Physical Address

304 North Cardinal St.
Dorchester Center, MA 02124

Nested Quantifiers

In Section 7 of this chapter we examined examples of propositional functions with two and three variables. For each variable in an open statement, we needed to substitute some value from the respective universe in order to determine the truth value.

Here, we look at quantifying statements with two or more variables.

Bound and Free Variables

Bound Variable, Free Variable

For some universe 𝒰, consider a propositional function p(x, y) where x and y are both constrained by universe 𝒰.

Consider the following quantified statements:

∀x [p(x, y)]
∃x [p(x, y)].

In both statements since x is bound by a quantifier, and y is not bound by any quantifier, we say that x is a bound variable, whereas y is a free variable.

Up to now, we’ve only dealt with quantified statements that have only one variable. As such, we would have considered those variables bound by whatever quantifier was used. In all of those examples, every quantified statement had a definite truth value.

In our definition, we specify that any variable not bound by a quantifier is called a free variable. As such, statements such as

∀x [p(x, y)]
∃x [p(x, y)]

do not have definite truth variables. In both cases, we still need to either bound variable y by a quantifier, or substitute in a value from 𝒰 in order for the truth value to be determined. This is important enough to highlight.

Quantified Statements with Free Variables Are Open Statements

For some universe 𝒰, consider a propositional function p(x, y) where x and y are both constrained by universe 𝒰.

Consider the following quantified statements:

∀x [p(x, y)]
∃x [p(x, y)].

The truth value of both statements is undetermined. As such, they are open statements whose truth values depend on what values are substituted in for variable y.

Example 1.10.1

Consider the universe N of all integers along with the following open statement:

p(m, n):m + n = 0

What happens when we quantify the m variable? With the existential quantifier, we get

∀m [p(m, n)]

which can be translated to “For every integer m, we have that m + n = 0.” Is this true? We would need to know what value n holds to be certain. We can’t just pick any value of n. The above statement would be false when m = 1 and n = 2. We just can’t be sure. As such, ∀m [p(m, n)] does not have a truth value.

Now consider the statement

∃m [p(m, n)]

which can be translated to “There exists an integer m such that m + n = 0.” Again, until a value for n is known, we can’t be sure. Of course, we could figure out what value n would need to be in order for m + n = 0. For example, when m = 5, we could make n = -5. But again, the value of n is unknown. As such, the statement ∃m [p(m, n)] is open, and has no definite truth value.

Repeated Quantifiers

In order for a quantified statement to have a definite truth value, there can be no free variables. We can bind all variables to the same kind of quantifier, which will be useful as there are many kinds of mathematical statements we can make that rely on more than one variable.

Example 1.10.2

A basic law of arithmetic is the commutative law, where, in a sum or product, the two constituent parts can be swapped without affecting the result. In other words, for any two real numbers x and y, we have that

x + y = y + x.

Consider the universe 𝒰 of all real numbers. We can express the commutative laws of arithmetic using quantifiers like this:

∀x∀y [x + y = y + x]
∀x∀y [x * y = y * x]

The order of the quantified variables does not matter, so we could also express the commutative laws of arithmetic like this:

∀y∀x [x + y = y + x]
∀y∀x [x * y = y * x]

Example 1.10.3

Consider the universe N consisting of all integers.

Consider the statement

“The integer 100 is the sum of two perfect squares.”

The presence of the definite article the suggests that the existential quantifier is the implied quantifier. As such, we could express the statement using quantifiers like this:

∃m∃n [100 = m2 + n2].

Again, the order of the quantification does not matter in this case, so we could also write

∃n∃m [100 = m2 + n2].

One more way to write this is to specify which universe each variable belongs to using the ∈ symbol discussed on this page:

∃m∈N ∃n∈N [100 = m2 + n2].

Notice that when the same quantifier is used, the order in which we express the quantification did not matter, as such, we have our first logical equivalencies.

Order Doesn’t Matter in Repeated Quantifiers

Consider a universe 𝒰 for some open statement p(x, y) with variables x and y.

∀x ∀y [p(x, y)] ⟺ ∀y ∀x [p(x, y)]
∃x ∃y [p(x, y)] ⟺ ∃y ∃x [p(x, y)]

This also applies to open statements involving three, four, or more variables.

There is a short-hand notation that is commonly used on repeated quantifiers. To demonstrate it’s use, we of course start with some universe 𝒰 where the variables will take on values. Consider the open statement p(x, y). We can use the following conventions:

∀x ∀y [p(x, y)] ⟺ ∀x, y [p(x, y)]
∃x ∃y [p(x, y)] ⟺ ∃x, y [p(x, y)].

Naturally, we can extend this to three variables. If we consider the open statement q(a, b, c), then we can use

∀a ∀b ∀c [q(a, b, c)] ⟺ ∀a, b, c [q(a, b, c)]
∃x ∃y ∃y [q(a, b, c)] ⟺ ∃a, b, c [q(a, b, c)].

Of course, these can be extended to as many variables as needed.

Example 1.10.4

Consider the universe of all real numbers, which we’ll denote R.

When dealing with real numbers, another commonly used arithmetic law is the distribution of multiplication over addition. Typically, this is expressed as follows:

“For every real number x, y, and z, we have that x(y + z) = xy + xz.”

Here, the quantifier being used is the universal quantifier, and it is being applied to all variables, so we can concisely represent this statement with the following proposition:

∀x ∀y ∀z [x(y + z) = xy + xz].

Using the short-hand notation discussed above, we could also write this as

∀x, y, z [x(y + z) = xy + xz].

Mixed Quantifiers

Some statements involve both the existential quantifier and the universal quantifier.

Example 1.10.5

Consider the universe consisting of all real numbers.

It is known that every number has an additive inverse. What this basically means is this:

“For every real number x, there exists some other number y where x + y = 0.”

Notice that we used the phrases “For every….” and “There exists….”, which suggests both a universal quantifier and an existential quantifier. The phrase “For every…” comes first, suggesting it is the first quantifier, and so the existential quantifier comes second. As such we can rewrite the above statement as

∀x ∃y [x + y = 0].

Since x is universally quantified, we can pick arbitrary values to test and see if this works.

If x = 3008, then we know that y = -3008 will work.

Similarly, taking x = -π2, we can set y = π2.

Indeed, as soon as a value of x is selected, we can just take y = -x, because since x is a real number, so is -x. Thus, no matter what real number we pick for x, we can always find an appropriate value for y.

Example 1.10.6

Let’s reconsider Example 1.10.5 regarding additive inverses.

Suppose we swapped quantifiers like we did when we examined repeated quantifiers. In doing so, we form the following statement:

∃y ∀x [x + y = 0].

This proposition can be translated into English as “There exists some real number y, such that for every real number x, we have that x + y = 0.”

Here, the proposition is asserting the existence of some such real number. Then, once that real number is determined, supposedly, every real number added to it will yield a sum of 0.

We know that y can’t be 17, because 3 + 17 ≠ 0 (here, y = 17 and x = 3.)

Similarly, we know that y can’t be -0.745 because 20 – 0.745 ≠ 0 (here, y = -0.745 and x = 20.)

As a matter of fact, no such value of y can exist, because as soon as we pick a value for y, there are infinitely many ways to pick the value of x (since x is universally quantified, the statement asserts that every value of x will work, which so far is not the case.) Since there are infinitely many values to substitute in for x, there are infinitely many sums.

Out of those infinitely many sums, only one sum will be 0. That happens when we pick x to be equal to -y. Of course, since x is universally quantified, we could also pick the real number (-y + 1), which will yield a sum of 1, and not 0.

Surprisingly, if this statement were true, every single real number would need to be equal to 0, an obviously ridiculous scenario!

Based on Example 1.10.5 and Example 1.10.6, we know that we can’t just swap mixed quantifiers arbitrarily. Perhaps under very specific circumstances, it could be done, but in general, that is not something we can do. This is worth highlighting.

Mixed Quantifiers are Generally Unswappable

Consider a given universe 𝒰, and open statement p(x, y) defined on that universe.

In general, we can’t swap mixed quantifiers, and so they are not logically equivalent:

∀x ∃y [p(x, y)] ⇐/⇒ ∃y ∀x [p(x, y)].

As such, we need to exercise caution regarding the order of quantifiers.

Negating Multiple Quantifiers

One more thing to talk about is how to negate propositions involving multiple quantifiers.

For a given universe 𝒰, and an open statement p(x) defined on that universe, we have that

¬∀x [p(x)] ⟺ ∃x [¬p(x)]
¬∃x [p(x)] ⟺ ∀x [¬p(x)]

Figure 1.10.1: A reminder on how to negate quantified statements.

Recall that from Section 9 of this chapter, we noted how to negate propositions involving only a single quantifier. Figure 1.10.1 to the left gives us a reminder of how to do so.

In essence, when we try to negate a multiply quantified statement, we just need to keep punting the negation inside the quantified statement, until all that’s left to do is negate the contained proposition like so:

¬∀x ∀y [p(x, y)]Laws of Logic Used
∃x [¬∀y [p(x, y)]]¬∀x [p(x)] ⟺ ∃x [¬p(x)]
∃x ∃y [¬p(x, y)]¬∀x [p(x)] ⟺ ∃x [¬p(x)]

So, after punting the negation inside each nested layer of the quantified statement, we determined that whatever open statement p(x, y) represents, we have that

¬∀x ∀y [p(x, y)] ⟺ ∃x ∃y [¬p(x, y)].

The strategy of punting the negation further inside each nested layer will work for the other forms of quantified statements as well, so we summarize the important combinations in a highlight.

Negating Multiple Quantifiers

For a given universe 𝒰, and an open statement p(x, y) defined on that universe, we have that

¬∀x ∀y [p(x, y)] ⟺ ∃x ∃y [¬p(x, y)]
¬∀x ∃y [p(x, y)] ⟺ ∃x ∀y [¬p(x, y)]
¬∃x ∀y [p(x, y)] ⟺ ∀x ∃y [¬p(x, y)]
¬∃x ∃y [p(x, y)] ⟺ ∀x ∀y [¬p(x, y)]

Example 1.10.7

Suppose we have universe 𝒰, and that the open statements a(x, y), b(x, y), and c(x, y) are defined on that universe.

We want to negate the following quantified statement:

∃x ∀y [a(x, y) ∧ b(x, y) → c(x, y)].

Before we start, we can make the following short-hand substitutions.

a(x, y): a
b(x, y): b
c(x, y): c

We do this to make the proposition a little less messy:

∃x ∀y [(a ∧ b) → c].

Now we perform the negation:

¬∃x ∀y [(a ∧ b) → c]Law of Logic Used
∀x ∃y [¬[(a ∧ b ) → c]]¬∃x ∀y [p(x, y)] ⟺ ∀x ∃y [¬p(x, y)]
∀x ∃y [¬[¬(a ∧ b ) ∨ c]]p → q ⟺ ¬p ∨ q
∀x ∃y [¬¬(a ∧ b ) ∧ ¬c]DeMorgan’s Laws
∀x ∃y [(a ∧ b ) ∧ ¬c]Law of Double Negation
∀x ∃y [a ∧ b ∧ ¬c]Associative Law of ∧

Substituting a(x, y), b(x, y), c(x, y) back in for a, b, and c respectively, we get the negation of

∃x ∀y [a(x, y) ∧ b(x, y) → c(x, y)]

is the statement

∀x ∃y [a(x, y) ∧ b(x, y) ∧ ¬c(x, y)].

A Final Word on Propositions in General

Throughout this chapter, we have spent a great deal of time working with logical expressions in abstract ways. We’ve modeled them with truth tables to see when they are satisfiable, and we’ve used logical equivalencies to simplify complicated propositions, as well as to expand simple propositions into complex propositions. We’ve also examined how work with variables, and what effects quantifying those variables has on the truth value of a proposition.

Remember that all of the letters we have used in our logical expressions, including x, y, z, as well as a, b, c, and even p, q, r, s, and t all represent statements. Those statements could be primitive as in

“Thomas Jefferson was the second president of the United States”
“2 + 2 = 4”
“Quadrilateral ABCD has a right angle”

or they could be non-primitive as in

“If the sun is shining, and Mr. Wilson is not fishing at the lake, then Dennis will take his R.C. boat out to the lake.”

The purpose was to get a grip on mathematical logic. Translating English sentences representing mathematical statements into a symbolic form gives us the ability to manipulate those statements using the logic studied in this chapter.

Note that translating a mathematical statement into symbolic form forces us to be precise in our use of language in describing the exact problem we want to solve. Any misunderstanding will invalidate any reasoning ability we may be able to employ because at that point, the intended problem is no longer being discussed. Instead, a (even if slightly) different problem is being reasoned on.

Having these logical tools at our disposal, we are better able to accurately use language to precisely describe a problem of interest, and to then use the logical laws to analyze the problem of interest. Mathematical logic is the foundation of all mathematics to be learned, and will be employed heavily throughout the remainder of this book and in future books.