4
$\begingroup$

I noticed that $\displaystyle \int_{-a}^{b} \frac{x^3}{(x^2+4)(x^2+1)}$ will converge to $0$ whenever $a=b$ and will converge to some value whenever $a,b$ are in the reals (excluding infinity). How would you show that this integral does not converge over the interval $(-\infty, \infty)$ to someone who is not convinced?

When I evaluated this integral using complex residues for the interval $(-\infty, \infty)$ $(Res[z=2i] + Res[z=i])$,I got a complex value as an answer.

$\endgroup$
2
  • $\begingroup$ the highest power in the nominator the nominator is three, in the denominator it's four. therefore as $x->\infty$, the integral behaves as $\int (1/x)dx=\ log(x)$ which is clearly divergent. $\endgroup$ – tired May 17 '15 at 13:43
  • $\begingroup$ last comment: the integral is also zero on the infinite real line in a slightly more general sense: adding $1/x-1/x$ and split the integral in an intelligent way $\endgroup$ – tired May 17 '15 at 14:08
3
$\begingroup$

$\displaystyle \frac{x^3}{(x^2+4)(x^2+1)} \geq \frac{x^3}{4x^4}=\frac{1}{4x}$

for sufficiently large $x$.

OBS: And what do you mean by "will converge to $0$ whenever $a=b$ and will converge to some value whenever a,b are in the reals (excluding infinity)"? The integral on those values is a bona fide, well-defined, riemann-integral. There is no "convergence".

$\endgroup$
5
  • $\begingroup$ What about the case when $x<0$? $\endgroup$ – Samrat Mukhopadhyay May 17 '15 at 13:43
  • $\begingroup$ you can apply the same reasoning, switching x to -x $\endgroup$ – tired May 17 '15 at 13:45
  • $\begingroup$ btw, it should be 1/x instead of 1/4x because the highest order polynomial in the denominator has coefficent 1 $\endgroup$ – tired May 17 '15 at 13:47
  • $\begingroup$ @tired my reasoning is: $\frac{x^2+4}{x^2} \rightarrow 1$, so $x^2 +4\leq 2x^2$ for sufficiently large $x$. Analogously, $x^2+1 \leq 2x^2$ $\endgroup$ – Aloizio Macedo May 17 '15 at 14:08
  • $\begingroup$ If $a=b$ you integrate an odd function on $[-a,a]$ so you get $0$. $\endgroup$ – Beni Bogosel May 17 '15 at 15:28
3
$\begingroup$

Beside the simple point from Aloizio Macedo which would apply to many cases, may be, going further, you could use the fact that $$\frac{x^3}{(x^2+4)(x^2+1)}=\frac{4 x}{3 \left(x^2+4\right)}-\frac{x}{3 \left(x^2+1\right)}=\frac 23 \frac{2 x}{x^2+4}-\frac 16 \frac{2 x}{x^2+1}$$ So $$\int\frac{x^3}{(x^2+4)(x^2+1)}\,dx=\frac{2}{3} \log \left(x^2+4\right)-\frac{1}{6} \log \left(x^2+1\right)$$ Already, the doefficients show what would be the problem $\frac 23 -\frac 16=+\frac 12$.

Since you already noticed that, if $a=b$, the result is $0$ then $$\int_{-a}^b\frac{x^3}{(x^2+4)(x^2+1)}\,dx=\int_{a}^b\frac{x^3}{(x^2+4)(x^2+1)}\,dx=\frac{1}{6} \log \Big(\frac{(b^2+4)^4\,(a^2+1)}{(a^2+4)^4\,(b^2+1)}\Big)$$ which, forgetting a constant which depend on $a$, is basically $\log(b)$ for large values of $b$.

$\endgroup$

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.