Real zeros of random cosine polynomials with palindromic blocks of coefficients (Q822642)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 7399103
Language Label Description Also known as
default for all languages
No label defined
    English
    Real zeros of random cosine polynomials with palindromic blocks of coefficients
    scientific article; zbMATH DE number 7399103

      Statements

      Real zeros of random cosine polynomials with palindromic blocks of coefficients (English)
      0 references
      22 September 2021
      0 references
      Classical results on a random cosine polynomial \(V_n=\sum_{j=0}^{\infty} a_j cos(jx),\) \(x\in (0, 2\pi)\) state that if the coefficients \(a_j\) are independent and identically distributed (i.i.d.) real-valued standard Gaussian random variables, then the expected value \(\mathbb{E}[N_n(0, 2\pi)] \sim \frac{2n}{\sqrt{3}}\) as \(n\to \infty,\) where \(N_n(0, 2\pi)\) denotes the number of real zeros of \(V_n\) in \((0, 2\pi).\) This paper is one from the series of the author's works devoted to the question, on how many real zeros, compared with the case of independent coefficients, should be expected if a certain restriction is imposed upon the coefficients. \par In the present paper the author investigates how many real zeros a random cosine polynomial \(V_n\) with identically and normally distributed coefficients possesses if the coefficients are sorted in palindromic blocks of a fixed length \(l.\) More precisely, assume that \(l\in \mathbb{N}\) is a fixed number and \(n=2l m-1+r,\) where \(m\in\mathbb{N},\) and \(r \in \{0,1,\ldots, 2l-1\}.\) The coefficients \(A=(a_j)_{j=0}^n\) are divided into \(2m\) blocks of the length \(l\) in the following fashion: \( (a_0, a_1, \ldots, a_n)=\cup_{j=0}^{2m-1} A_{j} \cup \tilde{A}, \) where \(A_j=(a_{ l j}, a_{ l j+1}, \ldots, a_{ l j+l-1})\) if \( j=0, \ldots, m-1;\) \(A_j=(a_{ l j+r}, a_{ l j+1+r}, \ldots, a_{ l j+l-1+r})\) if \( j=m, \ldots, 2m-1;\) \(\tilde{A}=(a_{ l m}, a_{ l m+1}, \ldots, a_{ l m+r-1})\) if \( 1\leq r\leq 2l-1,\) and \(\tilde{A}=\emptyset\) if \(r=0.\) The author proves that if \(\cup_{j=0}^{m-1} A_{2j} \cup \tilde{A}\) is a family of i.i.d. random variables with Gaussian distribution \( \mathcal{N}(0, \sigma^2),\) and \(a_{l(2m-1-j)+r+k}=a_{l j+k}\) for \(j=0,1, \ldots, m-1,\) \(k =0,1, \ldots, l-1,\) (i.e. \(A_{2m-1-j}=A_j\)), then \(\mathbb{E}[N_n(0, 2\pi)] = \frac{2n}{\sqrt{3}}K_l+\mathcal{O}(n^{3/4}), \) as \(n \to \infty,\) where the constant \(K_l>1\) depends only on \(l\) and can be explicitly represented by a double integral formula. That is in this case slightly more real zeros should be expected compared with the classical case of random cosine polynomials with i.i.d. coefficients.
      0 references
      random trigonometric polynomials
      0 references
      random cosine polynomials
      0 references
      dependent coefficients
      0 references
      expected number of real zeros
      0 references
      palindromic blocks
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references