MAT. SBORNIK MATH. USSR SBORNIK
TOM 102(144)(1977), NO. 2 Vol. 31 (1977), No. 2
ON ADMISSIBLE RULES OF INTUITIONISTIC
PROPOSITIONAL LOGIC
UDC 517.12
A. I. CITKIN
ABSTRACT. This paper studies modus rules of deduction admissible in intuitionistic
propositional logic (a rule is called a modus rule if it corresponds to some sequence and
allows passage from the results of any substitution in the formulas in its antecedent to the
result of the same substitution in its succedent). Examples of such rules are considered, as
well as the derivability of certain rules from others by means of the intuitionistic
propositional calculus. An infinite independent system of admissible modus rules is
constructed. It is proved that a finite Godel pseudo-Boolean algebra in which all modus
rules are valid (i.e. the quasi-identities corresponding to them are valid) is isomorphic to a
sequential union of Boolean algebras of power not greater than 4.
Bibliography: 17 titles. Figures: 3.
In the foundations of mathematics an important role is played by the intuitionistic
propositional logic which is given by the well-known calculus [2]-a list of ten axiom
schemes and the deductive rule of modus ponens. This calculus yields not only the logic,
understood extensionally as the set of formulas true or valid in it, i.e. deducible
(provable) in the calculus, but also a relation of deducibility (by means of this calculus)
of certain formulas from others. It is natural to raise the question of the possibility of
altering the relation of deducibility by postulating other rules of deduction without
altering the logic itself. It is known (see [9], for example) that for intuitionistic proposi-
tional logic there exist admissible rules of deduction (i.e. rules relative to which the logi©
is closed) that are not derivative^) (i.e. the conclusion of the rule cannot be deduced
from the premises of the rule with the aid of only the axioms and the rule of modus
ponens). The application of admissible rules does not extend the set of formulas valid in
intuitionistic logic, but broadens the possibilities for constructing proofs. The problem
arises of studying the relation of deducibility of certain formulas from others specified
by the collection of all deductive rules (including the axiom schemes) admissible in
intuitionistic logic; in particular, the problem of whether this relation can be given by a
finite number of rules of an easily discernible form.(
2
)
AMS (MOS) subject classifications (1970). Primary 02B05, 02C15, 02J05, 02D99, 02E05, 06A35; Secondary
02B99, 02E99, 06A25, 02H10, 06A40, 08A15.
(')Sometimes, in comparing admissible with derivative rules, the latter are called deducible, for example in
[16]. The term "derivative" is taken from [9].
(
2
)Made appropriately precise (see [9] and below), the latter question is equivalent to the following: can the
quasi-variety of pseudo-Boolean algebras, generated by the free algebras, be specified by a finite number of
quasi-identities? Cf. Problem 3.30 in [4], posed by D. M. Smirnov, concerning the analogous quasi-variety of
groups, and also Problem 7 in [ 3 ].
© American Mathematical Society 1978
279