Isomorphisms of algebras of binary functions (Q1893630)

The algebras of binary functions discussed in this paper are analogues and generalizations of incidence algebras defined on `structured sets' \((D, \leq)\). (The binary relation \(\leq\) is mostly taken to be reflexive in this paper.) The mappings \(f: D^2\to K\), require \(K\) to be equipped with associative operations \(+\), \(\cdot\) and an element 0 (ringoid) and the algebra of such functions is defined with sum and scalar multiplication as obvious, while \[ (f * g) (x,y)= \sum_z f(x,z) \cdot M(x,y, z)\cdot g(z,y), \] with the (kernel) function \(M: P^3\to K\) defined such that \(f * g\) is well defined. The general problem is to demonstrate that if suitably closely related algebraic structures of a pair of such algebras are (anti)-isomorphic in their category, then suitably closely related structures associated with the structured sets are correspondingly (anti)-isomorphic in their own category. This type of converse of an ``obvious result'' is usually nontrivial to obtain and almost always useful in the setting of its own theory. E.g., \((P, \leq)\) a locally finite poset, \(M\equiv 1\), \(K\) a field of characteristic 0 yields a well-known isomorphism theorem due to R. Stanley. Stechkin has obtained results on Möbius inversion and applications in a general setting. Baclawski has studied the \(K\)-space of derivations on incidence algebras and given applications. Other authors, including the author of this paper have added significantly to the information already available. Here, the author requires pairs of kernels \(M_1\), \(M_2\) associated with algebras \(A_1\), and \(A_2\) to be associative or \(\zeta\) similar, making use of some careful counting arguments needed to replace the (local) finiteness of intervals in the more traditional setting and adapting traditional arguments to the changed circumstances simultaneously in a proficient balancing act.