Polynomial functions on upper triangular matrix algebras (Q2408003)

Let \(D\) be an integral domain, \(K\) its field of quotients and \(B=A\otimes_DK\), where \(A\) is \(D\)-algebra, assumed to be finitely generated and torsion-free. One defines two sets of integral-valued polynomials on \(A\) by putting \[ \mathrm{Int}_K(A)=\{f\in K[X]:\;f(A)\subset A\},\quad \mathrm{Int}_B(A)=\{f\in B[X]:\;f(A)\subset A\}. \] It has been shown in 2013 by the author [J. Algebra 373, 414--425 (2013; Zbl 1273.13037)] that if \(A\) is the ring \(M_n(D)\) of \(n\times n\) matrices over \(D\), then one has \[ Int_B(A)=Int_K(A)\otimes_DA,\eqno(1) \] and later \textit{G. Peruginelli} and \textit{N. J. Werner} [``Decomposition of integer-valued polynomial algebras'', Preprint, \url{arXiv:1604.08337}] determined when (1) holds in the case when \(D\) is either a discrete valuation ring or a Dedekind domain with the finite norm property. The author now treats the analogous question in the case when \(A\) is the algebra of upper triangular matrices over a commutative ring \(R\) and describes the relation between polynomial functions on \(A\) induced by polynomials with matrix coefficients and those induced by polynomials over \(R\) (Theorem 4.2 and Remark 1.6). Reviewer's remark: Note that in the item 8 of the references the arXiv number is incorrect, it should be 1604.06912. Actually this paper has already appeared in [\textit{G. Peruginelli} and \textit{N. J. Werner}, Monatsh. Math. 183, No. 1, 177--189 (2017; Zbl 1364.13018)].