A Dense Model Theorem for the Boolean Slice

Gil Kalai,Noam Lifshitz,Dor Minzer,Tamar Ziegler

2024-08-01

Abstract:The (low soundness) linearity testing problem for the middle slice of the Boolean cube is as follows. Let $\varepsilon>0$ and $f$ be a function on the middle slice on the Boolean cube, such that when choosing a uniformly random quadruple $(x,y,z ,x\oplus y\oplus z)$ of vectors of $2n$ bits with exactly $n$ ones, the probability that $f(x\oplus y \oplus z) = f(x) \oplus f(y) \oplus f(z)$ is at least $1/2+\varepsilon$. The linearity testing problem, posed by David, Dinur, Goldenberg, Kindler and Shinkar, asks whether there must be an actual linear function that agrees with $f$ on $1/2+\varepsilon'$ fraction of the inputs, where $\varepsilon' = \varepsilon'(\varepsilon)>0$. We solve this problem, showing that $f$ must indeed be correlated with a linear function. To do so, we prove a dense model theorem for the middle slice of the Boolean hypercube for Gowers uniformity norms. Specifically, we show that for every $k\in\mathbb{N}$, the normalized indicator function of the middle slice of the Boolean hypercube $\{0,1\}^{2n}$ is close in Gowers norm to the normalized indicator function of the union of all slices with weight $t = n\pmod{2^{k-1}}$. Using our techniques we also give a more general `low degree test' and a biased rank theorem for the slice.

Combinatorics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the performance of the linearity testing problem in the case of low soundness on the middle slice of the Boolean cube. Specifically, when a function \( f \) satisfies \( f(x \oplus y \oplus z) = f(x) \oplus f(y) \oplus f(z) \) with a probability of at least \( \frac{1}{2}+\varepsilon \) for uniformly randomly selected quadruples \((x, y, z, x \oplus y \oplus z)\), does there exist an actual linear function that agrees with \( f \) on \( \frac{1}{2}+\varepsilon' \) proportion of the inputs, where \( \varepsilon'=\varepsilon'(\varepsilon)>0 \).

The main result of the paper is to prove that such an \( f \) is indeed associated with some linear function. To achieve this, the authors prove a dense model theorem for the middle slice of the Boolean hyper - cube, especially a theorem regarding Gowers uniformity norms. Specifically, they show that for any \( k\in\mathbb{N} \), the normalized indicator function of the middle slice is close in the Gowers norm to the normalized indicator function of the union of all slices with weight \( n\ mod\ (2^{k}-1) \). Through these techniques, they also present a more general "low - degree test" and a biased rank theorem.

### Main Contributions
1. **Linearity Testing Problem**: Solved the linearity testing problem on the middle slice of the Boolean cube in the case of low soundness.
2. **Dense Model Theorem**: Proved a dense model theorem for the middle slice of the Boolean hyper - cube, especially applicable to Gowers uniformity norms.
3. **Low - Degree Test**: Provided a more general low - degree test method applicable to higher - order polynomials.
4. **Biased Rank Theorem**: Studied the rank problem of low - degree polynomials in the biased case on the middle slice of the Boolean cube.

### Method Overview
- **Dense Model Theorem**: Use the dense model theorem to transform the problem of the middle slice of the Boolean hyper - cube into the problem of the Boolean hyper - cube.
- **Gowers Uniformity Norms**: Use Gowers uniformity norms to measure the uniformity of functions.
- **Fourier Analysis**: Use Fourier analysis techniques to handle the properties of Boolean functions.

### Applications
- **Low - Degree Test**: Extended the results of linear testing to higher - order polynomials.
- **Biased Rank Problem**: Studied the rank problem of low - degree polynomials in the biased case.

Through these methods, the paper not only solves the linearity testing problem but also provides new tools and perspectives for other related problems.