Isolated Singularities: Removable, Poles and Essential Types

An isolated singularity of f at z = a is a point where f fails to be analytic, yet f is analytic at every nearby point — that is, on some punctured neighbourhood 0 < |z − a| < r. The word "isolated" means the bad point stands alone, with no other singularities crowding against it.

Classifying isolated singularities matters because it decides everything that follows: a removable singularity can be patched, a pole contributes a computable residue to integrals, and an essential singularity behaves so wildly that only deep theorems describe it. The classification is read directly from the Laurent series.

Key Concepts

• Isolated vs non-isolated: singularities at z = 1/(nπ) of 1/sin(1/z) accumulate at 0, so z = 0 is a non-isolated singularity and the classification below does not apply to it.
• Removable singularity: the Laurent principal part is zero; defining f(a) as the limit makes f analytic there.
• Pole of order m: the principal part has finitely many terms, with a₋ₘ ≠ 0 the lowest; |f(z)| → ∞ as z → a.
• Essential singularity: the principal part has infinitely many non-zero terms; behaviour near a is chaotic.
• Riemann's theorem: if f is bounded near an isolated singularity, the singularity is removable.
• Casorati–Weierstrass: near an essential singularity, f comes arbitrarily close to every complex value.

Formulas

• Removable test: lim (z → a) f(z) exists and is finite. Example: (sin z)/z at z = 0.
• Pole test: lim (z → a) (z − a)ᵐf(z) exists, is finite and non-zero, for some positive integer m — the order of the pole. Example: 1/(z − 2)³ has a pole of order 3 at z = 2.
• Essential test: the limit as z → a neither exists finitely nor is infinite. Example: e^(1/z) at z = 0.
• Zeros and poles: if f has a zero of order m at a, then 1/f has a pole of order m at a.

Worked Example

Classify the singularity of f(z) = (eᶻ − 1)/z at z = 0.

1. The function is undefined at z = 0, but analytic for every z ≠ 0, so z = 0 is an isolated singularity.
2. Expand the numerator: eᶻ = 1 + z + z²/2! + z³/3! + …, so eᶻ − 1 = z + z²/2! + z³/3! + ….
3. Divide by z: f(z) = 1 + z/2! + z²/3! + … — the principal part is empty.
4. Hence the singularity is removable, with lim (z → 0) f(z) = 1.
5. Setting f(0) = 1 extends f to an analytic function. Numerically, at z = 0.01: (e^0.01 − 1)/0.01 ≈ 1.00502, matching 1 + 0.01/2 = 1.005 from the series.

Common Misconceptions

• "A removable singularity means the function is fine there." Almost — the function is undefined at the point itself; only after filling in the limit value does it become analytic.
• "Poles always look like 1/z." A pole of order m behaves like 1/(z − a)ᵐ, and the order matters: residues and integral contributions depend on it.
• "Essential singularities are just very bad poles." No — they are qualitatively different. By the Casorati–Weierstrass theorem, e^(1/z) attains values arbitrarily close to any complex number in every neighbourhood of 0, something no pole does.

Real-World Applications

• Residue theory: only poles (and essential singularities) contribute residues, so classification is step one of every contour integral.
• Control engineering: poles of a transfer function determine system stability; their orders and locations are the whole design problem.
• Numerical analysis: removable singularities must be handled by series expansion rather than direct evaluation to avoid 0/0 errors.

Isolated singularities come in exactly three flavours: removable (patchable, empty principal part), poles (finite principal part, function blows up), and essential (infinite principal part, wild behaviour). The limit tests above classify any example in seconds, and the Laurent series always confirms the verdict.