1

I have this integral, and the solution gives an indeterminate form for the value $\alpha = 1$, can you explain to me how to solve the indeterminate form?

$$\int_{\beta}^{+\infty} x^{-\alpha} dx = \frac{1}{1-\alpha} \left[ x^{-\alpha + 1}\right]_{\beta}^{+\infty} = \frac{1}{0} \left[ +\infty^{0} - 1\right] =?$$

Little side note, I know the solution of this integral I am really just curious od knowing how to tackle the indeterminate form.

Matteo
  • 187
  • 1
    do you know what is $ \int \frac{dx}{x} $ ? – what'sup Oct 13 '13 at 18:21
  • @what'sup - Yeah, log(x), so? – Matteo Oct 13 '13 at 18:22
  • @what'sup - Maybe I should specify in the question that I know what shoul be the result, I am interested in understanding how to solve the indeterminate form... – Matteo Oct 13 '13 at 18:24
  • 1
    http://math.stackexchange.com/questions/64432/what-is-so-special-about-alpha-1-in-the-integral-of-x-alpha?lq=1 and http://math.stackexchange.com/questions/498339/demystify-integration-of-int-frac1x-dx?lq=1 – what'sup Oct 13 '13 at 18:26
  • @what'sup - very interesting links, thks a lot! But suppose I omit the previous part and just give you the indeterminate form, any clue on how to continue from there? – Matteo Oct 13 '13 at 18:30
  • Such intgrals converge only for $\alpha>1$ – Aman Mittal Oct 13 '13 at 18:47
  • 1
    @Matteo: the trick is to replace $;\displaystyle\frac{1}{1-\alpha} \left[ x^{-\alpha + 1}\right]{\beta}^{+\infty};$ by $;\displaystyle\frac{1}{1-\alpha} \left[ x^{-\alpha + 1}-1\right]{\beta}^{+\infty};$ to get a correct result at the limit (as shown in the two links proposed by what's up). – Raymond Manzoni Oct 13 '13 at 19:18

1 Answers1

1

It's not really indeterminate. What you are trying to do is $$ \lim_{\alpha\to 1^+}\int_\beta^\infty x^{-\alpha}\,\, dx = \lim_{\alpha\to 1^+}\frac{x^{1-\alpha}}{1-\alpha}. $$ But this is of the form $\displaystyle \frac{1}{0}$ which is not indeterminate.

Your issue is that to reach that last formula in your post you have to assume the integral exists.

J126
  • 17,707
  • but the integral does exist for $\alpha \to 1^{+}$ – Matteo Oct 13 '13 at 19:01
  • @Matteo: The integral exists for all $\alpha >1$. So, we can try to see if the limit exists. It makes no sense to say that the integral exists for $\alpha\rightarrow 1^+$. – J126 Oct 13 '13 at 20:28