FirstOrderLogic.jl | Overview | Installation | Usage | Why not PROLOG? | Roadmap

FirstOrderLogic.jl

This is a Julia package for parsing, manipulating and evaluating formulas in first-order logic. It may prove useful to

• the mathematician who wants to quickly verify a formula and enter it in plain mathematical notation.
• the student studying first order logic and despairing over his homework.
• the programmer looking for a clean implementation of first-order logic algorithms.

For example, to check if ∀x(P(x) ∧ ¬P(f(x))) is unsatisfiable, run

is_satisfiable("*{x}(P(x) & ~P(f(x)))")

Overview

Linux	Windows	Coverage

FirstOrderLogic.jl follows the standard structure for Julia projects. Currently, it provides the following:

• An intuitive type system for formulas in first-order logic.

• Various methods for converting formulas in special forms:

  • prenex normal form
  • skolem normal form
  • conjunctive normal form
  • ...

• A prover for formulas in first-order logic.

Installation

1. Run

julia

2. In your Julia shell, run

Pkg.clone("[email protected]:roberthoenig/FirstOrderLogic.jl.git")

Usage

Quick walkthrough

Say we have a formula ∀x(P(x) ∧ ¬P(f(x))). We want to verify that this formula is unsatisfiable. To do so, we do the following:

1. Type the formula in FirstOrderLogic.jl syntax: *{x}(P(x) & P(f(x))).
2. Run

julia

3. Import FirstOrderLogic.jl:

using FirstOrderLogic

4. Run is_satisfiable(<our formula>):

is_satisfiable("*{x}(P(x) & ~P(f(x)))")

If everything works correctly, the program should print "false" and terminate. This means that ∀x(P(x) ∧ ¬P(f(x))) is unsatisfiable. We have verified what we wanted to verify.

Formula syntax

FirstOrderLogic.jl currently uses its own syntax for representing formulas in first-order logic. A parser for latex symbols is in progress.

The following is a list of all currently supported syntactic elements.

Mathematical syntax	FirstOrderLogic.jl syntax	Semantic
x	x	variable
f(x,y)	f(x,y)	function
f(x,y)	f(x,y)	predicate
∃x	?{x}	existential quantification
∀x	*{x}	universal quantification
¬x	~x	negation
x ∧ y	x & y	conjunction
x ∨ y	x | y	disjunction
x → y	x > y	implication
(x ∧ y) ∨ z	(x & y) | z	precedence grouping

Example

Formula in mathematical notation:

∀x∃yA(x,g(y,z)) ∧ (¬B() ∨ ∃dC(d))

The same formula in FirstOrderLogic.jl notation:

*{x}?{y}A(x,g(y,z)) & (~B() | ?{d}C(d))

Some important functions

Function	Purpose
is_satisfiable(formula; maxsearchdepth)	Checks if formula is satisfiable.
get_conjunctive_normal_form(formula)	Returns formula in the conjunctive normal form, expressed as clauses.
get_prenex_normal_form(formula)	Returns the prenex normal form of formula.
get_skolem_normal_form(formula)	Returns the skolem normal form of formula.
Formula(str)	Parse str to a Formula object and return that object.

Each available function comes with a concise docstring.

How is this different from PROLOG?

The PROLOG programming language can be used to prove seemingly arbitrary statements in first-order logic. PROLOG differs from FirstOrderLogic.jl in two ways:

• Syntax: PROLOG code is written in facts. These facts represent horn clauses. In doing so, PROLOG hides the first-order logic syntax from the user. This proves convenient for real world applications, but it's not mathematically pure. FirstOrderLogic.jl uses a direct mathematical syntax for expressing formulas.
• Expressive power: PROLOG works on a limited set of all valid formulas in first-order logic, namely only formulas that can be expressed as a conjunction of horn clauses. This is sufficient for many real world applications, but it's not mathematically complete. PROLOG thereby trades completeness for computation time. FirstOrderLogic.jl does not make that trade. It is slow, but complete.

Roadmap

The following features are work in progress:

• A highly optimized, parallelizable version of the satisfiability checker is_satisfiable().
• A deadsimple webapp for the satisfiability checker is_satisfiable().

Any and all external contributions are most welcome!