CS103 Handout 10

Winter 2012-2013 February 1, 2013

Problem Set 4
This fourth problem set explores propositional and first-order logic, along with its applications.
Once you've completed it, you should have a much stronger understanding of mathematical logic
and its properties.
In any question that asks for a proof, you must provide a rigorous mathematical proof. You
cannot draw a picture or argue by intuition. If we specify that a proof must be done a certain way,
you must use that particular proof technique; otherwise you may prove the result however you
wish.
As always, please feel free to drop by office hours or send us emails if you have any questions.
We'd be happy to help out.
This problem set has 150 possible points. It is weighted at 7% of your total grade. The earlier
questions serve as a warm-up for the later problems, so be aware that the difficulty of the
problems does increase over the course of this problem set.
Good luck, and have fun!

Checkpoint due Monday, February 4 at 2:15PM

Assignment due Friday, February 8 at 2:15PM
Write your solutions to the following problems and submit them by Monday, February 4th at the start of
class. These problems will be graded based on whether or not you submit solutions, rather than the
correctness of your solutions. We will try to get these problems returned to you with feedback on your
proof style this Wednesday, February 6th. Submission instructions are on the last page of this problem
set.

Checkpoint Problem: The Biconditional Operator (25 Points if Submitted)

Suppose that you want to prove the statement p ↔ q. In lecture, we saw that this is equivalent to prov-
ing p → q and q → p, but there are many other ways to prove this statement.
Below are four propositional formulas. For each of those formulas, write a truth table for that formula.
Based on the truth table, decide whether or not that formula is equivalent to p ↔ q. If the formula is
equivalent to p ↔ q, then any time that you need to prove p ↔ q, you could instead prove the other for-
mula instead. (This will be extremely useful in the later problem sets!)
i. (p → q) ∧ (¬p → ¬q)
ii. (p → q) ∧ (¬q → ¬p)
iii. (p ∧ q) ∨ (¬p ∧ ¬q)
iv. p → (q → p)
Problem One: The Epimenides Paradox (12 points)
In logic, we assume that every statement is either true or false. However, some statements called
logical paradoxes break this rule and can be neither true or false. For example, the statement “this
statement is false” is a paradox – if it were true, it would have to be false, and if it were false, it would
have to be true. The statement is therefore a paradox – it must be either true or false, but it can be
neither true nor false.
One of the earliest paradoxes is the called the Epimenides Paradox, which is stated as follows:
Epimenides, a Cretan, says “All Cretans always lie.”
According to the ancient Greeks, this statement is a paradox because Epimenides can neither tell the
truth nor lie. A sketch of the argument is as follows:
“If Epimenides tells the truth, then all Cretans always lie. Since Epimenides is himself a
Cretan, then he must be lying, which is impossible because we know that Epimenides is
telling the truth. Thus it is not possible for Epimenides to be telling the truth.
If, on the other hand, Epimenides is lying, then his statement is false and all Cretans
never lie. Since Epimenides himself is a Cretan, then he must be telling the truth, which
is impossible because we know that he was lying. Thus it is not possible for Epimenides
to be lying.
Thus Epimenides must be neither lying nor telling the truth – a paradox!”
However, there is a flaw in the above line of reasoning, and despite its name the Epimenides Paradox is
not a paradox.
Identify the flaw in this reasoning. Since this is not really a paradox, Epimenides must either be lying
or telling the truth. Is Epimenides lying or telling the truth? If he's telling the truth, why doesn't his
statement contradict itself? If he's lying, why doesn't his statement contradict itself? Justify your an-
swer.

Problem Two: Propositional Negations (16 points)

For each of the following propositional logic statements, find another statement that is the negation of
the given statement. The statement you choose should only have ¬ directly applied to propositions.
For example, to get the contradiction of the statement
p→q→r
you might use the following line of reasoning:
¬(p → q → r)
p ∧ ¬(q → r)
p ∧ q ∧ ¬r
Once you have found your negation, prove that is is correct by constructing a truth table for the
negation of the original statement and showing it is equal to the truth table for your resulting statement.
For the above case, we would construct truth tables for ¬(p → q → r) and p ∧ q ∧ ¬r as follows:
 p q r ¬(p → q → r) p q r p ∧ q ∧ ¬r
F F F F F T F T F F F F F F FFTF
F F T F F T F T T F F T F F FFF T
F T F F F T T F F F T F F F TTT F
F T T F F T T T T F T T F F TFF T
T F F F T T F T F T F F T F FFTF
T F T F T T F T T T F T T F FFF T
T T F TT F T F F T T F T T TTT F
T T T F T T T T T T T T T F TFF T

Since these truth tables have the same truth values, the formulas are equivalent. The truth tables above
are very detailed and you don't need to provide this level of detail in yours. However, you should at
least specify the truth value of each connective.
i. p ↔ q
ii. r ∨ (¬p ∧ q)
iii. ¬(p → q) → (p → ¬q)

Problem Three: Sufficient Connectives (20 Points)

As we saw in lecture, some propositional connectives can be written in terms of other connectives. For
example, p → q is equivalent to ¬p ∨ q, so it's possible to rewrite all formulas in propositional logic
without ever using → by replacing all instances of φ → ψ with ¬φ ∨ ψ.
This shows that it's possible to drop down from our seven initial quantifiers to just six (¬, ∧, ∨, ↔, ⊤,
and ⊥) without losing any expressive power – any propositional formula that can be written with the
initial seven quantifiers can be rewritten using just six of them. Is it possible to drop down to an even
smaller number? Could we get by with just five quantifiers? Or four? Or even fewer? In this
problem, we will explore the following question:
How many logical connectives are necessary to express all propositional formulas?
It turns out that we can drop down from the six connectives ¬, ∧, ∨, ↔, ⊤, and ⊥ to a set of just five,
since the ↔ operator can be replaced by expressions involving just ¬, ∧, ∨, ⊤, and ⊥ (as you saw in
the checkpoint problem.) From this set of five operators, we can drop down to just three by eliminating
⊤ and ⊥.
i. Find a formula that is logically equivalent to ⊤ that uses only the connectives ¬, ∧, and ∨. You
may introduce other propositional variables if you would like. Prove that your formula is
equivalent to ⊤ by showing that its truth table always evaluates to ⊤.
Given that you now have a new formula for ⊤, you can easily obtain one for ⊥ by just negating the
formula for ⊤. As a result, at this point you've eliminated all connectives except for ¬, ∧, and ∨. But
why stop there? Let's see just how much redundancy there is.
ii. Find a formula that is logically equivalent to p ∨ q that uses just ¬ and ∧. Do not introduce any
new propositional variables. Prove that your formula is equivalent to p ∨ q by showing that its
truth table is the same as the truth table for p ∨ q.
We are now down to just ¬ and ∧, and it might seem like we can't go any further. However, it's
possible to replace these two connectives with just one connective. The nor connective, denoted p ↓ q,
has the following truth table:
p q p↓q
F F T
F T F
T F F
T T F
Notice that the truth table for p ↓ q is the same as the truth table for ¬(p ∨ q), hence the name “nor” (as
in “neither p nor q”)
iii. Find a formula logically equivalent to ¬p that uses just ↓. Do not introduce any new variables.
Prove that your formula is equivalent to ¬p by showing that its truth table is identical to the
truth table for ¬p.
iv. Find a formula logically equivalent to p ∧ q that uses just ↓. Do not introduce any new
variables. Prove that your formula is equivalent to p ∧ q by showing that its truth table is
identical to the truth table for p ∧ q.

You have just shown that every propositional logic formula can be written purely in terms of ↓, since
you can
• Eliminate ↔, →, ⊤, and ⊥ by converting to ¬, ∧, and ∨,
• Eliminate ∨ by converting to ¬ and ∧, and
• Eliminate ¬ and ∧ by converting to ↓.
Because ↓ is powerful enough to derive the rest of the connectives, it is sometimes referred to as a sole
sufficient operator.
An interesting aside: because any propositional formula can be
written entirely in terms of ↓, it is possible to build any digital
circuit using only wires and hardware implementations of ↓ called
NOR gates. Many early computers were assembled with all the
logic done purely in NOR gates. One particularly impressive
such computer was the Apollo Guidance Computer (pictured to
the left; image courtesy of the Computer History Museum), which
helped guide the Apollo spacecraft to safe landings on the moon.
There are actually two sole sufficient operators: the ↓ connective
you just saw, and a second connective call nand whose truth table
is the negation of the truth table for the ∧ connective. Most
modern computers are build primarily of NAND gates rather than
NOR gates, since they're slightly easier to manufacture.
Problem Four: Translating into Logic (32 points)
In each of the following, you will be given a list of first-order predicates and functions along with an
English sentence. In each case, write a statement in first-order logic that expresses the indicated
sentence. Your statement may use any first-order construct (equality, connectives, quantifiers, etc.), but
you must only use the predicates, functions, and constants provided.
i. Given the predicate
Natural(x), which states that x is an natural number,
the function
Product(x, y), which yields the product of x and y,
and the constants 1 and 137, write a statement in first-order logic that says “137 is prime.”
ii. Given the predicates
Word(x), which states that x is a word,
Definition(x), which states that x is a definition, and
Means(x, y), which states that x means y,
write a statement in first-order logic that says “some words have exactly two meanings.”
iii. Given the predicates
x ∈ y, which states that x is an element of y, and
Set(S), which states that S is a set,
write a statement in first-order logic that says “every set has a power set.”
iv. Given the predicates
Lady(x), which states that x is a lady,
Glitters(x), which states that x glitters,
IsSureIsGold(x, y), which states that x is sure that y is gold,
Buying(x, y), which states that x buys y,
StairwayToHeaven(x), which states that x is a Stairway to Heaven,
write a statement in first-order logic that says “There's a lady who's sure all that glitters is gold,
and she's buying a Stairway to Heaven.”*
v. Given the predicates
String(x), which states that x is a string,
Set(S), which states that S is a set,
Regular(L), which states that L is a regular language,
DFA(D), which states that D is a DFA,
Accepts(D, w), which states that D accepts w, and
x ∈ S, which states that x is an element of S,
write a statement in first-order logic that says “If L is a language, then it is a regular language if
and only if there is a DFA that accepts exactly the strings in L.” (If you're starting this problem
set early, we'll define DFAs in Monday's lecture).

* Let's face it – the lyrics to Led Zeppelin's “Stairway to Heaven” are impossible to decipher. Hopefully we can gain
some insight by translating them into first-order logic!
Problem Five: First-Order Negations (12 points)
Proof by contradiction can be difficult because it is often tricky to determine what the negation of the
theorem is. In this problem, you'll use first-order logic to explicitly determine the negation of
statements in first-order logic.
For each of the first-order logic formulas below, find a first-order logic formula that is the negation of
the original statement. Your final formula must not have any negations in it, except for direct negations
of predicates. For example, the negation of
∀x. (p(x) → ∃y. (q(x) ∧ r(y)))
would be found by pushing the negation in from the outside as follows:
¬(∀x. (p(x) → ∃y. (q(x) ∧ r(y))))
∃x. ¬(p(x) → ∃y. (q(x) ∧ r(y)))
∃x. (p(x) ∧ ¬∃y. (q(x) ∧ r(y))
∃x. (p(x) ∧ ∀y. ¬(q(x) ∧ r(y)))
∃x. (p(x) ∧ ∀y. (q(x) → ¬r(y)))
You must show every step of the process of pushing the negation into the formula (along the lines of
what is done above), but you do not need to formally prove that your result is correct.
i. ∃S. (Set(S) ∧ ∀x. x ∉ S)
ii. ∀p. ∀q. (p ∈ ℕ ∧ q ∈ ℕ ∧ q ≠ 0 → q × q × 2 ≠ p × p)
iii. ∀x. (x ∈ ℝ → ∀y. (y ∈ ℝ → (x < y → ∃q. (q ∈ ℚ ∧ x < q ∧ q < y))))

Problem Six: Prenex Normal Form (8 Points)

A statement in first-order logic is said to be in prenex normal form (PNF) iff all the quantifiers in the
formula appear at the very front. For example, the statement
∀x. ∃y. ∀z. (Person(x) → (Person(y) ∧ y ≠ x ∧ (Flowers(z) → Gives(x, y, z))))
is in PNF, but this statement
∀x. (Thing(x) → ∃t. (Time(t) ∧ TimeFor(t, x)))
is not.
Consider the following procedure for converting first-order logic formulas into PNF: directly move all
quantifiers in the formula to the front of the formula, in the order in which they appear. For example,
this procedure would rewrite the formula
∀x. (x ∈ ℝ → ∀y. (y ∈ ℝ → (x < y → ∃q. (q ∈ ℚ ∧ x < q ∧ q < y))))
as
∀x. ∀y. ∃q. (x ∈ ℝ → (y ∈ ℝ → (x < y → (q ∈ ℚ ∧ x < q ∧ q < y))))
Although this procedure will always convert a first-order logic formula into PNF, in some cases it
changes the meaning of the formula. Give an example of a first-order logic formula such that this pro-
cedure yields a formula that is not equivalent to the starting formula. You should justify your answer,
but you don't need to formally prove it.
Problem Seven: SAT Solving Algorithms (24 Points)
Suppose that you are handed a “black-box” SAT-solving algorithm – that is, a device that takes in a
propositional logic formula φ and returns whether or not φ is satisfiable. You don't know anything
about the workings of this algorithm – perhaps it converts the formula to CNF and uses DPLL, or
perhaps it constructs a truth table – but you are assured that given a formula φ it returns whether or not
φ is satisfiable. Let's denote this algorithm A, so A(φ) is true iff φ is satisfiable.
i. Create an algorithm that uses A as a subroutine to determine whether φ is a tautology. You can
describe your algorithm in pseudocode or as a high-level series of steps (such as “first, create
the formula ψ by doing X, then run A on ψ, then do Y”). Then, prove that your algorithm is
correct by showing that your algorithm outputs that φ is a tautology iff φ is indeed a tautology.
Do not list all possible assignments and check each individually; leverage off of algorithm A to
get an answer directly.
ii. Suppose that you have two propositional formulas φ and ψ. You want to see if φ ≡ ψ; that is,
whether φ and ψ always have the same truth values. Create an algorithm that uses A as a
subroutine to answer this question, and prove that your algorithm is correct. As in part (i), you
should not list all possible assignments and should instead leverage A to get the answer directly.

Problem Eight: Course Feedback (5 Points)

We want this course to be as good as it can be, and we'd really appreciate your feedback on how we're
doing. For a free five points, please answer the following questions. We'll give you full credit no
matter what you write (as long as you write something!), but we'd appreciate it if you're honest about
how we're doing.
i. How hard did you find this problem set? How long did it take you to finish? Does that seem
unreasonably difficult or time-consuming for a five-unit class?
ii. Did you attend a recitation section this week? If so, did you find it useful?
iii. How is the pace of this course so far? Too slow? Too fast? Just right?
iv. Is there anything in particular we could do better? Is there anything in particular that you think
we're doing well?

Submission instructions
There are three ways to submit this assignment:
1. Hand in a physical copy of your answers at the start of class. This is probably the easiest way
to submit if you are on campus.
2. Submit a physical copy of your answers in the filing cabinet in the open space near the handout
hangout in the Gates building. If you haven't been there before, it's right inside the entrance
labeled “Stanford Engineering Venture Fund Laboratories.” There will be a clearly-labeled
filing cabinet where you can submit your solutions.
3. Send an email with an electronic copy of your answers to the submission mailing list
( [email protected] ) with the string “[PS4]” in the subject line. If
you do submit electronically, please submit your assignment as a single PDF if at all possi-
ble. Sending multiple files makes it harder to grade your submission.
Extra Credit Problem: Insufficient Connectives (5 Points Extra Credit)
In Problem Three, you proved that the ¬ and ∧ connectives are sufficient to express all propositional
logic formulas with at least one variable.
Prove that there exists a propositional logic formula φ containing at least one variable that cannot be
rewritten as a formula using just the connectives ¬ and ↔.