A necessary and sufficient condition for an algebraic integer to be a Salem number (Q2199496)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | A necessary and sufficient condition for an algebraic integer to be a Salem number |
scientific article |
Statements
A necessary and sufficient condition for an algebraic integer to be a Salem number (English)
0 references
11 September 2020
0 references
A Salem number is a real algebraic integer greater than \(1\) whose other conjugates lie in the closed unit disc, with at least one on the boundary. Then, the (normalized) minimal polynomial \(P_{\alpha}(x)=\sum_{k=0}^{d}a_{k}x^{k}\) of a Salem number \(\alpha \) is reciprocal, i.e., \(a_{k}=a_{d-k}\) for all \(0\leq k<d/2\) (or equivalently, \(P_{\alpha}(x)=x^{d}P_{\alpha }(1/x)),\) and has two real roots, namely \(\alpha \) and \(1/\alpha ,\) and \((d-2)\) roots with modulus \(1;\) thus \(d\) is even and \(d\geq 4.\) There is no known general rule for testing whether a reciprocal polynomial with rational integer coefficients is irreducible. In the paper under review, the author shows that a real algebraic integer \(\theta >1\) of degree \(d\) is a Salem number if and only if there is a natural number \(n\) such that the minimal polynomial \(P_{\theta ^{n}}(x)\) of \(\theta ^{n}\) is of the form \(x^{d}+b_{d-1}x^{d-1}+\cdot \cdot \cdot +b_{1}x+1,\) where \(b_{k}=b_{d-k}\) for all \(\ 0<k<d/2,\) and \[ 2(d-2)\left\vert b_{1}\right\vert >d(2+\sum_{k=2}^{d-2}\left\vert b_{k}\right\vert ).\tag{*} \] The direct implication in this equivalence follows from a theorem of C. Pisot, saying that if \(e^{\pm i2\pi \omega _{1}},\ldots,e^{\pm i2\pi \omega_{d/2-1}}\) are the conjugates with modulus \(1\) of a Salem number then the numbers \(1,\omega _{1},\ldots,\omega _{d-2}\) are \(\mathbb{Q}\)-linearly independent (see for instance [\textit{R. Salem}, Algebraic numbers and Fourier analysis. Reprint. Orig. publ. 1963 by Heath, Boston. Belmont, California: Wadsworth International Group, a Division of Wadsworth, Inc (1983; Zbl 0505.00033)]), while the implication inverse is a corollary of a result of \textit{R. S. Vieira} [Ramanujan J. 42, No. 2, 363--369 (2017; Zbl 1422.30013)], asserting that the reciprocal polynomial \(P_{\theta ^{n}}\) has exactly \((d-2)\) roots with modulus \(1,\) when its coefficients satisfy (*). Also, for a given Salem number \(\alpha ,\) with degree \(4\) or \(6\) (resp. \(8\) or \(10),\) the probability that the coefficients of the minimal polynomial of an arbitrary power of \(\alpha \) satisfy (*) is determined (resp. is approximated) in the second part of the present manuscript.
0 references
special algebraic numbers
0 references
Salem numbers
0 references
reciprocal polynomials
0 references