Objects of categories as complex numbers (Q703807)

For complex numbers, it is easy to deduce, that \(x^2+1=x\) implies \(x^7=x\). \textit{A. Blass} [J. Pure Appl. Algebra 103, 1--21 (1995; Zbl 0846.18002)] proved that this also holds for the category of sets in the following sense: Given an explicit isomorphism between \(T\) and \(T^2+1\) (\(T^2\) is Cartesian product of \(T\) with itself and \(1\) is an added basepoint), this provides an explicit isomorphism between \(T\) and \(T^7\). In the present paper, a much more general statement is proven. A category with a symmetric additive structure \((\oplus,0)\) and a multiplicative structure, \((\otimes,1)\) such that the latter distributes over the former up to coherent isomorphism is called a {rig} category. A polynomial \(p(x)\) over the natural numbers may be interpreted as a polynomial in a rig category \(\mathcal{A}\), since an object \(T\) from \(\mathcal{A}\) gives rise to a new object \(p(T)\) in the category, well-defined up to canonical isomorphism. Hence polynomial equations may be interpreted as isomorphisms. The main result is stated in two versions of which we give the less general version first: Given polynomials \(p\), \(q_1\) and \(q_2\) in \(\mathbb{N}[x]\) such that \(p\) has non-zero constant term and moreover \(q_1\) and \(q_2\) have degree at least one, for such polynomials, if \(p(x)-x\) is primitive in \(\mathbb{Z}[x]\) and has no repeated complex roots and if each complex root, \(r\), satisfies \(q_1(r)=q_2(r)\), then \(x=p(x)\) implies \(q_1(x)= q_2(x)\) {rig theoretically}, i.e., for any rig category, \(\mathcal{A}\) and any \(T\in\mathcal{A}\), given an explicit isomorphism between \(T\) and \(p(T)\) one can build an isomorphism \(q_1(T)=q_2(T)\). Another version says: Let \(p\), \(q_1\) and \(q_2\) be as above. Suppose \(x=p(x)\) implies \(q_1(x)= q_2(x)\) ring-theoretically, i.e., for all rings \(A\) and all \(a\in A\), if \(a=p(a)\) then \(q_1(a)=q_2(a)\). Then \(x=p(x)\) implies \(q_1(x)= q_2(x)\) rig theoretically.