from $a$ to $0.5$ also does, but showing that one of the integrals is bad is enough.)

So we are looking at the integral $\int_0^a\frac{dx}{x^2-a^2}$. For no good reason, except for a preference for the positive, we look instead at $$\int_0^a \frac{dx}{a^2-x^2}.$$ Make the change of variable $w=a-x$. Note that $a^2-x^2=(a-x)(a+x)=w(2a-w)$. Quickly we arrive at $$\int_{w=0}^a \frac{dw}{w(2a-w)}.$$ This integral diverges, by comparison with $\int_0^a\frac{dw}{w}$, which, as pointed out by anonymous, diverges.

  • $\begingroup$ Thanks a lot! very helpful and clear. $\endgroup$ – Jozef Feb 2 '12 at 15:28

You have a mistake. The integral $\int_{0}^{0.5} dx/x$ does not converge! in fact, we can easily calculate it, as the antiderivative of $1/x$ is $lnx$, and $\lim_{x\to 0^+} ln(x) = -\infty$.

  • $\begingroup$ Right! so how can I prove that my original integral does converge? $\endgroup$ – Jozef Feb 2 '12 at 14:12
  • $\begingroup$ Well, your integral is actually a definite integral. The function $\frac{1}{x^2-1}$ is continuous on the interval $[0,1/2]$, so there is no question of convergence at all. $\endgroup$ – the L Feb 2 '12 at 14:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.