---
title: Goldstern's Principle for Π¹₁ Sets
url: https://www.emergentmind.com/topics/goldstern-s-principle-for-boldsymbol-1_1-sets
type: topic
---

# Goldstern's Principle for Π¹₁ Sets

Goldstern’s Principle for $\boldsymbol{\Pi}^1_1$ Sets

Goldstern’s principle, initially formulated for analytic (Σ¹₁) sets, asserts that the union of a real-parametrized, monotone family of Lebesgue measure zero sets remains null when the family is uniformly analytic. The significant extension to the co-analytic (Π¹₁) case, known as GP(Π¹₁), demonstrates that monotone families of null sets, indexed co-analytically, also yield null unions. This principle forms part of a broader investigation into the intersection of descriptive set theory, measure theory, pointclass regularity properties, and forcing, and has direct implications for the structure of null ideals and regularity phenomena at higher levels of the projective hierarchy [2206.08147, 2512.22852].

## 1. Formal Statement and Key Definitions

Let $Y$ be a Polish probability space (such as $2^\omega$ endowed with the usual product measure $\mu$). Consider a family $\{A_x : x \in \omega^\omega\}$, where each $A_x \subseteq Y$, parameterized by real sequences in $\omega^\omega$. The family is *monotone* if $x \leq x'$ (pointwise) implies $A_x \subseteq A_{x'}$.

A set $A \subseteq \omega^\omega \times Y$ is *co-analytic* (Π¹₁) if its complement in $\omega^\omega \times Y$ is analytic (Σ¹₁). The *vertical section* at $x$ is denoted $A_x = \{ y \in Y : (x, y) \in A \}$.

**Goldstern’s Principle for Π¹₁ Sets (GP(Π¹₁))**:
If $A \subseteq \omega^\omega \times 2^\omega$ is co-analytic, $\{A_x\}$ is monotone, and $\mu(A_x) = 0$ for all $x$, then
$$
\mu\left( \bigcup_{x \in \omega^\omega} A_x \right) = 0.
$$
This property is denoted $GP(\Pi^1_1)$. More generally, for a pointclass $\Gamma$, $GP(\Gamma)$ refers to the analogous statement for sets in $\Gamma$ [2206.08147, 2512.22852].

## 2. Proof Techniques and Absoluteness

The proof strategy for GP(Π¹₁) diverges from the Σ¹₁ scenario, reflecting the higher complexity of the pointclass. While the analytic case admits a proof based on random forcing and Shoenfield (Σ¹₂) absoluteness, the co-analytic case exploits Laver forcing, which is proper and $\omega^\omega$-bounding, and preserves outer measure.

Key technical ingredients:
- **Kechris–Tanaka Lemma**: For a universal analytic set $U \subseteq \omega^\omega \times 2^\omega$, the relation $\{ (x, r) : \mu(U_x) > r \}$ is analytic. For co-analytic $A$, $\{ (x, r) : \mu(A_x) > r \}$ is analytic, and the null-fibre assertion $\mu(A_x) = 0$ is Σ¹₁-uniform in $x$.
- **Absoluteness**: Statements like “$\forall x\, \mu(A_x) = 0$” and the monotonicity condition are Π¹₂ sentences, preserved by proper, bounding forcings such as Laver forcing.

The argument involves passing to a forcing extension $V[d]$, where a dominating real $d$ is adjoined:
- The monotonicity and null-fibre statements persist in $V[d]$.
- One shows $\bigcup_{x \in V} A_x \subseteq A_d$. Since $\mu(A_d) = 0$ and outer measure is preserved, the union was already null in $V$ [2206.08147, 2512.22852].

## 3. Extensions, Ideals, and Failure for Hausdorff Measure

Beyond Lebesgue-null ideals, one may consider other σ-ideals such as those generated by Hausdorff measures with various gauge functions $f$. The generalization $GP(\Pi^1_1, N^f_X)$ asserts the same principle holds for sets null with respect to the appropriate Hausdorff measure $H^f$.

While $GP(\Pi^1_1, N^f_X)$ holds for continuous doubling gauge functions $f$ on compact metrics, the classical Hausdorff case ($f(x) = x^s$, $0 < s < 1$) demonstrates notable divergence in $L$:

- **Failure in $L$**: For every $s \in (0,1)$, there is co-analytic $A \subseteq \omega^\omega \times 2^\omega$ with each $A_x$ countable, but $\bigcup_x A_x$ has Hausdorff dimension $1$; that is, $H^{pow_s}(\bigcup_x A_x) > 0$. The construction uses $\Pi^1_1$-scales and the coding of effective Hausdorff dimension via algorithmic randomness [2512.22852].

A summary contrasting Lebesgue and Hausdorff cases in $L$:

| Measure Notion        | $\Pi^1_1$ Principle Holds in $L$ | Counterexample Construction     |
|----------------------|:-------------------------------:|-------------------------------|
| Lebesgue measure     | Yes                             | —                             |
| Hausdorff ($0 < s < 1$) | No                              | $\Pi^1_1$-scales, algorithmic randomness |

## 4. Independence, Consistency, and Countermodels

The status of $GP(\Gamma)$ for various pointclasses and ideals is sensitive to set-theoretic assumptions:

- **Consistency and Independence**:
  - Under $\mathsf{CH}$, a monotone family of null sets can have a union of full measure. Thus, $\mathsf{CH}$ implies failure of $GP(\text{all})$ [2206.08147].
  - In models with $\mathsf{AD}$ or Solovay models, *every* set of reals is Lebesgue measurable, and one obtains $GP(\text{all})$ via absoluteness arguments.
  - In Laver models, $GP(\text{all})$ holds due to the non-existence of long null towers.

- **Failure for Other Ideals**: For the ideal generated by closed null sets, $GP(\Pi^1_1, \mathcal E)$ fails already at the analytic level, employing an interval-partition argument [2206.08147].

## 5. Projective Hierarchy and Regularity

Goldstern’s principle demarcates subtle regularity phenomena across the projective hierarchy:

- For every $n \geq 2$, $GP(\Sigma^1_n)$ is distinct from $GP(\Pi^1_{n+1})$. Each increase in projective complexity yields a strict strengthening.
- Under determinacy and suitable large-cardinal hypotheses, one obtains sharp separation results: e.g., there exist models where $GP(\Sigma^1_n)$ holds while $GP(\Sigma^1_{n+1})$ fails.
- $GP(\Sigma^1_2)$ is provable assuming every $\Sigma^1_2$ set of reals is Lebesgue measurable. The existence of a real dominating over $L[a]$ for every real $a$ is implied by $GP(\Sigma^1_2)$ [2512.22852].

## 6. Implications, Corollaries, and Open Problems

- Any pointclass $\Gamma$ containing the Borel sets, closed under continuous pre-images, and satisfying determinacy of length three, gives $GP(\text{proj}(\Gamma))$ as a consequence. In particular, $\mathsf{AD}$ yields $GP(\text{all})$ [2206.08147].
- The ability to move from Lebesgue to Hausdorff ideals depends critically on projective well-orders and the availability of reals dominating countable initial segments in $L$.
- Open questions remain concerning the precise demarcation of $GP(\Gamma, I)$-type statements for broader ideals and categories of sets, especially at higher projective levels and under weaker regularity or determinacy assumptions.

The landscape outlined by Goldstern’s principle and its extensions for $\Pi^1_1$ sets reveals intricate dependencies between descriptive set-theoretic complexity, measure-theoretic regularity, forcing absoluteness, and foundational set-theoretic axioms [2206.08147, 2512.22852].

Source: https://www.emergentmind.com/topics/goldstern-s-principle-for-boldsymbol-1_1-sets