I came across a convergence test for improper integrals referred to as the $\mu$-test while I was looking through a textbook. I'm interested in understanding the idea behind the test since no explanation is given in the textbook.

Let $f(x)$ be unbounded at $a$ and integrable in the interval $[a+ \epsilon, b]$ where $0 < \epsilon < b-a$.

If there is a number $\mu$ between $0$ and $1$ such that $\lim_{x \to a^{+}} (x-a)^{\mu} f(x)$ exists, then $\int_{a}^{b} f(x) \ dx$ converges absolutely.

If there is a number greater than or equal to $1$ such that $\lim_{x \to a^{+}} (x-a)^{\mu} f(x)$ exists and is nonzero, then $\int_{a}^{b} f(x) \ dx$ diverges, and the same is true if $\lim_{x \to a^{+}} (x-a)^{\mu} f(x) = \pm \infty$.

In the case $\int_{a}^{b} f(x) \ dx$ is unbounded at $b$, we should find $\lim_{x \to b^{-}} (x-b)^{\mu} f(x)$, other conditions remaining the same.

If $\mu =1$ (or any positive integer) and the limit exists and is nonzero, then $f(x)$ would have a pole at $a$ if $f(x)$ were a function of a complex variable. So from that perspective I can see why the integral would diverge.

  • $\begingroup$ It seems that the second case covers all the conditions possible for $\mu\ge 1$ ? $\endgroup$ Oct 8, 2013 at 13:19

1 Answer 1


If $\lim_{x\to a^+} (x-a)^\mu f(x) = L$, then $|f(x)| < 2L|x-a|^{-\mu}$ near $x=a$.

Since $$\int_a^b |x-a|^{-\mu}\,dx$$ converges for $0 < \mu < 1$, the integral $$\int_a^b f(x)\,dx$$ converges absolutely. The other assertions are similar.


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.