Abstract:It is a classical observation that lacunary function systems exhibit many properties which are typical for systems of independent random variables. However, it had already been observed by Erdős and Fortet in the 1950s that probability theory's limit theorems may fail for lacunary sums if the sequence has a strong arithmetic "structure". The presence of such structure can be assessed in terms of the number of solutions of two-term linear Diophantine equations . As the first author proved with Berkes in 2010, saving an (arbitrarily small) unbounded factor for the number of solutions of such equations compared to the trivial upper bound, rules out pathological situations as in the Erdős–Fortet example, and guarantees that satisfies the central limit theorem (CLT) in a form which is in accordance with true independence. In contrast, as shown by the first author, for the law of the iterated logarithm (LIL) the Diophantine condition which suffices to ensure "truly independent" behavior requires saving this factor of logarithmic order. In the present paper we show that, rather surprisingly, saving such a logarithmic factor is actually the optimal condition in the LIL case. This result reveals the remarkable fact that the arithmetic condition required of to ensure that shows "truly random" behavior is a different one at the level of the CLT than it is at the level of the LIL: the LIL requires a stronger arithmetic condition than the CLT does.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is, in the application of an important theorem in probability theory - the Law of the Iterated Logarithm (LIL) to lacunary systems, what Diophantine conditions can ensure that these systems exhibit behavior similar to that of independent random variables.
Specifically, the paper explores under what conditions the lacunary function sum of the form \(\sum_{k = 1}^N f(n_kx)\) can satisfy the Law of the Iterated Logarithm. Here, \(f\) is a function with a period of 1, and \((n_k)_{k\geq1}\) is a sequence of natural numbers that satisfies the Hadamard gap condition, that is, there exists \(q>1\) such that \(n_{k + 1}/n_k\geq q\) for all \(k\in\mathbb{N}\).
### Main problems:
1. **Optimality of Diophantine conditions**: A core problem in the paper is to determine the optimal Diophantine conditions required for the lacunary function sum \(\sum_{k = 1}^N f(n_kx)\) to exhibit "truly independent" behavior within the LIL framework. Specifically, the authors prove that in order to ensure that the LIL holds, the required Diophantine conditions are stronger than those required for the Central Limit Theorem (CLT).
2. **Differences at different scales**: Another key problem is to explore how Diophantine conditions change at different deviation scales. The paper points out that at the CLT scale, weaker Diophantine conditions can ensure that the behavior of the lacunary function sum is similar to that of independent random variables; while at the LIL scale, stronger Diophantine conditions are required to ensure this behavior.
### Specific results:
- **CLT conditions**: For the CLT, if the number of solutions \(L(N, a, b, c)=o(N)\) of the Diophantine equation \(an_k - bn_\ell=c\) holds uniformly for all \(c\in\mathbb{Z}\setminus\{0\}\), then \(\sum_{k = 1}^N f(n_kx)\) satisfies the CLT.
- **LIL conditions**: For the LIL, if the number of solutions \(L(N, a, b, c)=O\left(\frac{N}{(\log N)^{1+\varepsilon}}\right)\) of the Diophantine equation \(an_k - bn_\ell=c\) holds uniformly for all \(c\in\mathbb{Z}_{\geq0}\), then \(\sum_{k = 1}^N f(n_kx)\) satisfies the LIL.
### Main contributions of the paper:
- **Proof of optimality**: The paper proves that the \((\log N)^{1+\varepsilon}\) factor in the above - mentioned LIL conditions is the best, that is, it cannot be further weakened. This reveals an interesting phenomenon: at the LIL scale, the required arithmetic conditions are stronger than those at the CLT scale.
- **Theoretical explanation**: The authors show, by constructing specific examples, why stronger Diophantine conditions are required at the LIL scale and explain the connection between these conditions and the tail behavior of the normal distribution.
### Conclusion:
Through rigorous mathematical analysis and constructive examples, this paper reveals the differences in arithmetic conditions for lacunary function sums within different probability limit theorem frameworks (such as CLT and LIL) and the underlying theoretical mechanisms. This research not only deepens our understanding of the probabilistic properties of lacunary systems but also provides an important theoretical basis for further research in related fields.