Hausdorff Dimension Games

• Hausdorff dimension games are two-player strategy games that characterize fractal sets by using covering arguments and set-theoretic operations.
• The games establish an equivalence between winning strategies and the Hausdorff dimension threshold, yielding powerful regularity and compactness results.
• They integrate classical Hausdorff measures with modern descriptive set theory and Schmidt-type games, informing optimal strategies and further developments in fractal geometry.

Hausdorff dimension games are a class of two-player games constructed to characterize and analyze the Hausdorff dimension of subsets in metric or topological spaces. Stemming from the foundational work of Das, Fishman, Simmons, and Urbański, they model the interplay between covering arguments and set-theoretic operations, providing both theoretical and constructive frameworks for dimension theory. These games connect classical Hausdorff measure and more recent developments in descriptive set theory, determinacy, and fractal geometry, often enabling sharp geometric and regularity statements for analytic, Borel, or arbitrary sets under strong set-theoretic assumptions.

1. Definitions and Variants of Hausdorff-Dimension Games

At the core is the original δ-dimension game $G^{\delta}_\beta(A)$ played with parameters $0 < \beta < 1/2$ and initial scale. Let $A \subset \mathbb{R}^d$ be a target set. At stage $n$, Player I selects a finite, $3\rho_n$-separated set $F_n \subset \mathbb{Q}^d$ with $F_n \subset B(x_{n-1}, (1-\beta)\rho_{n-1})$ (for $n>0$), $\rho_n = \beta\rho_{n-1}$, and an asymptotic "size rule":

$$\prod_{i < n}|F_i|^{-1}\cdot \beta^{\delta n} \leq c \quad \text{for some }c<\infty \text{ and all large }n.$$

Player II then chooses $0 < \beta < 1/2$0. The outcome is the (possibly unique) limit point $0 < \beta < 1/2$1. Player II wins if $0 < \beta < 1/2$2, otherwise Player I wins.

A more flexible version, $0 < \beta < 1/2$3, uses a sequence $0 < \beta < 1/2$4 satisfying suitable decay conditions (e.g., $0 < \beta < 1/2$5 for any $0 < \beta < 1/2$6 eventually) and adapts the size rule and separation accordingly.

These games can be unfolded further: given a relation $0 < \beta < 1/2$7 and $0 < \beta < 1/2$8, Player I can interleave "witness" moves spelling out $0 < \beta < 1/2$9; Player II wins if $A \subset \mathbb{R}^d$0. This unfolded game supports uniformization and additivity results within models admitting the Axiom of Determinacy (AD) (Crone et al., 2020).

2. Main Characterization Theorems

The fundamental result is the equivalence between determinacy of the game and the Hausdorff dimension threshold:

• Player II has a winning strategy in $A \subset \mathbb{R}^d$1 $A \subset \mathbb{R}^d$2 $A \subset \mathbb{R}^d$3.
• Player I has a winning strategy $A \subset \mathbb{R}^d$4 $A \subset \mathbb{R}^d$5 (Crone et al., 2020).

For $A \subset \mathbb{R}^d$6 with decaying $A \subset \mathbb{R}^d$7, the characterization remains valid, and in fact Player II's victory yields the existence of a compact subset $A \subset \mathbb{R}^d$8 with $A \subset \mathbb{R}^d$9—a strong regularity property.

The unfolded games, combining coordinate and witness plays, preserve this threshold: a winning strategy for Player II in the unfolded game translates into an ordinary $n$0-game victory for any $n$1.

3. Unfolding Principle and Its Consequences

The unfolded δ-dimension game allows Player I to "spell out" an independent witness sequence $n$2 in addition to controlling covers, modeling projection and uniformization problems. The key combinatorial lemma ensures any large enough family with $n$3 linear orders has an element whose initial segments are large in each order, providing the backbone for simulating multiple witnesses in parallel.

The unfolding theorem asserts: if Player II has a winning strategy in the unfolded δ-game for relation $n$4, then she can win in the ordinary $n$5-game for every $n$6 on $n$7 (Crone et al., 2020). This result is critical for transferring regularity properties from pairs $n$8 to projections and domains.

4. Applications in Regularity and Additivity

Under the Axiom of Determinacy (AD), Hausdorff-dimension games underlie several central dimension-theoretic results:

• Continuous uniformization: For $n$9 with $3\rho_n$0, for any $3\rho_n$1, there exists a compact $3\rho_n$2, $3\rho_n$3, and a continuous $3\rho_n$4 with $3\rho_n$5 for all $3\rho_n$6.
• Full additivity: Any well-ordered union $3\rho_n$7 with $3\rho_n$8 for each $3\rho_n$9, but $F_n \subset \mathbb{Q}^d$0, leads to a contradiction. Thus under AD, the supremum of the dimensions of a well-ordered family bounds the dimension of the union.
• Compact sets of nearly full dimension: Any analytic $F_n \subset \mathbb{Q}^d$1 of Hausdorff dimension $F_n \subset \mathbb{Q}^d$2 contains compact $F_n \subset \mathbb{Q}^d$3 with $F_n \subset \mathbb{Q}^d$4 for any $F_n \subset \mathbb{Q}^d$5 (Crone et al., 2020).

5. Generalizations and Connections to Schmidt-Type Games

The Hausdorff-dimension game is tightly related to Schmidt games and their numerous variants (classical, potential, matrix-potential, rapid) (Bellaïche et al., 13 Jan 2026, Broderick et al., 2017, Howat et al., 15 Aug 2025, Hatefi et al., 2024). All of these games share the principle of two players alternately refining covers while controlling geometric or measure-theoretic complexity.

Recent work has unified these tree-structured games under a game-theoretic covering model: the minimum Hausdorff dimension necessary for a "win" can be precisely recovered from the covering rates allowed to Player I (Bellaïche et al., 13 Jan 2026). In finite-branching settings, the value of the dimension game coincides with the Hausdorff dimension of the target set. These connections yield transparent proofs for sharp winning thresholds in alternating-move games in terms of Hausdorff dimension, e.g., $F_n \subset \mathbb{Q}^d$6 implies a Player II win in the infinite binary digit game.

6. Comparison with Other Dimension and Measure Games

Hausdorff-dimension games possess several features that make them distinct:

• Sharpness: The size-rule calibrates precisely to the δ-dimensional Hausdorff scaling, avoiding excess or loss in dimension estimates.
• Projective and analytic regularity: For pointclasses such as analytic or well-ordered unions under AD, these games yield dimension-regularity properties not accessible purely from classical covering arguments.
• Unfolding and uniformization: The ability to force witnesses through auxiliary coordinates underlines the utility of these games in projection and selection problems.
• Robustness to coding: Results extend from $F_n \subset \mathbb{Q}^d$7 to Polish spaces and to relations modeling descriptive set-theoretic complexity.

7. Open Problems and Directions

Several further avenues are open for Hausdorff-dimension games:

• Optimal strategies and constructive bounds: While the above results provide existence, quantifying the minimal cover-branching or separation needed for optimal strategies remains of interest.
• Extensions to other fractal dimensions: The extension of characterization theorems to packing, box, or Assouad dimensions within the same framework may yield finer dichotomies.
• Descriptive set-theoretic hierarchies: The interface with determinacy, Borel complexity, and higher projective levels offers a rich field for dimension games and uniformization phenomena.
• Connection to potential and matrix-potential games: Integrating anisotropic scales and Alice/Bob deletion mechanisms points toward a more unified synthesis of geometric fractal games over general metric structures (Howat et al., 15 Aug 2025).

Hausdorff dimension games thus provide a fundamental and flexible toolkit for both the structural and quantitative analysis of fractal geometry, with deep implications for descriptive set theory and geometric measure theory (Crone et al., 2020, Bellaïche et al., 13 Jan 2026).