Poisson statistics via the Chinese remainder theorem (Q936542)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Poisson statistics via the Chinese remainder theorem |
scientific article |
Statements
Poisson statistics via the Chinese remainder theorem (English)
0 references
14 August 2008
0 references
Suppose that for each prime \(p\) a subset \(\Omega_p\subset\mathbb Z/p\mathbb Z\) is given. For each squarefree \(q\) use the Chinese remainder theorem to define \(\Omega_q\subset \mathbb Z/q \mathbb Z\) by \(x\in\Omega_q\) if and only if \(x\in\Omega_p\) for each prime \(p\mid q\). Let \(s_q= q/|\Omega_q|\) be the average spacing between elements of \(\Omega_q\), and \(r_q= 1/s_q\) be the probability that a randomly chosen integer belongs to \(\Omega_q\). Let \(Q\) be an infinite set of squarefree positive integers ordered so that \(s_q\to\infty\). The aim of this paper is to establish conditions under which the spacings between consecutive elements of \(\Omega_q\) for \(q\in Q\) become Poisson distributed as \(s_q\to\infty\). Given a vector of integers \({\mathbf h}= (h_1,h_2,\dots, h_{k-1})\) and \(h_0= 0\), define \[ N_k({\mathbf h},\Omega_q)= \#\{t\pmod q: t+ h_i\in \Omega_q,\;0\leq i\leq k-1\}, \] and assume that for each integer \(k\) and prime \(p\) \[ N_k({\mathbf h},\Omega_p)= r^k_p p(1+ O((1- r_p) p^{-\varepsilon})) \] provided that \(0,h_1,\dots, h_{k-1}\) are distinct \(\pmod p\). Theorem 1 asserts that if also \(s_p= p^{(o(1)}\) for all primes \(p\) then the spacings between elements of the sets \(\Omega_q\) for \(q\in Q\) become Poisson distributed as \(s_q\to\infty\). A key step in the proof of this result is to show in Theorem 13 that under certain conditions the \(k\)-level correlation \(R_k(X,\Omega_q)\) for a compact set \(X\subset\{{\mathbf x}\in\mathbb R^{k-1}: 0< x_1<\cdots< x_{k-1}\}\), defined by \[ R_k(X, \Omega_q)= |\Omega_q|^{-1} \sum_{{\mathbf h}\in S_q X\cap\mathbb Z^{k-1}} N_k({\mathbf h},\Omega_q), \] satisfies as \(s_q\to \infty\), \(R_k(X, \Omega_q)= \text{vol}(X)+ o_{X,k}(1)\). Special cases have been considered previously in the literature. By taking \(\Omega_q\) to be the integers coprime to \(q\), the authors recover a result established by \textit{C. Hooley} [Publ. Math. 12, 39--49 (1965; Zbl 0142.29201) and by Math. Z. 90, 355--364 (1965; Zbl 0142.29202)]. \textit{P. Kurlberg} and \textit{Z. Rudnick} in [Duke Math. J. 100, No. 2, 211--242 (1999; Zbl 0985.11038)] and \textit{P. Kurlberg} in [Isr. J. Math. 120, Pt. A, 205--224 (2000; Zbl 1026.11074)] considered the case when \(\Omega_q\) is the set of squares \(\pmod q\). The present authors generalize this in Corollary 2 which states that for a fixed integer \(d\) and \(\Omega_q\) the set of \(d\)th powers \(\pmod q\) the spacings between elements of \(\Omega_q\) become Poisson distributed as \(s_q\to\infty\). Another result in the paper concerns polynomials \(f\in\mathbb Z[x]\) of degree \(n\) such that the set \(\{ f(\xi): \xi\in\mathbb C, f'(\xi)= 0\}\) has exactly \(n- 1\) distinct elements, i.e. \(f\) has \(n-1\) distinct critical values. Let \[ \Omega_q= \{x\,\pmod q: \exists y \pmod q\text{ such that }f(y)\equiv x \pmod q\}. \] Theorem 3 asserts that the spacings between elements of \(\Omega_q\) become Poisson distributed as \(s_q\to\infty\). The authors remark that this result has been extended to all non-constant polynomials by the second author in a separate paper [Int. J. Number Theory 5, No. 3, 489--513 (2009; Zbl 1231.11118)]. Another problem considered concerns determining conditions on the spacings in \(\Omega_{q_1}\) and \(\Omega_{q_2}\) in order that spacings in \(\Omega_{q_1q_2}\) become Poisson distributed, where \((q_1,q_2)= 1\). Some counterexamples are also discussed.
0 references
Poisson spacings
0 references
correlation functions
0 references
distribution modulo one
0 references