A formal system is an ordered quadruple
where:
- is a finite, nonempty alphabet (a set of symbols);
- is the set of well-formed-formulas (wffs), defined inductively by formation rules specifying which finite strings over belong to
- is a distinguished subset of , whose members are called axioms.
- is a finite set of rules of inference, each of which is a relation on determining from which formulas other formulas may be derived.
Referenced by (1 direct)
Direct references:
A derivation (or proof) in is a finite sequence of formulas
such that for each ,
- Either , or
- follows from earlier formulas in the sequence by an inference rule in .
Referenced by (1 direct)
Direct references:
A formula is a theorem (or provable formula) of if there exists such a derivation ending with . We then write
The system is said to be recursive (or effectively generated) if , , and are all @recursively-enumerable sets, so that derivations can be verified by a finite mechanical procedure.
No @semantics are assumed in this definition; a formal system is purely @syntactic.
A formal system operates purely on @form, not content.
Its symbols are inert marks; its rules manipulate these marks according to syntactic shape alone.
No step depends on what any symbol "means."
A proof in such a system is therefore a finite, mechanical transformation of strings; a sequence of moves in a symbol game governed entirely by explicit rules.
Semantics enter only afterward, when we interpret the symbols externally (e.g. as numbers, sets, or truth values).
Until then, the system is a syntax engine; a rule-bound procedure that generates certain strings (theorems) from others (axioms), without regard to interpretation or truth.
When an interpretation (model) and a satisfaction relation are supplied, one obtains a formal theory.