Scribd
Upload
90 views

Assignment 2

The document describes an assignment to complete several logic problems, including: 1) Converting a formula to disjunctive normal form 2) Converting a formula to conjunctive normal form 3) Expressing logical connectives using only the NAND connective 4) Manually running the Davis-Putnam algorithm on a given formula The student is asked to show the steps of running the Davis-Putnam algorithm on the formula (¬x ∨ y) ∧ (x ∨ ¬y) ∧ (y ∨ z) ∧ (x ∨ ¬z).

Uploaded by

compiler&automata
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
90 views

Assignment 2

The document describes an assignment to complete several logic problems, including: 1) Converting a formula to disjunctive normal form 2) Converting a formula to conjunctive normal form 3) Expressing logical connectives using only the NAND connective 4) Manually running the Davis-Putnam algorithm on a given formula The student is asked to show the steps of running the Davis-Putnam algorithm on the formula (¬x ∨ y) ∧ (x ∨ ¬y) ∧ (y ∨ z) ∧ (x ∨ ¬z).

Uploaded by

compiler&automata
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

15-414 HW 2 1

Instructor: Edmund M. Clarke Teaching Assistant: Michael Carl Tschantz


Assignment 2

1 DNF
Put (x → y) ∧ (z ↔ x) into disjunctive normal form.

2 CNF
Let f be the DNF formula (u ∧ v) ∨ (w ∧ x).

(a) Find a formula in conjunctive normal form that is logically equivalent to f .


(b) Find an equi-satisfiable formula in CNF to f by introducing a new variable a for u ∧ v
and a new variable b for w ∧ x.

3 Universal Sets of Connectives


We call a set of connectives universal if for any logical formula, there exists an equivalent formula
using only connectives in that set. For example, {∧, ∨, ¬} is a universal set.
Let ⊗ be the NAND connective with the following truth table

x y x⊗y
F F T
F T T
T F T
T T F

(a) Produce a formula equivalent to ¬x using only the connective ⊗ and variable x.
(b) Produce a formula equivalent to x ∧ y using only the connective ⊗ and the variables x
and y.
(c) Produce a formula equivalent to x ∨ y using only the connective ⊗ and variables x and
y.
(d) Briefly explain how this shows that {⊗} is a universal set of connectives.

4 Davis-Putnam
In this problem you will run the Davis-Putnam algorithm by hand showing the value of the input
variable S and the return value for each call to Satisfiable.

1
15-414 HW 2 2

Background. The Davis-Putnam algorithm is provided in the lecture “Propositional Logic”


(page 22) as the function Satisfiable. The function Satisfiable has one input, the clause
list S. Since the algorithm operates on a formula in CNF, the clauses of S are implicitly conjoined,
and each clause of S represents a disjunction literals. For example, the formula (¬x ∨ y) ∧ (x ∨
¬y) ∧ (y ∨ z) ∧ (x ∨ ¬z) is represented as the list of lists [[¬x, y], [x, ¬y], [y, z], [x, ¬z]].
Note that line 12 of the algorithm
choose a literal L occurring in S
introduces nondeterminism into the system since it does not state which literal L to choose. To
remove this nondeterminism, presume that the algorithm always selects the first literal of the first
clause in S.
Treat S ∪ {L} as appending the list S to the list [[L]]. For example, [[x, ¬y], [x, z]] ∪ {z} is
[[x, ¬y], [x, z], [z]].

What You Must Do. Run the function Satisfiable by hand when called with value of
[[¬x, y], [x, ¬y], [y, z], [x, ¬z]] for S. Since Satisfiable is a recursive function (see lines
13 and 14), it will call itself. To prove that you ran the function by hand, show the value of the
input S at the beginning of each of these recursive calls to Satisfiable. Also, note the return
value of each call.

Examples. For example, if the problem asked you to run Satisfiable with the initial value of
[[¬x, y], [y, ¬z], [¬y, x]] for S instead of [[¬x, y], [x, ¬y], [y, z], [x, ¬z]], a valid solution to this
problem would be (with optional comments explaining the solution in parentheses):
(Satisfiable is initially called with S = [[¬x, y], [y, ¬z], [¬y, x]]. It selects ¬x for L on line
12 for the first recursive call on line 13.)
On the first and only recursive call of Satisfiable, S = [[¬x, y], [y, ¬z], [¬y, x], [¬x]]. ([¬x]
is a unit clause so Satisfiable deletes every clause containing ¬x, and x is deleted from every
remaining clause. This yields [[y, ¬z], [¬y]]. Now [¬y] is a unit clause so every clause containing
¬y is deleted, and y is deleted from every remaining clause. This yields [[¬z]]. Since [¬z] is a
unit clause it is deleted leaving []. Since S is now empty,) true is returned by this recursive call.

Another example: if the problem instead asked you to run Satisfiable with the initial value of
[[¬x, y, z], [x, ¬y, ¬z], [y, z], [x, y, ¬z]] for S, a valid solution would be:
(Satisfiable is initially called with S = [[¬x, y, z], [x, ¬y, ¬z], [y, z], [x, y, ¬z]]. It selects
¬x as L for the first recursive call.)
On the first recursive call, S = [[¬x, y, z], [x, ¬y, ¬z], [y, z], [x, y, ¬z], [¬x]]. ([¬x] is a unit
clause so every clause containing ¬x is deleted, and x is deleted from every remaining clause.
This yields [[¬y, ¬z], [y, z], [y, ¬z]]. ¬y is selected as L for the second recursive call.) True is
returned by this recursive call. (The reason being that true is returned the third recursive call,
which is described below.)
On the second recursive call, S = [[¬y, ¬z], [y, z], [y, ¬z], [¬y]]. ([¬y] is a unit clause so this
yields [[z], [¬z]] after the removals. z is a unit clause yielding [[]]. Thus, S contains the null
clause.) False is returned by this recursive call. (y is now tried by the first recursive call since
[¬y] lead to false.)
On the third and final recursive call, S = [[¬y, ¬z], [y, z], [y, ¬z], [y]]. ([y] is a unit clause
yielding [[¬z]]. Since [[¬z]] is a unit clause, it is removed leaving [].) True is returned by this
call.

You might also like