Recipe
Expand a function as a Laurent series in a given annulus
Use when: you're asked for the Laurent series of f centered at z_0, valid specifically on one named annulus between two of f's singularities
Goal: the two-sided series \sum c_n(z-z_0)^n that actually converges to f on that one annulus, not some other one centered at the same point
Taught in Laurent's theorem
Quick recipe
-
1
Find every point where f fails to be analytic. Record each one's distance from the center z_0. These distances cut the plane into the largest annuli that Laurent's theorem allows.
Example
For f(z)=\dfrac{1}{(z+1)(z-2)} centered at z_0=0: the singularities sit at distance 1 and distance 2 from 0, giving three regions, |z|<1, 1<|z|<2, and |z|>2.
Why: a different annulus around the same center gives a different Laurent series. So you must name the target region before any algebra.
-
2
Write f as a sum of simple fractions. Each one has exactly one of the singularities from Step 1 in its denominator.
Example
\dfrac{1}{(z+1)(z-2)}=-\dfrac13\cdot\dfrac{1}{z+1}+\dfrac13\cdot\dfrac{1}{z-2}.
Why: the geometric-series trick works on one linear factor at a time.
-
3
For a term \dfrac{1}{z-a}, compare |z| on the target annulus with |a|. Then pick the geometric series that converges there.
Example
On 1<|z|<2: z is farther from 0 than the singularity at -1, so that term takes the outside form. But z is still closer than the singularity at 2, so that term takes the inside form.
- |z|<|a| on the target annulus — use \dfrac{1}{z-a}=-\dfrac1a\sum_{n=0}^\infty\dfrac{z^n}{a^n}, all non-negative powers.
- |z|>|a| on the target annulus — use \dfrac{1}{z-a}=\dfrac1z\sum_{n=0}^\infty\dfrac{a^n}{z^n}, all negative powers.
Why: a geometric series only converges when its ratio has modulus below 1. Which quantity plays the ratio flips at each singularity's own distance.
-
4
Write out each chosen expansion. Then add them into one two-sided series in (z-z_0).
Example
-\dfrac13\cdot\dfrac1{z+1}=-\dfrac1{3z}+\dfrac1{3z^2}-\cdots and \dfrac13\cdot\dfrac1{z-2}=-\dfrac16-\dfrac{z}{12}-\cdots, so on 1<|z|<2: f(z)=\cdots+\dfrac1{3z^2}-\dfrac1{3z}-\dfrac16-\dfrac{z}{12}-\cdots.
Why: the two pieces add term by term to give f on the one annulus where both are valid.
Common mistakes
- Giving a Laurent series with no annulus named, as if a function had only one such series.
- Using an inside expansion for a term when the target annulus lies outside that singularity, or the reverse.
- Forgetting to redo Step 3 for every singularity. Each one gets its own inside or outside choice.
- Assuming the coefficients from one annulus still work on a different annulus of the same function.