2
$\begingroup$

How can I prove that $$\int_0^1(1+\epsilon^{-1}e^{-\beta x/p\epsilon})dx\leq C,$$ where $\epsilon$ is a small parameter. In the limit $\epsilon^{-1}$ is unbounded so how is this integral bounded?

$\endgroup$
  • $\begingroup$ It seems to me that you can calculate the integral explicitly, and then see what happens to your $\varepsilon$. $\endgroup$ – R.T. Aug 8 '13 at 16:39
5
$\begingroup$

Just calculate:

$$\int_0^1 \! (1 + \epsilon^{-1}e^{-\beta x/p\epsilon}) \, dx$$ $$ = 1 + \epsilon^{-1}\int_0^1 \! e^{-\frac{\beta}{p \epsilon}x} \, dx$$ $$ = 1 + \frac{1}{\epsilon}\frac{-p\epsilon}{\beta}(e^{\frac{-\beta}{p\epsilon}} - 1)$$ $$ = 1 + \frac{p}{\beta}(1-e^{\frac{-\beta}{p\epsilon}})$$ This last expression is bounded in absolute value as long as $\epsilon$ does not become arbitrarily negative. Basically the reason the integral stays bounded is that the exponential term becomes small much faster than the $\epsilon^{-1}$ term becomes large as $\epsilon \to 0$ from the right. Notice also that you are only asking for an upper bound $$\int_0^1 \! (1 + \epsilon^{-1}e^{-\beta x/p\epsilon}) \, dx \le C$$ and such an expression is valid for any $\epsilon$ with $C = 1 + \frac{p}{\beta}$.

$\endgroup$
2
$\begingroup$

If I am not mistaken, your integral equals $$1+\frac{p}{\beta}(1-e^{-\frac{\beta}{p\varepsilon}})$$

Assuming that your parameters are positive, the result follows.

$\endgroup$
2
$\begingroup$

Change variables, $x=\epsilon u$ in your integral: $$ \begin{align} \int_0^1\left(1+\epsilon^{-1}e^{-\frac{\beta x}{p\epsilon}}\right)\,\mathrm{d}x &=\int_0^{1/\epsilon}\left(\epsilon+e^{-\frac{\beta u}{p}}\right)\,\mathrm{d}u\\ &\to1+\int_0^\infty e^{-\frac{\beta u}{p}}\,\mathrm{d}u\\ &=1+\frac{p}{\beta} \end{align} $$ as $\epsilon\to0$. So for $\epsilon$ near $0$, the integral is near $1+\frac{p}{\beta}$.

$\endgroup$

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.