Bauhaus

bauhaus is a Python package for spinning up propositional logic encodings from object-oriented Python code.

Features

• Create propositional variables from Python classes
• Build naive SAT encoding constraints from propositional variables
  • At most one
  • At least one
  • Exactly one
  • At most K
  • Implies all
• Compile constraints into a theory in conjunctive or negation normal form
• With python-nnf, submit a theory to a SAT solver
• Theory introspection

Installation

Install bauhaus by running

pip install bauhaus

How is it used?

Create Encoding objects that you intend to compile to an SAT. Encoding objects will store your model's propositional variables and constraints on the fly.

from bauhaus import Encoding, proposition, constraint
e = Encoding()

Create propositional variables by decorating class definitions with @proposition.

@proposition(e) # Each instance of A is stored as a proposition
class A(object): pass

Create constraints by decorating classes, methods, or invoking the constraint methods.

# Each instance of A implies the right side
@constraint.implies_all(e, right=['hello'])
# At most two of the A instances are true
@constraint.at_most_k(e, 2)
@proposition(e)
class A(object):

    def __init__(self, val):
        self.val = val

    def __repr__(self):
        return f"A.{self.val}"

    # Each instance of A implies the result of the method
    @constraint.implies_all(e)
    def method(self):
        return self.val

# At most one of the inputs is true.
constraint.add_at_most_one(e, A, A.method, Var('B'))

Compile your theory into conjunctive or negation normal form (note: the theory is truncated),

objects = [A(val) for val in range(1,4)]
theory = e.compile()
>> And({And({Or({Var(3), ~Var(A.3)}), Or({Var(1), ~Var(A.1)}),
        ...
        And({Or({~Var(A.1), ~Var(A.2), ~Var(A.3)})})})

And view the origin of each constraint, from the propositional object to the final constraint. (Note: the introspection is truncated)

e.introspect()
>>
constraint.at_most_k:  function = A  k = 2:

(~Var(A.3), ~Var(A.1)) =>
Or({~Var(A.1), ~Var(A.2), ~Var(A.3)})


(~Var(A.3), ~Var(A.2)) =>
Or({~Var(A.1), ~Var(A.2), ~Var(A.3)})


(~Var(A.1), ~Var(A.2)) =>
Or({~Var(A.1), ~Var(A.2), ~Var(A.3)})


Final at_most_k: And({Or({~Var(A.1), ~Var(A.2), ~Var(A.3)})})
...
...

Contribute

Head over to our code of conduct and get a feel for the library by reading our architecture design and documentation.

• Issue Tracker: https://github.com/QuMuLab/bauhaus/issues
• Source Code: https://github.com/QuMuLab/bauhaus
• Join us! http://mulab.ai/

Support

If you are having issues, please let us know. Reach out to us at [email protected] or by creating a GitHub issue.

License

The project is licensed under the MIT license for the Queen's Mu Lab

Citing This Work

bauhaus was created by Karishma Daga under mentorship of Christian Muise at Queen's University, Kingston.