2
$\begingroup$

How to get the asymptotic expansion for the integral $$\int_{0}^{1}\exp(-x/t)dt$$ in the limit $x\rightarrow 0$ ? I took $x/t=u$ and did integration by parts (IP) but if I keep doing IP, I get a series in $\exp(-x)$ times increasing powers of $x^{-1}$; this does not look like a valid asymptotic expansion. Another way I tried to use is to write $1/t$ as a geometric series in $1-t$ but it does not help either. Can I just take the integrand to be $\exp(-x)$ because the maximum contribution to the integral is from $t=1$ and then just expand $\exp(-x)\approx 1-x$ in the limit $x\rightarrow 0$ and call that as the asymptotic expansion ?

$\endgroup$
2
  • 1
    $\begingroup$ Substituting $u = x/t$ gets you something you can work with. $\endgroup$ Oct 29 '14 at 15:54
  • $\begingroup$ Your integral equals $e^{-x}-\Gamma(0,x)$ where $\Gamma(a,z)$ is the incomplete gamma function. Mathematica gives the expansion $1+x\log x - (1-\gamma)x-x^2/2+O(x^3)$, where $\gamma$ is the Euler–Mascheroni constant. $\endgroup$ Jan 5 '18 at 11:11
3
$\begingroup$

If you suppose $x>0$ and put $v=x/t$, you get $$F(x)=x\int_x^{+\infty}\frac{\exp(-v)}{v^2}dv=x\int_1^{+\infty}\frac{\exp(-v)}{v^2}dv+x\int_x^1\frac{\exp(-v)}{v^2}dv$$ Hence $$F(x)=x\int_1^{+\infty}\frac{\exp(-v)}{v^2}dv+x\int_x^1\frac{\exp(-v)-1+v}{v^2}dv+x\int_x^1\frac{1-v}{v^2}dv$$

And

$$F(x)=x\int_1^{+\infty}\frac{\exp(-v)}{v^2}dv+x\int_x^1\frac{\exp(-v)-1+v}{v^2}dv-x+1+x\log x$$

Now $$\frac{\exp(-v)-1+v}{v^2}=\sum_{k\geq 2}\frac{(-1)^k}{k!}v^{k-2}$$ and $$ \int_x^1\frac{\exp(-v)-1+v}{v^2}dv=\sum_{k\geq 2}\frac{(-1)^k}{k!}(\frac{1-x^{k-1}}{k-1})$$

Finally: $$F(x)=1+x\log x+cx+\sum_{k\geq 2}\frac{(-1)^{k-1}}{k!(k-1)}x^k$$ with $$c=\sum_{k\geq 2}\frac{(-1)^k}{k!(k-1)}+\int_1^{+\infty}\frac{\exp(-v)}{v^2}dv-1$$

$\endgroup$

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.