You are on page 1of 6

Practice Problem Set

Michael Burton

January 20, 2023

1 Problems
1.1 Basic set theory
Definitions 1. A set A is a subset of a set B (in symbols: A ⊆ B) iff every
member of A is also a member of B. That is, for all u, if u is a member of A (in
symbols: u ∈ A) then u ∈ B. Note that every set is a subset of itself according
to this definition. The empty set ∅ is the unique set with no members; nothing
is a member of the empty set.
Problem 1. Explain why the empty set is a subset of every set, i.e. ∅ ⊆ B for
all sets B.
Definitions 2. The intersection of two sets A and B (in symbols: A ∩ B) is
the set of all things that are members of both A and B. A set is transitive iff it
contains as members all the members of its own members. That is, a transitive
set has as members all of its “grand-members”. If A is transitive, then

x ∈ y ∈ A ⇒ x ∈ A,

i.e. if x is a member of y and y is a member of A, then x itself is a member of


A.
Problem 2. Show that the intersection of two transitive sets is itself a transitive
set.

1.2 Classical Sentential Logic


Notation. If v is an assignment and ϕ a formula, we write v(ϕ) = 1 to mean “v
assigns ϕ to the value 1” and likewise for v(ϕ) = 0. This notation makes sense;
recall that v is really a function that maps formulas to {0, 1}. This notation is
used below.
Problem 3. A formula is valid (according to classical sentential logic) iff every
assignment makes it true. Show that this is equivalent to the formula being a
consequence of the empty set.

1
Problem 4. Really prove that

|= (ϕ ∧ ¬ϕ) → ψ,

where |= is logical consequence as defined by classical sentential logic, and ϕ


and ψ are arbitrary formulas.
Problem 5. (Explosion in classical sentential logic.) Let Γ be any theory.
(Recall that a theory is a set of formulas that is closed under consequence, so
Γ |= ϕ implies ϕ is a member of Γ, in symbols ϕ ∈ Γ.) Suppose (ϕ ∧ ¬ϕ) ∈ Γ,
for some ϕ. Show that ψ ∈ Γ for every ψ, i.e. Γ contains every formula. (Hint:
first prove that, if (ϕ → ψ) ∈ Γ and ϕ ∈ Γ, then ψ ∈ Γ, if Γ is a theory.
Problem 6. The method of truth tables can be used to determine whether
a formula is valid in classical sentential logic (i.e. is true according to all as-
signments) or not. If the formula contains n unique sentence symbols, the truth
table for the formula will contain 2n rows. It is as if we only consider all possible
partial assignments from those n unique sentence symbols to the values 0 and
1 (representing true and false). Prove, using induction on (the construction
of) formulas, that if two assignments v and u “agree on” the value of each of the
sentence symbols occurring in ϕ, then v(ϕ) = u(ϕ), i.e. the assignments agree
on the value of ϕ itself.

1.3 Classical First-order Logic


Set-up. Consider the following (silly) structure A whose signature only has two
predicates, one unary (M ) and one binary (S):
domain of A = {Alice, Bob, Charlie}
M A = {Alice, Bob}
S A = {hAlice, Bobi, hBob, Charliei}
Note that M A is a set of members of the domain, i.e. a set of objects, i.e. a
subset of the domain, whereas S A is a set of ordered pairs of objects.
Problem 7. Is the following sentence true in A?

∃x∃y(M x ∧ (¬M y ∧ Sxy))

You can explain why or why not in a nonformal or intuitive way.


Problem 8. How would you “translate” the (English) argument about Socrates’
mortality into the (formal) language of first-order logic?
Problem 9 should be done before Problem 10.
Definition. Say that two variable assignments “agree” with respect to the term
t iff they agree at every variable in t. For example, if the term t is

f (x, y)

2
where f , of course, is a binary function symbol, then s1 and s2 agree with respect
to t iff s1 (x) = s2 (x) and s1 (y) = s2 (y), i.e. the two variable assignments map
x to the same object and y to the same object.
Below, we work directly with variable assignments and imagine that we have
some fixed structure A in mind.
Problem 9. Show that, if s1 and s2 agree with respect to t, then s1 (t) = s2 (t),
i.e they agree at t. Use induction on (the construction of) terms.
Problem 10. Let ϕ be a formula. Suppose s1 and s2 agree at every variable
that is free in ϕ. Prove that s1 satisfies ϕ iff s2 satisfies ϕ. Use induction
on (the construction of) formulas. Use the result of Problem 9. Note that this
shows that all variable assignments “agree” on all sentences; sentences are those
formulas with no free variables, recall.

3
2 Answers
Answer to Problem 1. Let A be any set. Suppose instead that the empty set
is not a subset of A. Then not every member of the empty set is a member of
A. This means there must be something with the following two properties: (1)
it is a member of the empty set and (2) it is not a member of A. But nothing
has property (1); no thing is a member of the empty set. So, nothing has both
properties. So, we reject what we supposed: the empty set is a subset of A.
And A was arbitrary, so every set has the empty set as a subset.
Answer to Problem 2. Let A and B be arbitrary transitive sets. To show that
A ∩ B is itself transitive, we must show that, for every y ∈ A ∩ B and x ∈ y,
x ∈ A ∩ B. Let y be an arbitrary member of A ∩ B and let x be an arbitrary
member of y. Thus, y ∈ A and y ∈ B (from definition of intersection). Thus,
x ∈ A and x ∈ B (because x ∈ y and y ∈ A and A is transitive; same for B).
Thus, x ∈ A ∩ B (from definition of intersection).
Answer to Problem 3. We are asked to prove that a formula is valid iff it
is a consequence of the empty set. We prove both directions in turn. First,
assume that ϕ is valid. Thus, every assignment makes ϕ true. Thus, every
assignment that makes every member of the empty set true makes ϕ true, i.e. |=
ϕ. (Key point: the empty set has no members, so every assignment makes every
one of its members true. This is an example of something being “vacuously”
true.) Second, assume that ϕ is a consequence of the empty set. Thus, every
assignment that makes every member of the empty set true makes ϕ true. As
noted above, this means every assignment makes ϕ true, i.e. ϕ is valid.
Answer to Problem 4. This problem asks us to prove that
(ϕ ∧ ¬ϕ) → ψ
is a consequence of the empty set. As we showed in the previous answer, this
just requires showing that it is true according to every assignment. Let v be an
arbitrary assignment. If v(ϕ) = 1, then v(¬ϕ) = 0 (from semantics for ¬) and
so v(ϕ ∧ ¬ϕ) = 0 (from semantics for ∧). Thus, every assignment makes every
contradiction false. From the semantics for →, this means that
v((ϕ ∧ ¬ϕ) → ψ) = 1
and therefore the formula in question is, indeed, true according to any assign-
ment.
Answer to Problem 5. We follow the hint. Suppose
Γ |= ϕ → ψ
and
Γ |= ϕ.
We wish to show now that Γ |= ψ. Let v be an arbitrary assignment that makes
every member of Γ true. Thus, by the suppositions, v makes ϕ → ψ true and

4
ϕ true. Thus, by the semantics for →, v makes ψ true. Thus, because v was
arbitrary, every assignment that makes every member of Γ true makes ψ true,
i.e. Γ |= ψ. We now note that every theory contains every valid formula: if
|= χ, then Γ |= χ, for any χ, any Γ. Thus, we have both (ϕ ∧ ¬ϕ) ∈ Γ and
((ϕ ∧ ¬ϕ) → ψ) ∈ Γ. Thus, we have ψ ∈ Γ by what we showed following the
hint. And ψ was arbitrary, so every formula is a member of Γ.
Answer to Problem 6. We have a single base case to consider, when ϕ is just
a sentence symbol. We will consider just one inductive case. The others are
similar.
1. If ϕ is just a sentence symbol, say, p, then v(p) = u(p) as we assume v
and u agree on every sentence symbol in ϕ, which here is just to say that
they agree on p, i.e. ϕ itself.
2. Suppose ϕ is ψ1 ∧ψ2 . Our inductive hypothesis is that, for any assignments
v, u, if they agree on all sentence symbols in ψ1 , they agree with respect
to ψ1 , i.e. v(ψ1 ) = u(ψ1 ), and the same holds for ψ2 , as well. Then we
have: v(ψ1 ∧ ψ2 ) = 1 iff v(ψ1 ) = 1 and v(ψ2 ) = 1 (semantics) iff u(ψ1 ) = 1
and u(ψ2 ) = 1 (inductive hypothesis) iff u(ψ1 ∧ ψ2 ) = 1 (semantics).
3. ...
The cases for ¬, ∨, →, and ↔ are very similar to the case for ∧ above.
Answer to Problem 7. This sentence is true in this structure iff there exist two
objects such that (1) the first “is M ”, (2) the second “is not M ”, and (3) the
first “does S to” the second. In fact, we can find such a pair of objects: Bob
and Charlie. Bob is M , Charlie is not M , and Bob does S to Charlie. You can
verify this by inspecting the definition of the structure in the statement of the
problem above. As an aside, note that the sentence is still true if we remove
Bob from the interpretation of M , so only Alice “is M ”. The new pair would
be Alice and Bob.
Answer to Problem 8. There are other ways, but the following is the most
straightforward. We let the predicate H stand for “is human”, the precidate M
for “is mortal” and the constant s for “Socrates”. Then we have:
1. ∀x(Hx → M x) (All humans are mortal; if human, then mortal.)
2. Hs (Socrates is human.)
3. M s (Socrates is mortal.)
Answer to Problem 9. We have two base cases to consider: variable and con-
stant. We have one case in the inductive step: functional terms.
1. When t is simply a variable x, then we have s1 (x) = s2 (x) and therefore
s1 (x) = s2 (x). (Recall the definition of s̄ from s on the handout.)
2. When t is a constant c, then we have s1 (c) = s2 (c). Every pair of vari-
able assignments agrees with respect to every constant c, as there are no

5
variables in a constant, of course.
3. Now, consider when t is a functional term, i.e. of the form

f (t1 , ..., tn )

for some terms t1 etc. This is the inductive step. Our inductive hypothesis
is that the claim holds for t1 etc. That is, we assume s1 (t1 ) = s2 (t1 ), for
example.1 Now, looking back at the handout, we can see the following:

s1 (f (t1 , ..., tn )) =f A (s1 (t1 ), ..., s1 (tn )) (1)


A
=f (s2 (t1 ), ..., s2 (tn )) (∗) (2)
=s2 (f (t1 , ..., tn )) (3)

Note that in step (∗) we use the inductive hypothesis when we substitute
identical terms, s2 (t1 ) for s1 (t1 ) etc.
Answer to Problem 10. We have one base case, when ϕ is an atomic formula.
We consider two inductive cases only; the rest are similar to one of these.
1. Suppose ϕ is P t1 ...tn . Note that every variable in ϕ is free as there are no
quantifiers in this formula. Thus, if s1 and s2 agree on all free variables
in ϕ (as they in fact do), then they agree with respect to t1 etc. We
can therefore use the result of Problem 9 here. We have then that s1
satisfies ϕ iff hs1 (t1 ), ..., s1 (tn )i ∈ P A (semantics) iff hs2 (t1 ), ..., s2 (tn )i ∈
P A (Problem 9 justifies this equivalence) iff s2 satisfies ϕ (semantics).
2. Suppose ϕ has the form ψ1 ∧ ψ2 . Our inductive hypothesis is that the
claim holds for ψ1 and ψ2 . Thus s1 satisfies ψ1 ∧ ψ2 iff s1 satisfies ψ1
and s1 satisfies ψ2 (semantics) iff s2 satisfies ψ1 and s2 satisfies ψ2 (in-
ductive hypothesis, see the footnote for Problem 9) iff s2 satisfies ψ1 ∧ ψ2
(semantics).
3. ...
4. Suppose ϕ has the form ∀xψ. Our inductive hypothesis is that the claim
holds for ψ. We should emphasize something at this point: the inductive
hypothesis is that, for any two variable assignments—call them s∗ and
s∗∗ —if they agree at every variable free in ψ, then s∗ satisfies ψ iff s∗∗
satisfies ψ. Now, s1 satisfies ∀xϕ iff, for all objects d, (s1 )xd satisfies ψ
(semantics) iff, for all objects d, (s2 )xd satisfies ψ (∗) iff s2 satisfies ∀xψ.
Note that the step marked (∗) is justified by the inductive hypothesis. If
s1 and s2 agree as we’ve assumed they do, then of course (s1 )xd and (s2 )xd
agree as well at the variables free in ψ (which may now include x).
5. ...

1 Strictly speaking, what we actually assume is that this equality holds if s and s agree
1 2
with respect to t1 . But of course they do, because they agree with respect to f (t1 , ..., tn ), and
this term contains at least all of the variables of, for example, t1 .

You might also like