The paper develops a general method for building "free algebras" in two related families of mathematical structures called Hilbert algebras and Brouwerian semilattices. These structures come from logic: Hilbert algebras capture the implication connective in certain logics, while Brouwerian semilattices also include a conjunction operation. A free algebra on n generators is essentially the most general possible algebra of its type built from n starting elements, with no extra assumptions beyond the defining rules. Free algebras are important because they encode all the logical identities that hold in a given class of structures, and understanding them tells you a great deal about the underlying logic.
The construction works by exploiting a classical result in algebra: any algebra can be broken down into simpler building blocks called subdirectly irreducible algebras. The authors show that if you have a clear enough description of these building blocks for a particular class of Hilbert algebras or Brouwerian semilattices, you can systematically assemble the corresponding free algebra. The method is uniform, meaning the same recipe applies across many different classes at once, rather than requiring a custom argument each time. They also extend the approach to versions of these structures that include a "bottom" or zero element.
To illustrate the method's reach, the authors work out several concrete examples: free algebras in classes defined by a bound on height or width (geometric properties of the algebra's structure), and a tidy closed formula counting the size of free algebras in the class of linear Hilbert algebras, where elements are totally ordered. As a bonus, the work sheds light on a property called structural completeness, which asks whether every rule of inference that preserves all the valid statements in a logic is already derivable from its axioms. The authors find a condition under which a conditional identity (a quasi-equation) in Hilbert algebras with zero is actually equivalent to an unconditional one, which is a useful simplification in the study of these logics.