Skip to content

Hartogs's Extension Theorem

Extend a holomorphic function uniquely across a compact hole in a connected domain of at least two complex dimensions.

Version
v1 · 2026-10-03 · History
Domain-specific #
13296
Domain group
Formal Sciences
Origin domain
Mathematics
Subdomains
Complex Analysis, Several Complex Variables → Mathematics
Aliases
Hartogs extension theorem, Hartogs phenomenon compact-hole theorem

Core Idea

Hartogs's extension theorem says that in at least two complex dimensions a holomorphic function cannot sustain a compact interior “hole” as its singularity when the remaining domain is connected. Precisely, let Ω be a domain in ℂⁿ with n≥2, let K⊂Ω be compact, and assume Ω\K is connected. Every function f holomorphic on Ω\K then has a unique holomorphic extension F to all of Ω with F=f outside K.[1][2]

This differs sharply from one complex variable: 1/z is holomorphic on ℂ\{0} but cannot extend holomorphically through zero. In several variables, a puncture or a compact inner region does not grant the same analytic freedom. The result is a theorem about holomorphic functions under stated conditions, not a blanket statement that every missing region of any differential equation is fillable.[2]

Structural Signature

Sig role-phrases:

  • Several-complex-variable domain — Open connected Ω⊂ℂⁿ with n≥2. Dimension is a strict hypothesis, not a decorative parameter: the one-variable pole is a counterexample.[1][2]
  • Compact interior hole — K is compact and contained in Ω, so its removal does not run to the domain boundary or infinity. A deleted noncompact hypersurface can carry different singularity behavior.[1]
  • Connected complement — Ω\K is one connected region on which the original holomorphic function is specified. Without that condition, separately assigned components need not agree as one continuation.[1]
  • Holomorphic input — f is complex analytic throughout the complement, not merely continuous or smooth there. The analytic rigidity is what forces extension.[2]
  • Unique holomorphic output — F exists on all of Ω and agrees with f outside K; uniqueness follows from the identity principle on the connected domain.[1]

Condensed: n≥2 + compact interior K + connected Ω\K + holomorphic f → unique holomorphic F on Ω.

What It Is Not

  • Not a theorem in one complex dimension. 1/z at an isolated puncture shows the dimension hypothesis is essential.[2]
  • Not extension across every excluded set. A noncompact deleted hypersurface or disconnected complement falls outside this compact-hole form.
  • Not a claim about arbitrary smooth functions. Ordinary smooth data on a shell need not be forced to one analytic interior.
  • Not Hartogs's separate-holomorphicity theorem. That concerns joint holomorphy from holomorphy in each variable, a different hypothesis and conclusion.
  • Not every Hartogs-type PDE phenomenon. Extension results for other overdetermined systems need their own operators and theorems; the frozen seed's PDE setting is not automatically an example of this exact statement.[3]

Scope of Application

The theorem applies to compact holes in domains of ℂⁿ for n≥2, provided the complement stays connected. A removed point gives the local isolated-singularity consequence. A closed ball removed from a larger complex ball gives a shell whose holomorphic functions are forced inward. The hole's geometry can vary; compactness, dimension, connectedness and holomorphy are the load-bearing conditions.[1][2]

Several proof styles exist, including Cauchy-integral constructions and solutions to a compact-support ∂̄ problem. Those routes illuminate why the result holds but do not add new hypotheses or turn the statement into a universal PDE extension principle. Generalizations to complex manifolds or singular spaces require additional assumptions and should be cited under their own theorems.[2][4]

Clarity

The theorem separates absence of a formula inside K from freedom to choose values inside K. The former is present by definition: f is given only on Ω\K. The latter is absent under the theorem's hypotheses: any holomorphic continuation must be the same F. One-variable intuition suggests a puncture can hide a pole, but in several variables such an isolated compact obstruction cannot persist.[1][2]

It also clarifies why “singular set” needs care. Under a local punctured-neighborhood setup, an isolated singularity is removable. It does not follow that a function with arbitrary noncompact exceptional set extends, nor that all other analytic or PDE systems obey identical rules.

Manages Complexity

Without the theorem, one might analyze each shape of hole and each candidate function separately to ask whether a continuation exists. The hypothesis checklist compresses those cases into a single result: once dimension, compactness, connectedness and holomorphy are established, existence and uniqueness follow.[1]

The compression is exact, not vague. The theorem's force comes from analytic rigidity in several complex variables, while the excluded cases mark where separate analysis is needed. Replacing the checklist with “high-dimensional holes always disappear” would erase the boundary information that makes the result useful.

Abstract Reasoning

Given f on a punctured or shelled region, identify Ω, K and the complex dimension. Verify K is compactly inside Ω, Ω\K is connected, and f is holomorphic there. Then infer a unique holomorphic F on the whole domain. For an isolated point in ℂ², these conditions hold locally, so the missing value is determined by the continuation rather than assignable at will.[1]

Counterfactual checks prevent overreach. Change n to one and 1/z defeats extension. Change the excluded set to a noncompact complex hypersurface and a genuine pole-like behavior can survive. Drop connectedness and two pieces of data may be incompatible. None of these cases is refuted by the theorem because each removes a required condition.[2]

Knowledge Transfer

Within several complex variables, the same compact-hole argument applies to punctured balls and connected shells in different dimensions n≥2. The theorem also motivates broader questions about domains of holomorphy and extension across exceptional sets. Those are consequences or research directions, not literal copies of the theorem unless their exact hypotheses match.[2]

The phrase Hartogs phenomenon appears in more general analytic and PDE settings, but transferring this theorem's conclusion requires a new proof for the new operator. The broad pattern of rigidity may be analogous; the precise theorem remains domain-specific and is not promoted to a prime on that basis.[3]

Examples

Punctured ball in two complex variables

Take Ω={(z,w)∈ℂ²: |z|²+|w|²<1} and K={(0,0)}. Define the worked input only on Ω\K by f(z,w)=z²+w. The punctured ball is connected, and f is holomorphic there; the forced extension is F(z,w)=z²+w on all of Ω, with F(0,0)=0. The formula makes this particular continuation easy to see, while the theorem gives the same conclusion for any holomorphic input on this connected punctured domain, not just a polynomial chosen to be transparent.[1][2]

Mapped back: domain = unit ball in ℂ²; hole = compact point (0,0); complement = connected punctured ball; input = z²+w restricted to it; output = unique F=z²+w on the full ball, assigning value zero at the omitted point.

Holomorphic function on a shell in three complex variables

For a genuinely different hole, take Ω={(z,w,u)∈ℂ³: |z|²+|w|²+|u|²<4} and K={(z,w,u): |z|²+|w|²+|u|²≤1}. On the connected shell 1<|z|²+|w|²+|u|²<4, let g(z,w,u)=e^z+wu. It is holomorphic there and has the explicit extension G(z,w,u)=e^z+wu across the entire closed inner ball, with G(0,0,0)=1. Although this chosen formula displays the answer, the theorem's force is that a shell-defined holomorphic function cannot specify an incompatible interior value. The shell is not a one-variable annulus on which a 1/z term could persist.[1]

Mapped back: domain = radius-two ball in ℂ³; hole = compact closed radius-one ball; complement = connected shell; input = restriction of e^z+wu; output = unique G=e^z+wu on all of Ω, including G(0,0,0)=1.

Near miss: one complex variable

Take Ω=ℂ, K={0} and f(z)=1/z. The complement is connected and the hole compact, but n=1; f has a pole and no holomorphic extension through zero. This demonstrates why dimension is constitutive, not optional.[2]

Structural Tensions

This theorem does not present an intrinsic design tradeoff with two benefits and two costs inside a valid instance. Holomorphy, n≥2, compact K⊂Ω, and connected Ω\K are hypotheses, not a balance to optimize. The real analytical tension is between one's one-variable intuition that a deleted set might support a pole and the several-variable rigidity that forbids it here; that is an epistemic contrast, not a structural opposed-cost choice. Diagnostic: before invoking extension, verify each hypothesis and ask whether a purported singular term is actually holomorphic on the entire complement.[1][2]

Likewise, calling an extension for a different operator “Hartogs-like” does not spend one property to gain another; it changes the theorem being used. Such transfers require independent proof and belong to the boundary, not to an invented second intrinsic tension.[3]

Structural–Framed Character

The theorem is highly structural within several complex variables: a small set of topological and analytic conditions forces a unique extension. Its vocabulary travels among punctured balls, shells and domains in ℂⁿ, but using the name for a generic missing-data or PDE problem would import complex-analytic conditions not already present. Its proof and application require mathematical practice but not an institution's prescription. Evaluative weight is essentially absent: “removable” is a mathematical property, not a preference that holes ought to disappear. The portable skeleton of constraint-driven completion could recur elsewhere, yet the named theorem's force is the specific analytic guarantee. Its character: strongly structural in complex analysis, but mathematically domain-specific rather than prime.

Structural Core vs. Domain Accent

The skeleton is a forced unique completion from data outside an omitted interior region when strong compatibility constraints hold. The mechanism is holomorphic rigidity in at least two complex variables with compact hole and connected complement. Without that mechanism the skeleton is merely a completion analogy, not Hartogs's theorem. The named entry does not clear the prime bar because each hypothesis is mathematically essential and one complex variable already supplies a counterexample. A broader completion or analytic-rigidity prime would need separate cross-domain evidence; none is silently asserted here.

DAG placement. No verified live extension-theorem genus covers this exact statement. Analytic Function names the quantified object, not the theorem; Continuity is a related property, not the necessary genus. The live Ohsawa–Takegoshi L² Extension Theorem has different data and hypotheses. Dimension, domain and deleted-set conditions remain load-bearing, so the result does not extend by name to one-variable holes or unrelated PDEs.

Neighborhood in Abstraction Space

Hartogs's Extension Theorem sits in a sparse region of the domain-specific corpus (75th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Complex & Holomorphic Function Theory (8 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With

Hartogs's extension theorem is not Hartogs's theorem on separate holomorphicity, not an unqualified assertion that all higher-dimensional singularities disappear, and not the Ohsawa–Takegoshi theorem. The one-variable 1/z counterexample and noncompact excluded sets delimit the result. General Hartogs-type PDE extensions require their own proof and are not counted as examples of the compact-hole theorem.

References

[1] UC San Diego, Lecture 3, “Hartog's Extension Theorem”, exact domain, compact-hole and connected-complement statement. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l

[2] Jiří Lebl, “Several Complex Variables are Better than One”, Hartogs phenomenon, one-variable counterexample and proof outline. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m

[3] “A Geometrical Proof of the Hartogs Extension Theorem”, original alternate-proof abstract and historical scope. registry ↩a ↩b ↩c

[4] “Elementary Approach to the Hartogs Extension Theorem”, original proof-paper abstract. registry ↩