Geek Answers Handbook

Z3py: convert formula Z3 to positions used by picosate

LINKS:
Launching python-bound python theorem Z3

I used Z3 as a SAT solver. For larger formulas, there seems to be performance issues, so I wanted to check picosate to see if this is a faster alternative. My existing python code generates a propositional formula in z3 syntax:

from z3 import *

import pycosat
from pycosat import solve, itersolve

#
#1 2  3  4  5  6  7   8  (variable names in picosat are numbers!)
#
C, G, M, P, R, S, SN, B = Bools('C G M P R S SN B')
C = (And(*(S,Or(Not(S),P),Or(Not(P),S),Or(Not(P),B),Or(Not(C),P),Or(Not(G),P),Or(Not(M),P),Or(Not(R),P),Or(Not(SN),P),Or(Not(B),P),True,Not(False),Or(R,SN,B,G,M,C,True))))

# formula in Z3:
f = simplify(C)

print f

OUTPUT / RESULT

And(S,
    Or(Not(S), P),
    Or(Not(P), S),
    Or(Not(P), B),

    Or(Not(C), P),
    Or(Not(G), P),
    Or(Not(M), P),
    Or(Not(R), P),
    Or(Not(SN), P),
    Or(Not(B), P))

However, Picosat uses lists / arrays of numbers, as shown in the following example ("clauses1": 6 refers to the variable P, -6 means "not P", etc.):

import pycosat
from pycosat import solve, itersolve
#
# use pico sat
#
nvars = 8
clauses =[
    [6],
    [-6, 4],   ##  "Or(Not(S), P)" from OUPUT above
    [-4, 6],
    [-4, 8],
    [-1, 4],
    [-2, 4],
    [-3, 4],
    [-5, 4],
    [-7, 4],
    [-8, 4]]

#
# efficiently find all models of the formula
#
sols = list(itersolve(clauses, vars=nvars))
print "result:"
print sols
print "\n\n====\n"

Z3- (, varaible "f" ), CNF, , picosat CNF? API- python Z3, .

( , . , C, , z3 )

+4
1

Z3 CNF. .

Z3 CNF Dimacs, , . , Z3

, ,

+3

More articles:


All Articles