The Laurent Series That Refuses to Converge Everywhere
Let .
Part (a): Find the Laurent series expansion of valid in the annulus .
Part (b): Find the Laurent series expansion of valid in the region .
Part (c): Both series represent the same function , yet they look completely different. What is the conceptual reason they must differ?
Answer: Two Laurent Series for One Function
Key Idea / Intuition
A Laurent series is not just a property of a function โ it is a property of a function on an annulus. The same meromorphic function can have completely different Laurent series in different annular regions, because each region "sees" different poles as being "inside" vs "outside." The coefficients are determined by residues and integrals that depend essentially on which singularities are enclosed. This is the heart of why Laurent series are attached to domains, not just to functions.
Formal Proof / Solution
Setup: Partial Fractions
First, decompose via partial fractions:
(Check: . โ)
The singularities are at (pole) and (pole).
Part (a): Laurent Series on
In this region, , so can be expanded as a geometric series:
Therefore:
This has a simple pole at (the term), as expected.
Part (b): Laurent Series on
In this region, , so should be expanded in powers of :
Also, is already a power of . Combining:
The and cancel!
This series has no negative powers beyond โ in particular, the residue at is zero, which is consistent with as .
Part (c): Conceptual Reason They Differ
The Laurent series on an annulus is unique โ there is exactly one such series converging there. The two annuli and are separated by the singularity at .
- On : the singularity at is outside the disk, so expands in non-negative powers of .
- On : the singularity at is inside the circle, so must be expanded in negative powers of .
In the language of the Cauchy integral formula: the Laurent coefficients depend on . Crossing the singularity at changes which poles are enclosed, changing all the coefficients.
The two series are genuinely different functions of their respective variables โ they just happen to represent the same meromorphic function in their respective domains of validity.
Source: Complex Analysis (SteinโShakarchi), Chapter 3; standard folklore