Kronecker conjugacy of polynomials (Q4382984)

Two polynomials \(f\), \(g\) with rational coefficients are said to be properly Kronecker conjugate if they are not linearly related (i.e. one does not have \(f(X)=g(aX+b)\) with rational \(a\neq 0\), \(b\)) and for all except finitely many rational primes \(p\) the value sets \(f({\mathbb{Z}}\bmod\;p)\) and \(g({\mathbb{Z}}\bmod\;p)\) coincide. It was shown by \textit{M. Fried} [Ill. J. Math. 17, 128-146 (1973; Zbl 0266.14013); J. Number Theory 6, 211-231 (1974; Zbl 0299.12002)] that if \(f(X)\in{\mathbb{Z}}[X]\) is functionally indecomposable (i.e. it cannot be written as a composition of polynomials of lower degree) then there are no polynomials \(g\) properly Kronecker conjugate to \(f\). The author shows (Theorem 1.2) that the same assertion holds if one assumes that \(f(X)=a(b(X))\), where the polynomials \(a\) and \(b\) are both functionally indecomposable. The considered problem may be generalized to number fields: let \(K\) be a finite extension of the rationals and call two polynomials \(f,g\) (with coefficients in \(K\)) properly Kronecker conjugate resp. strongly Kronecker conjugate over \(K\) if they are not linearly related over \(K\) resp. over the algebraic closure of \(K\) and for all except finitely many prime ideals \(P\) of the ring \({\mathbb{Z}}_K\) of integers of \(K\) the sets \(f({\mathbb{Z}_K}\bmod\;P)\) and \(g({\mathbb{Z}_K}\bmod\;P)\) coincide. It was proved by \textit{M. Fried} (preprint) and \textit{W. Feit} [J. Comb. Theory, Ser. A 14, 221-247 (1973; Zbl 0278.05016)] that if \(f\in {\mathbb{Z}}_K\) is functionally indecomposable and \(g\) is properly Kronecker conjugate to \(f\) then \(f\) and \(g\) are strongly Kronecker conjugate and the degree of \(f\) is contained in the finite set \(\{7,11,13,15,21,31\}\). The author obtains (Theorem 1.4) an analogue of this result in the case when \(f(X)=a(b(X))\) and \(a(X),b(X)\) are functionally indecomposable and neither \(a(X)\) nor \(b(X)\) is strongly conjugate over \(K\) to another polynomial. It turns out that in this case \(f(X)\) and \(g(X)\) are strongly Kronecker conjugate and one has \[ (\deg a,\deg b)\in\{(4,2),(6,2),(8,2),(10,2),(5,3),(8,3)\}. \] In each of these cases there exist explicitly given examples of suitable \(f\) and \(g\) provided the field \(K\) is chosen sufficiently large. Moreover it is shown (Theorem 3.31) that there are only seven explicitly listed possibilities for the Galois group \(G\) of the polynomial \(f(X)-t\) over \(K(t)\) (with \(t\) transcendental over \(K\)). The proofs are based on the study of the group \(G\) as well as \(\dot G\), the Galois group of \(f(X)-t\) over the field \({\bar K}(t)\) (with \({\bar K}\) being the algebraic closure of \(K\) and \(t\) transcendental over \(K\)). An essential tool is the classification of possible groups \(\dot G\) in the case when \(f\) is functionally indecomposable given by \textit{W. Feit} (loc. cit.) and the author [Contemp. Math. 186, 385-401 (1995; Zbl 0840.12001)].