Kleene Algebra of Partial Predicates

Abstract

Summary We show that the set of all partial predicates over a set D together with the disjunction, conjunction, and negation operations, defined in accordance with the truth tables of S.C. Kleene’s strong logic of indeterminacy [17], forms a Kleene algebra. A Kleene algebra is a De Morgan algebra [3] (also called quasi-Boolean algebra) which satisfies the condition x ∧¬:x ⩽ y ∨¬ :y (sometimes called the normality axiom). We use the formalization of De Morgan algebras from [8]. The term “Kleene algebra” was introduced by A. Monteiro and D. Brignole in [3]. A similar notion of a “normal i-lattice” had been previously studied by J.A. Kalman [16]. More details about the origin of this notion and its relation to other notions can be found in [24, 4, 1, 2]. It should be noted that there is a different widely known class of algebras, also called Kleene algebras [22, 6], which generalize the algebra of regular expressions, however, the term “Kleene algebra” used in this paper does not refer to them. Algebras of partial predicates naturally arise in computability theory in the study on partial recursive predicates. They were studied in connection with non-classical logics [17, 5, 18, 32, 29, 30]. A partial predicate also corresponds to the notion of a partial set [26] on a given domain, which represents a (partial) property which for any given element of this domain may hold, not hold, or neither hold nor not hold. The field of all partial sets on a given domain is an algebra with generalized operations of union, intersection, complement, and three constants (0, 1, n which is the fixed point of complement) which can be generalized to an equational class of algebras called DMF-algebras (De Morgan algebras with a single fixed point of involution) [25]. In [27] partial sets and DMF-algebras were considered as a basis for unification of set-theoretic and linguistic approaches to probability. Partial predicates over classes of mathematical models of data were used for formalizing semantics of computer programs in the composition-nominative approach to program formalization [31, 28, 33, 15], for formalizing extensions of the Floyd-Hoare logic [7, 9] which allow reasoning about properties of programs in the case of partial pre- and postconditions [23, 20, 19, 21], for formalizing dynamical models with partial behaviors in the context of the mathematical systems theory [11, 13, 14, 12, 10].

Partial Correctness of a Factorial Algorithm

“…Then ∼ (Equality(A, loc /1 , loc /3 ) ∧ factorial-inv(A, loc, n 0 )), Asg z ((loc /4 ) ⇒ a ), valid-factorial-output(A, z, n 0 ) is an SFHT of ND SC (V, A). (14) Partial correctness of a FACTORIAL algorithm: Suppose V is not empty and A is complex containing and V is without nonatomic nominative data w.r.t. A and loc /1 , loc /2 , loc /3 , loc /4 are mutually different and loc and val are compatible w.r.t.…”

An Inference System of an Extension of Floyd-Hoare Logic for Partial Predicates

“…The formalized rules can be used for reasoning about sequential programs, and in particular, for sequential programs on nominative data [4]. Application of these rules often requires reasoning about partial predicates representing preand post-conditions which can be done using the formalized results on the Kleene algebra of partial predicates given in [8]. In the sequel D denotes a non empty set, d denotes an element of D, f , g denote binominative functions of D, and p, q, r, s denote partial predicates of D.…”