SAT

A simple Python module for the 3SAT problem.

Usage

This module implements two classes:

Boolean clauses
Boolean CNF formulas

In order to load the module simple import it:

from sat import *

Clause

The general syntax for creating a clause is:

cl = Clause(pos_mask, neg_mask, n)

Where n is the number of variables in the clause, and pos_mask, neg_mask represent the positive (resp. negative) variables in the clause using a length n bit mask.

For example, say n = 5 and we want to represent the clause x_1 v !x_0 v x_4, then we represent the positive variables via pos_mask = 2**1 + 2**4 and the negative mask via neg_mask = 2**0.

So we would initialize the clause as follows:

cl = Clause(18, 1, 5)

Formula

A formula is simply the conjunction of one or more clause. Given a sequence clauses = {cl_1, ..., cl_k} of clauses, we initialize the formula cl_1 ^ cl_2 ^ ... ^ cl_k as follows:

F = Formula(clauses)

Full example

The following snippet constructs the formula F = (x_1 v x_2 v x_3) ^ (!x_2 v x_3 v x_4) over 5 variables:

cl1 = (7, 0, 5)
cl2 = (24, 4, 5)
F = Formula([cl1, cl2])

Functionality

In what follows, let F be a formula and cl be a clause. Things you can do with this module include:

Assignment testing a clause:

assignment = (pos_mask, neg_mask) # Same representation as for the clause
cl(assignment) # Evaluate clause over assignment
F(assignment) # Evaluate formula over assignment

Random elements:

cl = Clause.random(n) # Random 3-variable clause over n variables
F = Formula.random(n, k) # Random 3CNF formula with k clauses over n variables
ass = Formula.random_assignment(n) # Random assignment of n variables
ass_hash = Formula.random_hashed(n) # Random 2-wise indpendent assignment

Brute force solutions:

sol = F.brute_force() # Return solution for a formula F usign brute force search in (pos_mask, neg_mask) format
count = F.brute_force(count=True) # Count number of solutions to a formula F

Approximations:

apx_cnt = F.approximate_count() # Count probability of acceptance for random assignment
apx_sat = F.approximate_sat() # Count avg. number of accepted clause in random assignment
max_sat = F.approximate_sat(avg=False) # Count max number of accepted clause in random assignment

License

This project is distributed under the Apache license version 2.0 (see the LICENSE file in the project root).

nparzan/sat

SAT