I am trying to find an analytical solution of an integral of the form $$ \int_a^{\infty} x^p [e^{\epsilon(x^{-m} - x^{-n})}-1]dx $$ where $0<a<+\infty$ and $p\in[0, 1, 2, 3]$

I understand that there is no way to express this integral using standard mathematical functions and that is fine. That said...

Is it possible to use a series expansion for the exponential term?

The term in square brackets can be expressed as: $$ \sum_{k=1}^\infty \frac{[\epsilon(x^{-m} - x^{-n})]^k}{k!} $$

but when I do this, I get an integral that diverges... I am most likely doing something wrong because I have computed the numerical solution of this integral and it does not diverge.


In the case where $m=2n$, we can use the exponential generating function of Hermite polynomials $H_n(x)$ : $$ \exp (2xt-t^2) = \sum_{n=0}^\infty H_n(x) \frac {t^n}{n!} $$

  • $\begingroup$ What are m, n, and p that get you that graph, as it stands it incredibly general. Or did you sample n and m also to plot it $\endgroup$ – Triatticus Oct 27 '16 at 23:53

If $\min(m,n) > p+1$,

$$\eqalign{\int_a^\infty x^p \dfrac{(x^{-m} - x^{-n})^k}{k!} \; dx &= \sum_{j=0}^k \int_a^\infty {k \choose j} (-1)^{j} x^{p - j n - (k-j) m}\; dx\cr &= \sum_{j=0}^k {k \choose j} (-1)^{j+1} \dfrac{a^{p-jn-(k-j)m+1}}{p-jn-(k-j)m+1} \cr }$$

It's not pretty, but I don't see how you're getting a divergent integral out of it.

In the case $a=1$, this actually simplifies to an expression using the Gamma function:

$$ \dfrac{\Gamma \left( {\frac { \left( -k+1 \right) m-n+p+1}{m-n}} \right) k!}{\Gamma \left( {\frac { \left( -k+1 \right) m-n+p+1}{m-n}} + k \right) \left( mk-p-1 \right) } $$


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.