MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

The modified Bessel function (Macdonald function) $K_\alpha(z)$ is known to have the following asymptotic expansion for large positive $z$: $$ K_\alpha(z)=\sqrt{\frac{\pi}{2z}}e^{-z}\sum_{k=0}^\infty \frac{b_k(\alpha)}{z^ k} $$ where $b_1(\alpha)=1$, $b_2(\alpha)=\frac{\alpha^2-1^2}{1!8}$, $b_3(\alpha)=\frac{(\alpha^2-1^2)(\alpha^2-3^2)}{2!(8)^2}$ and so on. Is there any simple integral representation, for which it would be a perturbative expansion such that $$ K_\alpha(z)=h(z) \int_C \exp\left(\frac{f(y)}{z}\right) g(y)^\alpha d\mu(y) $$ where $f(x)$, $g(x)$, $h(x)$ and $d \mu(x)$ are $\alpha$-independent?

share|cite|improve this question
I think you want the integrand to have the factor $\exp(z f(y))$, since you are taking the $z\to\infty$ limit and want the argument of the exponential to vary quickly with $y$. – Igor Khavkine Jan 31 '13 at 15:38

The DLMF lists multiple integral representations of $K_\nu(z)$. Here's one that fits your bill:

$$\mathop{K_{{\nu}}}\nolimits\!\left(z\right)=\frac{\pi^{{\frac{1}{2}}}(\frac{1}{2}z)^{\nu}}{\mathop{\Gamma}\nolimits\!\left(\nu+\frac{1}{2}\right)}\int _{0}^{\infty}e^{{-z\mathop{\cosh}\nolimits t}}(\mathop{\sinh}\nolimits t)^{{2\nu}}dt .$$

For integer $\nu$, the contour could be extended by symmetry to all of $\mathbb{R}$.

share|cite|improve this answer
Thank you, I know this nice integral representation, but I want something (maybe) more complicated - with $\frac{1}{z}$ as I wrote. – Sasha Jan 31 '13 at 15:54
Sorry, I don't follow your logic about $1/z$ in $g(f(y)/z)$. What kind of expansion would you expect to perform? The integral formula I suggested should be amenable to a Laplace-type (or steepest descent) expansion for large $z$. In any case, most known integral representations are in fact listed on the DLMF page. – Igor Khavkine Jan 31 '13 at 20:53
I just want to have simple $1/z$ expansion for large $z$ (and not a usual steepest descent expansion). – Sasha Feb 1 '13 at 13:06

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.