1 MATH 131AH notes

1. P ROPOSITIONAL LOGIC
Mathematics depends on precise statements for which the common (English) language
is often too inaccurate and at times even plainly ambiguous. For this reason, mathemat-
ical statements are often cast in the dry language (and formalism) of propositional logic.
We thus start by reviewing the important concepts from propositional logic that we will
need throughout the rest of the course.

1.1 Logical propositions and operations.

A basic building block is a proposition which is simply a phrase or a statement (expressed

using the language and objects of the theory) that can be assigned a TRUE/FALSE value.
Examples of propositions is the statement 1 † 0 or the statement
x2 ` 2x ` 1 “ 0 has no solution over R. (1.1)
We will denote propositions P or Q in the sequel.
There are natural operations on propositions that produce new propositions. These
correspond (roughly) to English words
NOT, AND, OR, IMPLIES and IS EQUIVALENT TO (1.2)
For propositions P, Q these are denoted by
P, P ^ Q, P _ Q, P ñ Q and P ô Q, (1.3)
and are referred to as negation, logical conjunction, logical disjunction, implication and
equivalence, respectively. To define these new propositions precisely, we give a list of
their TRUE/FALSE values depending on those of P and Q. This results in

P Q P P^Q P_Q P ñ Q P ô Q
TRUE TRUE FALSE TRUE TRUE TRUE TRUE
TRUE FALSE FALSE FALSE TRUE FALSE FALSE
FALSE TRUE TRUE FALSE TRUE TRUE FALSE
FALSE FALSE TRUE FALSE FALSE TRUE TRUE

which we refer to as the truth table for these operations.

Two points are worthy of a note. First, the OR represented by logical disjunction _
is not what is usually meant by “or” in idiomatic English. Indeed, when a host offers
her guest a “beer or wine,” she presumably means exactly one of the two while the
logical OR — and likely also the thirsty guest — enjoys the possibility of taking both
drinks concurrently. This arises because we often mistake OR for exclusive-OR, a.k.a.
XOR, which is defined as
P XOR Q :“ pP ^ Qq _ p P ^ Qq (1.4)
The logical OR is sometimes translated into English by the phrase and/or but that re-
quires disambiguation with logical AND.
Another point to note is the somewhat unintuitive convention that if P is FALSE, then
P ñ Q is TRUE regardless of the truth value of Q. This subtle point shows the utmost
importance of soundness of logical reasoning: Indeed, it implies that the introduction of

Preliminary version (subject to change anytime!) Typeset: January 11, 2023

MATH 131AH notes 2

a single FALSE statement into the language or theory permits a logical inference of any
statement and, consequently, the collapse of the whole theory.
Caution is at times needed to properly interpret a logical statement written in English
as a statement in propositional logic. The most common situation arises with phrases
“if” and “only if”. These are meant to be read as follows
“P only if Q” means P ñ Q (1.5)
while
“P if Q” means Q ñ P (1.6)
One can easily err when converting the English phrase into the logical proposition so
we should pay extra attention to these in what follows.
The above phrases commonly occur jointly in statements of form “P if and only if Q”
which we take to mean P ô Q. To give an example of the above, the phrase “x is positive
only if x equals it absolute value” translates into
x ° 0 ñ x “ |x| (1.7)
while “x is positive if x equals its absolute value” translates into
x “ |x| ñ x ° 0 (1.8)
Note that the former statement is TRUE for all real-valued x, while the latter is FALSE
for at least one x (namely, x “ 0).
1.2 Primitive operations and equivalent forms.

Not all of the above operations are primitive; in fact, it would suffice to have just NOT
and, for instance, OR. Indeed, with “ ” and “_” defined by the corresponding columns
in the above truth table, we could then set
P ^ Q :“ p P _ Qq
(1.9)
P ñ Q :“ P _ Q
and, with ^ and ñ reduced to the primitives as above, let
P ô Q :“ pP ñ Qq ^ pQ ñ Pq (1.10)
where we employ the convention that the NOT operation is always taken first, unless
parentheses get in the way.
We also note that, here and henceforth,
“:“” means “defined as” (1.11)
where the latter (unlike for ordinary equality) requires only that the object on the right-
hand side is defined; the object on the left is then identified with that on the right (and
so an ordinary equality henceforth applies).
That the first line (1.9) gives the desired concept is verified with the help of the fact
that NOT is an involution,
p Pq ô P (1.12)
and that the de Morgan formulas
pP ^ Qq ô p P _ Qq
(1.13)
pP _ Qq ô p P ^ Qq

Preliminary version (subject to change anytime!) Typeset: January 11, 2023

3 MATH 131AH notes

hold true. The second line in (1.9) is then deduced from the following lemma:
Lemma 1.1 (Proof by contradiction) For any logical propositions P and Q,
pP ñ Qq ô pP ^ Qq (1.14)

Proof. This can be checked directly by working out the truth table of the proposition on
the right-hand side. Alternatively, note that P ñ Q is FALSE only if P is TRUE and Q is
FALSE, which is also the only situation in which P ^ Q is TRUE. ⇤
The title of the lemma refers to the fact that (1.14) is the logical basis for proofs by
contradiction. These demonstrate the implication “P implies Q” to be TRUE by way
of assuming P along with the negation of Q and using these and a sequence of logical
inferences to deduce an absurd (meaning: demonstrably FALSE) conclusion. That being
said, the assumption that Q is FALSE is at times redundant in this argument and its
omission leads to a somewhat different proof strategy:
Lemma 1.2 (Proof by contrapositive) For any logical propositions P and Q,
pP ñ Qq ô p Q ñ Pq (1.15)

Proof. We again readily check that both sides are FALSE only for the situation in which P
is TRUE and Q is FALSE. ⇤
While the unary operation NOT must be included among the primitives, the reliance
on OR as the primitive binary operation is a matter of choice. Indeed, by de Morgan for-
mulas, AND could be used instead and, in light of (1.9), IMPLIES could be used as well.
Note that although 16 distinct binary operations exist on propositions P and Q, they can
all be written as strings of logical disjunctions of logical conjunctions involving P, P,
Q and Q and are thus reducible to NOT and OR above. That being said, the reduction
to primitives is important mainly for theory building, and perhaps implementation on a
computer, so we will keep using the full set above in the sequel.

1.3 Predicates and quatifiers.

Most of mathematics deals with propositions that depend on parameters. Such proposi-
tions are called predicates or Boolean functions. The TRUE/FALSE value of the predicate
Ppxq is generally not determined until x is specified. In order to turn predicates into
proper propositions, we thus need to specify, or curb, x somewhat. This is done using
the following quantifiers
@ :“ for all
(1.16)
D :“ there exists
With the help of these we define propositions
@x : Ppxq :“ Ppxq is TRUE for every choice of x (1.17)
where, alternatively, the phrase on the right means that the Boolean function x ބ Ppxq
has no FALSE in its range. Similarly,
Dx : Ppxq :“ Ppxq is TRUE for at least one choice of x (1.18)

Preliminary version (subject to change anytime!) Typeset: January 11, 2023

MATH 131AH notes 4

which in the language of Boolean functions means x ބ Ppxq has TRUE in its range.
Another word used in the above context is satisfiability. The phrase Dx : Ppxq means
that Ppxq is satisfiable while @x : Ppxq means that Ppxq is NOT satisfiable.
We may often need to further restrict the x for which the quantifier applies. This
requires the language of set theory which, for the sake of easier explanation, we assume
the reader to be familiar with. Given a set E, we write
@x P E : Ppxq :“ @x : x P E ñ Ppxq (1.19)
and
Dx P E : Ppxq :“ Dx : x P E ^ Ppxq (1.20)
Examples of these are the predicates (depending on a set E Ñ R)
@x P E : x2 ° 0 (1.21)
which is TRUE if 0 R E and FALSE otherwise and
Dx P E : x2 ° 0 (1.22)
which is FALSE if E Ñ t0u and TRUE otherwise.
Having both quantifiers around is again done mainly for comfort of expression as one
can be reduced to the other. Indeed, we have:
Lemma 1.3 For any predicate Ppxq,
˘
p@x : Ppxq ô Dx : Ppxq
˘ (1.23)
pDx : Ppxq ô @x : Ppxq

Proof. We will only prove the first equivalence employing the formalism of Boolean
functions. Indeed, @x : Ppxq means that the range of the map x ބ Ppxq is the singleton
tTRUEu. Since Ppxq takes only TRUE/FALSE values, the negation of @x : Ppxq means
that the range of x ބ Ppxq contains FALSE. But this is the same as saying that the range
of x fiÑ Ppxq contains TRUE, which is then identified with Dx : Ppxq. ⇤
Formulas can depend on more than just one parameter some of which may need to be
quantified. We use the convention that while quantifiers are read left to right, they are
applied right to left. Indeed, given a formula Ppx, yq in two variables, we have
´ ¯
@x Dy : Ppx, yq :“ @x : Dy : Ppx, yq (1.24)

and, similarly,
´ ¯
@x @y : Ppx, yq :“ @x : @y : Ppx, yq (1.25)
As with most other operations in mathematics, the order quantifiers are applied matters
for the result. We in fact have:
Lemma 1.4 For any predicate Ppx, yq,
@x@y : Ppx, yq ô @y@x : Ppx, yq (1.26)
and
DxDy : Ppx, yq ô DyDx : Ppx, yq (1.27)

Preliminary version (subject to change anytime!) Typeset: January 11, 2023

5 MATH 131AH notes

and
Dx@y : Ppx, yq ñ @yDx : Ppx, yq (1.28)
We leave the easy proof (based on similar arguments as those in the proof of Lemma 1.3)
to the reader. Note that for mixed quantifiers, only one implication is claimed above be-
cause the other does not hold in general. For instance,
@m P N Dn P N : m “ n (1.29)
is a TRUE statement while the one with the quantifiers reversed is not.
Parentheses have to be used whenever possible ambiguity arises. The position of the
parentheses can of course change the logical content of the statement. For instance,
´ ¯
@n P N : Dm P N : m “ n ñ 0 “ 1 (1.30)
is a TRUE statement yet
´ ¯
@n P N Dm P N : m “ n ñ 0 “ 1 (1.31)
is FALSE. (Here N are the naturals with the usual interpretation of 0 and 1.)

Preliminary version (subject to change anytime!) Typeset: January 11, 2023