Glicksberg's theorem¶
A continuous zero-sum game result guaranteeing equality of maximin and minimax expected payoff over Borel mixed strategies on compact Hausdorff strategy spaces under the stated semicontinuity condition.
Core Idea¶
Glicksberg's theorem extends minimax reasoning to certain continuous zero-sum games. Each player's pure strategies lie in a compact Hausdorff space, mixed strategies are Borel probability measures, and expected payoff is obtained by integrating the payoff function over both measures.
Under the theorem's semicontinuity condition, the supremum of one player's guaranteed payoff equals the infimum of the other's worst-case bound, so the game has a mixed-strategy value. The hypotheses do real work: the source explicitly warns that the continuity requirement cannot simply be omitted.
Structural Signature¶
Sig role-phrases:
- compact strategy spaces. Provide Hausdorff compact pure-strategy domains. Constitutive topological hypothesis. If altered: Noncompact escape can destroy attainment or value arguments.
- semicontinuous payoff. Controls limiting payoff behavior on the product space. Constitutive regularity hypothesis. If altered: Arbitrary discontinuity can produce no value.
- Borel mixed strategies. Convexify choices as probability measures. Identity-bearing strategy extension. If altered: Pure strategies alone may lack a saddle value.
- expected payoff integral. Extends K bilinearly to mixed choices. Constitutive evaluation. If altered: An undefined integral breaks the stated game.
- minimax equality. Equates the players' guaranteed expected-payoff bounds. Theorem conclusion. If altered: The equality is not an extra assumption.
What It Is Not¶
- Von Neumann minimax theorem. Are finite or convex-set hypotheses being used?
- Nash existence theorem. Is a non-zero-sum equilibrium rather than a game value sought?
- Sion minimax theorem. Which convexity and semicontinuity assumptions apply?
- Pure saddle point. Are mixed probability measures necessary?
Scope of Application¶
Use the theorem only after matching every topological, measure, payoff, and zero-sum hypothesis to the stated version.
- Continuous games. Establishes a value beyond finite matrices.
- Mixed strategies. Uses probability measures over compact spaces.
- Minimax analysis. Equates guaranteed expected payoffs.
- Economic models. Handles continuous action sets when assumptions hold.
- Counterexamples. Shows why discontinuity matters.
Clarity¶
The equation is a conclusion about optimized expected payoffs, not a pointwise interchange of sup and inf. The order of quantifiers and the passage from pure points to measures must remain visible.
Manages Complexity¶
Topology, integration, optimization, and strategic opposition meet in one result. Compactness controls escape, semicontinuity controls limits, and mixing supplies convexity-like structure; dropping one can invalidate the equality.
Abstract Reasoning¶
- Verify zero-sum payoff orientation.
- Check both pure-strategy spaces for compact Hausdorff structure.
- Match the exact upper or lower semicontinuity version.
- Define Borel probability measures and the expected payoff integral.
- Only then infer equality of maximin and minimax values.
Knowledge Transfer¶
The hypothesis-audit pattern transfers to existence theorems. The conclusion does not transfer to discontinuous, noncompact, non-zero-sum, or differently measurable games without a separate theorem. The nearest stopping boundary is explicit: Von Neumann's minimax theorem is closest: it covers finite or convex settings under different hypotheses, while Glicksberg addresses compact topological strategy spaces with measure-valued mixing. The inclusion test remains: A theorem application qualifies when the game is zero-sum, pure-strategy spaces meet the compact Hausdorff hypothesis, payoff has the required semicontinuity, and mixed payoffs use Borel measures. The structure no longer applies when the case exits when a required topological or regularity assumption fails or when the game is not zero-sum.
Examples¶
Canonical¶
Two players choose points in compact Hausdorff spaces, a bounded payoff meets the stated semicontinuity condition, and each may randomize with a Borel probability measure; the integrated game receives a minimax value.
Mapped back: compact strategy spaces → A and B; semicontinuous payoff → K on A×B; Borel mixed strategies → f and g; expected payoff integral → double integral of K; minimax equality → sup-inf equals inf-sup.
Applied / In Practice¶
A proposed continuous game has a jump payoff that violates the required regularity; the analyst withholds the theorem rather than inferring value from compact action sets alone.
Mapped back: compact strategy spaces → present; semicontinuous payoff → failed; Borel mixed strategies → available but insufficient; expected payoff integral → defined; minimax equality → not guaranteed.
Structural Tensions¶
T1: general strategy spaces vs. strong regularity. The theorem broadens actions beyond finite sets while retaining compactness and semicontinuity. Diagnostic: Which hypothesis supplies the missing control?
T2: pure discontinuity vs. mixed convexification. Randomization helps but cannot repair every payoff pathology. Diagnostic: Does the exact theorem cover this discontinuity?
Structural–Framed Character¶
Description turns on compact strategy spaces, semicontinuous payoff, Borel mixed strategies, expected payoff integral, minimax equality. Skeletal core. Regularity and compactness let opposing optimization commute after choices are convexified by probability measures. Domain-bound accent. Zero-sum games, Borel measures, semicontinuity, Hausdorff compactness, and expected payoff define the theorem. Transfer remains bounded because Why not prime. Minimax structure travels; this node is one named theorem with exact hypotheses. The negative boundary is concrete: Any finite minimax result, Nash equilibrium theorem, compact optimization claim, or continuous game is not automatically Glicksberg's theorem. Glicksberg's theorem is structural: its spaces, measures, integrals, hypotheses, and equality are formal. Its character: a compact-topological mixed-strategy minimax existence theorem.
Structural Core vs. Domain Accent¶
Skeletal core. Regularity and compactness let opposing optimization commute after choices are convexified by probability measures.
Domain-bound accent. Zero-sum games, Borel measures, semicontinuity, Hausdorff compactness, and expected payoff define the theorem.
Why not prime. Minimax structure travels; this node is one named theorem with exact hypotheses.
Instantiates / Related Primes¶
This entry is a kind of Minimax Theorem.
- Minimax. The conclusion equates opposed guaranteed bounds.
- Existence. Compactness and regularity prevent limiting failure.
- No strict parent is asserted.
Relationships to Other Abstractions¶
Current abstraction Glicksberg's theorem Domain-specific
Parents (1) — more general patterns this builds on
-
Glicksberg's theorem is a kind of Minimax Theorem Domain-specific
Glicksberg's theorem is a named member of the minimax theorem family, extending it to continuous games on compact Hausdorff strategy spaces.The minimax theorem is a family of results giving hypotheses under which opposed max-min and min-max values coincide, certifying a saddle value. Glicksberg's theorem is exactly such a member: under compactness of the (Borel, mixed-strategy) strategy spaces and a stated semicontinuity condition on the continuous zero-sum payoff, the maximin and minimax expected payoffs coincide. The differentia is the specific hypothesis set (compact Hausdorff spaces, Borel measures, semicontinuity) that lets the general minimax-equality pattern go through in the continuous setting. The equality holds whenever the stated hypotheses hold, so the qualifier is strict.
Hierarchy path (1) — routes to 1 parentless root
- Glicksberg's theorem → Minimax Theorem → Minimax Strategy → Optimization
Neighborhood in Abstraction Space¶
Glicksberg's theorem sits in a crowded region of the domain-specific corpus (35th percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.
Family — Game-Theoretic Models & Paradoxes (37 abstractions)
Nearest neighbors
- Mixed Strategy Equilibrium — 0.90
- Matching pennies — 0.88
- Bayesian Nash Equilibrium — 0.88
- Rubinstein bargaining model — 0.88
- Cognitive Hierarchy Theory — 0.88
Computed from structural-signature embeddings · 2026-10-08
Not to Be Confused With¶
- Von Neumann minimax theorem. Tell: Are finite or convex-set hypotheses being used?
- Nash existence theorem. Tell: Is a non-zero-sum equilibrium rather than a game value sought?
- Sion minimax theorem. Tell: Which convexity and semicontinuity assumptions apply?
- Pure saddle point. Tell: Are mixed probability measures necessary?
References¶
- Frozen Wikipedia discovery revision: https://en.wikipedia.org/wiki/Glicksberg%27s_theorem (revision 1350256736).
- Preserved source candidate: https://doi.org/10.2307/2032478
The frozen Wikipedia revision is discovery provenance. The retained source set was reviewed for identity, formal or operational relation, and scope. The encyclopedia's structural synthesis is bounded to those claims; a thin authority surface is recorded as a nonblocking source-strengthening repair rather than concealed.