Pisot numbers in the neighbourhood of a limit point. I (Q1064345)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Pisot numbers in the neighbourhood of a limit point. I |
scientific article |
Statements
Pisot numbers in the neighbourhood of a limit point. I (English)
0 references
1985
0 references
Let \(S\) be the set of Pisot numbers. The paper examines the problem of determining all elements of \(S\) in a neighbourhood of a limit point of \(S\). Using ideas of \textit{J. Dufresnoy} and \textit{Ch. Pisot} [Ann. Sci. Éc. Norm. Supér., III. Sér. 70, 105--133 (1953; Zbl 0051.02904)], an infinite tree \({\mathcal T}\) associated with \(S\) becomes the object of investigation. Each path to infinity in this tree corresponds to a rational function \(f\) bounded by one on the unit circle and with a single pole inside the circle at the reciprocal of some Pisot number \(\theta\). To determine a neighbourhood of \(\theta\), it is enough to determine a neighbourhood in \({\mathcal T}\) of each \(f\) associated with \(\theta\), provided the number of such \(f\) is finite. If \(f\) corresponds to a limit point of \(S\) and if \({\mathcal T}_ n(f)\) is the subtree of \({\mathcal T}\) which branches off from \(f\) at the height \(n\) then \({\mathcal T}_ n(f)\) converges (in a suitable sense) to a tree \({\mathcal T}'(f)\) called the derived tree of \(f\). If this tree is essentially finite (as defined in the paper) then a certain neighbourhood of \(f\) can be effectively determined. This gives an effective version of a theorem of \textit{F. L. Talmoudi} [C. R. Acad. Sci., Paris, Sér. A 287, 739--741 (1978; Zbl 0411.12002)] concerning the set \(S\cap [1,2[\). The ineffectiveness of that result arises from the use of a compactness result of \textit{Ch. Pisot} [Ann. Sci. Éc. Norm. Supér., III. Sér. 81, 165--188 (1964; Zbl 0139.28002)] at an early stage of the argument. In the author's approach, a compactness argument also appears in the guise of König's lemma at a later stage in the argument (Lemma 8.2). This is used to show that for each \(f\) associated with any \(\theta <2\) the tree \({\mathcal T}'(f)\) is essentially finite (Lemma 8.6, N.B. the misprint of \(>\) for \(<)\). However, for any given \(f\), the verification that \({\mathcal T}'(f)\) is essentially finite is a finite computation which does not require König's lemma. Knowing \({\mathcal T}'(f)\), the computation of a neighbourhood of \(f\) can then be done effectively. This is illustrated in a sequel [Math. Comput. 43, 593--602 (1984; Zbl 0554.12003)], where the algorithm is used to determine the smallest neighbourhood of the limit point \(\alpha_ 3\), the positive zero of \(z^ 4-2z^ 3+z-1\), within which all elements of \(S\) satisfy an equation of the form \(z^ n(z^ 4-2z^ 3+z-1)\pm A(z)\), where \(A(z)\) is one of \(z^ 3-z+1\), \(z^ 3-z^ 2+1\) or \(z^ 4-z^ 3+z-1\). This result was inaccessible to the author's earlier algorithm [Math. Comput. 32, 1244--1260 (1978; Zbl 0395.12004)].
0 references
Pisot numbers
0 references
limit point
0 references
infinite tree
0 references
effective version
0 references
algorithm
0 references
0 references