This deck covers Cantor's diagonal argument that the reals are uncountable and Cantor's theorem that every set is strictly smaller than its power set, then the cardinal arithmetic of the continuum, the Continuum Hypothesis, and its independence from ZFC. It ends with the collision between set theory and computation: there are only countably many programs but uncountably many reals, so most reals must be uncomputable. It targets the misconceptions that the diagonal number is already in the list, that uncountable simply means infinite, that the Continuum Hypothesis is merely unproven rather than independent, and that a bijection between the line and the plane is impossible.
Subject: Foundations of Higher Mathematics · 106 slides · symbolic lesson
Open the interactive version of this deck · Homework for this lesson
Objectives
By the end of this deck you can:
1. Run Cantor's diagonal argument to prove the real numbers are uncountable.
2. Prove Cantor's theorem: every set is strictly smaller than its power set, with no fixed point.
3. Do basic cardinal arithmetic and identify the cardinality of the continuum.
4. State the Continuum Hypothesis and say precisely what its independence from ZFC means.
5. Explain why most real numbers are uncomputable, and recognize the halting problem as the same diagonal argument.
Warm-up
Discussion prompt
Before we open Cardinality II: Uncountability & the Continuum: without looking back, what was the main idea of Cardinality I: Countable Sets, and what could you do by the end of it that you could not do before?
Hint: One sentence for the idea, one for the skill. If the second one is blank, that is the part to revisit.
Answer:
Comparing infinite sizes by bijection: equinumerosity, Dedekind-infinite sets, and countability. We build explicit bijections for the naturals, integers, pairs, and rationals, prove countable unions stay countable, and use Schroeder-Bernstein.
Concept
From the previous deck: a set is countable when it is finite or can be put in bijection with the natural numbers.
countable — A set A is countable if there is a bijection between A and some initial segment of the naturals, or with all of the naturals. Equivalently, A can be arranged as a (possibly infinite) list with no repeats missed.
We proved the integers and even the rationals are countable. The natural next question: is every infinite set countable?
Counterexample
Discussion prompt
From the previous deck: a set is countable when it is finite or can be put in bijection with the natural numbers.
That is stated as though it always holds. Do one of two things: produce a case where it fails, or say precisely what rules such a case out. "It just does" is not on the menu.
Hint: Hunt at the extremes first — zero, one, negative, empty, equal. If every extreme survives, the reason they survive is the proof.
Answer:
We proved the integers and even the rationals are countable. The natural next question: is every infinite set countable?
Intuition
To say a set is countable is to say you can enumerate it: a first element, a second, a third, forever.
An enumeration is exactly a surjection from the naturals onto the set. Row n of the list is the image of n.
\[ e : \mathbb{N} \twoheadrightarrow A, \qquad e(n) = \text{the } n\text{-th listed element} \]
So uncountable means the exact opposite: no matter what list you write down, it must miss something. Every enumeration fails to be onto.
Analogy
Discussion prompt
Explain A list is a function from the naturals by analogy to something with no Foundations of Higher Mathematics in it at all — a queue, a recipe, a map, a bank balance, whatever fits. Then say where your analogy breaks.
Hint: An analogy that never breaks is not an analogy, it is the same idea wearing a hat. Find the seam — that is the part that is actually new.
Answer:
To say a set is countable is to say you can enumerate it: a first element, a second, a third, forever.
Concept
We will show one concrete set is uncountable, and everything else follows. The cleanest target is the open unit interval.
\[ (0,1) = \{\, x \in \mathbb{R} : 0 < x < 1 \,\} \]
If this small interval already cannot be listed, then the whole real line certainly cannot either.
Explain it
Discussion prompt
Explain The target: the open interval to a student a year behind you. No notation, no jargon they have not met — and it still has to be true.
Hint: If your explanation needs a symbol they have never seen, you are describing the notation rather than the idea.
Answer:
If this small interval already cannot be listed, then the whole real line certainly cannot either.
Concept
Every number in the interval has an infinite decimal expansion after the point.
\[ x = 0.d_1 d_2 d_3 d_4 \ldots, \qquad d_k \in \{0,1,\ldots,9\} \]
The digit in position k is a function of x. This lets us talk about a real number one digit at a time, which is exactly what the diagonal argument needs.
Concept
Decimal expansions are almost unique. The only clashes are a tail of nines versus a tail of zeros.
\[ 0.4999\ldots = 0.5000\ldots \]
A single real can have two decimal names precisely when one ends in repeating nines. We will build our diagonal number to dodge this entirely, and revisit the point as a trap.
Picture it
Figure (svg): A four by four grid of decimal digits for r1 through r4. The diagonal cells, holding digits 1, 2, 5, 3, are boxed. Below, the constructed number x equals 0.5565, formed by changing each boxed diagonal digit.
Discussion prompt
Read the picture before the words. What is this showing, and what is the one thing it is built to make obvious? Commit to an answer, then read on.
Hint: Name the parts, then say what changes between them — and if nothing changes, say what is being held still.
Answer:
Suppose, for contradiction, that the interval is countable. Then its elements can be listed, and each row is an infinite decimal.
Worked example
Suppose, for contradiction, that the interval is countable. Then its elements can be listed, and each row is an infinite decimal.
\[ \begin{aligned} r_1 &= 0.\,\mathbf{1}\,3\,7\,5\ldots \\ r_2 &= 0.\,4\,\mathbf{2}\,6\,8\ldots \\ r_3 &= 0.\,8\,1\,\mathbf{5}\,2\ldots \\ r_4 &= 0.\,3\,3\,3\,\mathbf{3}\ldots \end{aligned} \]
Figure (svg): A four by four grid of decimal digits for r1 through r4. The diagonal cells, holding digits 1, 2, 5, 3, are boxed. Below, the constructed number x equals 0.5565, formed by changing each boxed diagonal digit.
Read the diagonal digits
Why: Take digit k from row k. That is the k-th digit of r_k, running down the highlighted diagonal.
\[ \text{diagonal}: \; d_1 = 1,\; d_2 = 2,\; d_3 = 5,\; d_4 = 3,\; \ldots \]
Change every diagonal digit by a fixed safe rule
Why: Define a new number x whose k-th digit disagrees with the k-th digit of row k. Use digits only from a safe pair to avoid nines and zeros.
\[ x_k = \begin{cases} 5 & \text{if the diagonal digit } d_k \neq 5 \\ 6 & \text{if } d_k = 5 \end{cases} \]
Write out the new number x
Why: Applying the rule to 1, 2, 5, 3 gives 5, 5, 6, 5. Every digit of x is a 5 or a 6, so x lies in the interval.
\[ x = 0.5\,5\,6\,5\ldots \]
Verify x is on the list nowhere
Why: By construction x disagrees with row k in digit k, for every k. So x is not r_1, not r_2, not r_3, and so on. The list was assumed complete, yet it misses x. Contradiction: the interval is not countable.
\[ \forall k \in \mathbb{N}: \; x_k \neq (r_k)_k \;\Rightarrow\; x \neq r_k \]
Notation
Annotate
From Cantor's diagonal argument on the interval — read this one piece at a time. What is each part doing?
On: \( \begin{aligned} r_1 &= 0.\,\mathbf{1}\,3\,7\,5\ldots \\ r_2 &= 0.\,4\,\mathbf{2}\,6\,8\ldots \\ r_3 &= 0.\,8\,1\,\mathbf{5}\,2\ldots \\ r_4 &= 0.\,3\,3\,3\,\mathbf{3}\ldots \end{aligned} \)
Concept
Two decimals that differ in even one digit are different numbers, once we have banned the nines-versus-zeros loophole.
The diagonal number is engineered to differ from row k in position k. One guaranteed disagreement per row is all it takes to be a different number from that row.
\[ x_k \neq (r_k)_k \;\Longrightarrow\; x \neq r_k \quad \text{for every } k \]
The single object x is simultaneously different from every listed number. No list can contain it, so no list is complete.
Pattern
1. Assume a complete list
Why: Suppose the set is countable, so every element appears as some row indexed by a natural number.
2. Build an object from the diagonal
Why: Define a new element whose k-th feature is read from row k, then deliberately changed.
3. Guarantee disagreement in row k at coordinate k
Why: The change ensures the new object differs from row k in the k-th slot, so it cannot equal row k.
4. Conclude the list was incomplete
Why: The new object is a genuine member yet appears nowhere, contradicting completeness. The set is uncountable. This one recipe also proves Cantor's theorem and the halting problem.
Real world
Discussion prompt
Outside this lesson: where does Cardinality II: Uncountability & the Continuum actually turn up? Name one concrete situation — a job, a piece of software someone ships, a decision somebody has to make — and say which part of The diagonalization recipe is doing the work in it.
Hint: Vague is the failure mode here. "Engineering" is not a situation; "deciding whether this build is fast enough to ship" is.
Answer:
That deck covers Cantor's diagonal argument that the reals are uncountable and Cantor's theorem that every set is strictly smaller than its power set, then the cardinal arithmetic of the continuum, the Continuum Hypothesis, and its independence from ZFC. It ends with the collision between set theory and computation: there are only countably many programs but uncountably many reals, so most reals must be uncomputable. It targets the misconceptions that the diagonal number is already in the list, that uncountable simply means infinite, that the Continuum Hypothesis is merely unproven rather than independent, and that a bijection between the line and the plane is impossible.
Elimination
Eliminate the wrong options
What are the first three digits of the constructed diagonal number x?
3 of these 4 are wrong. Strike them one at a time, and say what rules each one out before you strike the next. The survivor is the answer.
Survives elimination: A
Why: The diagonal digits are 5, 2, 7 (position k from row k). Apply the rule: 5 is already 5 so it becomes 6; 2 becomes 5; 7 becomes 5. So x starts 0.6 5 5, differing from each row in its own position.
Check
Use the same rule as the worked example: a diagonal digit becomes 5 unless it already is 5, in which case it becomes 6.
\[ \begin{aligned} r_1 &= 0.\,\mathbf{5}\,3\,8\ldots \\ r_2 &= 0.\,2\,\mathbf{2}\,1\ldots \\ r_3 &= 0.\,9\,4\,\mathbf{7}\ldots \end{aligned} \]
Check your understanding
What are the first three digits of the constructed diagonal number x?
Answer: A
Why: The diagonal digits are 5, 2, 7 (position k from row k). Apply the rule: 5 is already 5 so it becomes 6; 2 becomes 5; 7 becomes 5. So x starts 0.6 5 5, differing from each row in its own position.
Fill the middle
Fill in the blanks
From Trap: the diagonal number is further down the list — finish the line. Write what belongs on the right of the equals sign before you look.
\exists\, k:\; x = r_k \; ?
Why: Producing the right-hand side unprompted is the difference between recognising this line and being able to use it. If someone claims x is row k, look at digit k: x was built so that its k-th digit differs from the k-th digit of row k.
Trap
A tempting objection: sure, x is not among the first few rows, but the list is infinite. Surely x appears eventually, maybe as row one million.
The objector imagines x is really in the list, just at some large unspecified position.
\[ \exists\, k:\; x = r_k \; ? \]
It cannot be at any position. The construction guarantees x differs from row k in digit k, and this holds for k equal to one million just as much as for k equal to one.
Pin the disagreement at the claimed position
Why: If someone claims x is row k, look at digit k: x was built so that its k-th digit differs from the k-th digit of row k. So x is not row k, whichever k was named.
\[ x = r_{10^6} \;\Rightarrow\; x_{10^6} = (r_{10^6})_{10^6}, \;\text{but } x_{10^6} \neq (r_{10^6})_{10^6} \]
Trap
Suppose the diagonal rule were allowed to produce a nine or a zero. Then x might be a number like the one below, which has a second name.
\[ x = 0.4999\ldots = 0.5000\ldots \]
Now x could be absent under the name with nines but present in the list under its zeros name. The two decimals are the same real, so the argument leaks.
Close the loophole by never emitting a nine or a zero. Restrict the replacement digits to a safe pair such as five and six.
Force a unique decimal name
Why: A decimal made only of fives and sixes never ends in repeating nines and never ends in repeating zeros, so it has exactly one decimal representation. Digit-by-digit disagreement is then genuine number disagreement.
\[ x_k \in \{5,6\} \;\Rightarrow\; x \text{ has a unique expansion} \]
Concept
The diagonal argument shows no list of numbers from the interval can be complete. Therefore the interval is uncountable.
uncountable — A set that is neither finite nor in bijection with the naturals. Equivalently, no enumeration of it is onto: every list omits at least one element.
Since the interval sits inside the real line, and a subset of a countable set is countable, an uncountable subset forces the whole line to be uncountable too.
Definition probe
Sort into buckets
Every line below is part of the definition of countable or of uncountable — one or the other, never both. Put each where it belongs.
Ranking
Put in order
Put the moves of The interval and the whole line have the same size into the order they have to happen.
Why: These are the moves of the worked example in the order it makes them, and each one is set up by the one before it. As x approaches 0 from above the input approaches negative one half of pi, and tangent runs to negative infinity.
Worked example
We want a bijection from the open interval onto all of the reals. The tangent function, recentered, does the job.
\[ f : (0,1) \to \mathbb{R}, \qquad f(x) = \tan\!\big(\pi (x - \tfrac{1}{2})\big) \]
Check the endpoints in the limit
Why: As x approaches 0 from above the input approaches negative one half of pi, and tangent runs to negative infinity. As x approaches 1 the input approaches one half of pi and tangent runs to positive infinity.
\[ \lim_{x\to 0^+} f(x) = -\infty, \qquad \lim_{x\to 1^-} f(x) = +\infty \]
Argue injectivity by monotonicity
Why: On this interval the input runs strictly through the open interval from minus to plus one half of pi, where tangent is strictly increasing, so f is strictly increasing and therefore one to one.
Argue surjectivity by the intermediate value theorem
Why: f is continuous and takes values approaching both infinities, so it attains every real value in between. Hence f is onto.
Verify the correspondence at the center
Why: The midpoint should map to zero, the center of the line. Indeed f of one half is the tangent of zero, which is zero. A strictly increasing continuous surjection is a bijection, so the interval and the line are equinumerous, and the line is uncountable.
\[ f(\tfrac{1}{2}) = \tan(0) = 0 \]
Picture it
Animation
Shows: Each line of the worked example "The interval and the whole line have the same size", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: As x approaches 0 from above the input approaches negative one half of pi, and tangent runs to negative infinity. As x approaches 1 the input approaches one half of pi and tangent runs to positive infinity.
Step zero
Discussion prompt
Open and closed intervals have the same size — before any calculation: what is the plan? Name the moves in order, in plain English, without doing the arithmetic.
Hint: It starts with: Inject the open interval into the closed one
Answer:
Worked example
Do the two boundary points matter? We show the closed and open unit intervals are equinumerous using the Schroeder-Bernstein theorem: two injections, one each way, yield a bijection.
Inject the open interval into the closed one
Why: The open interval is literally a subset of the closed interval, so the identity map is an injection one way.
\[ \iota : (0,1) \hookrightarrow [0,1], \qquad \iota(x) = x \]
Inject the closed interval into the open one
Why: Shrink and shift the closed interval so it lands strictly inside the open one, avoiding both endpoints. This map is injective because it is a nonconstant linear function.
\[ g : [0,1] \to (0,1), \qquad g(x) = \tfrac{x+1}{3} \in [\tfrac{1}{3}, \tfrac{2}{3}] \]
Invoke Schroeder-Bernstein
Why: With an injection in each direction, the theorem manufactures a bijection between the two sets, even though no obvious explicit one is in sight.
Verify both maps are injective
Why: The identity is trivially injective. The linear map g has nonzero slope one third, so distinct inputs give distinct outputs, and its image lies inside the open interval. Both hypotheses hold, so the closed and open intervals have equal cardinality.
Picture it
Animation
Shows: Each line of the worked example "Open and closed intervals have the same size", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: The identity is trivially injective. The linear map g has nonzero slope one third, so distinct inputs give distinct outputs, and its image lies inside the open interval. Both hypotheses hold, so the closed and open intervals have equal cardinality.
Check
Sort out the relationship between being infinite and being uncountable.
Check your understanding
Which statement is true?
Answer: A
Why: Infinite and uncountable are different notions. The integers are infinite yet countable (they can be listed). The interval is infinite and additionally uncountable, since the diagonal argument defeats every list.
Concept
The rationals are countable, yet the reals are not. Splitting the reals into rationals and irrationals must therefore load all the uncountability onto the irrationals.
If the irrationals were countable, the reals would be a union of two countable sets, hence countable. They are not, so the irrationals are uncountable.
\[ \mathbb{R} = \mathbb{Q} \;\cup\; (\mathbb{R} \setminus \mathbb{Q}) \]
Estimation
Predict first
An algebraic number is a root of a nonzero polynomial with integer coefficients. Everything else is transcendental. We show the transcendentals are uncountable without exhibiting a single one.
Commit before you compute: what does Almost every real is transcendental come out to? A rough magnitude and the right form is enough — the point is to have something concrete to be wrong about.
Correct: Verify with the counting rule
Why: A prediction you can defend turns the computation into a check rather than a leap of faith — and an answer that contradicts it is caught on the spot. A union of two countable sets is countable; the reals are not countable; the algebraic reals are countable.
Worked example
An algebraic number is a root of a nonzero polynomial with integer coefficients. Everything else is transcendental. We show the transcendentals are uncountable without exhibiting a single one.
Count the integer polynomials
Why: For each degree and each bound on the coefficients there are finitely many integer polynomials. The set of all integer polynomials is a countable union of finite sets, hence countable.
\[ \mathbb{Z}[x] \;\text{is countable} \]
Count the algebraic numbers
Why: Each nonzero polynomial of degree n has at most n roots. The algebraic numbers are a countable union of finite root sets, so they are countable.
Apply the union principle in reverse
Why: If the transcendentals were countable, then the reals, being the algebraic numbers together with the transcendentals, would be a union of two countable sets, hence countable. But the reals are uncountable.
Verify with the counting rule
Why: A union of two countable sets is countable; the reals are not countable; the algebraic reals are countable. The only escape is that the transcendentals are uncountable. So almost all reals are transcendental, though naming even one takes real work.
\[ \lvert \mathbb{R} \setminus \overline{\mathbb{Q}} \rvert = \lvert \mathbb{R} \rvert \]
Picture it
Animation
Shows: Each line of the worked example "Almost every real is transcendental", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: A union of two countable sets is countable; the reals are not countable; the algebraic reals are countable. The only escape is that the transcendentals are uncountable. So almost all reals are transcendental, though naming even one takes real work.
Trap
One might hope: throw away the rationals, a small countable set, and perhaps the leftover irrationals become manageable, even countable.
The intuition treats the countable rationals as a large enough chunk to change the size of the whole.
\[ \mathbb{R} \setminus \mathbb{Q} \;\text{countable} \; ? \]
Removing a countable set from an uncountable set leaves an uncountable set. The countable part is negligible against the continuum.
Run the union argument
Why: If the leftover were countable, the reals would be the union of two countable sets and thus countable, contradicting the diagonal argument. So the irrationals stay uncountable.
Break the constraint
Discussion prompt
The rule this trap just fixed:
If the leftover were countable, the reals would be the union of two countable sets and thus countable, contradicting the diagonal argument. So the irrationals stay uncountable.
Now break it on purpose. Build a case that violates it and follow the consequences until something visibly fails. Where does the failure first show up — and would you have noticed it if you had not been looking?
Hint: The dangerous rules are the ones whose violation still produces an answer. If yours fails loudly, try to find one that fails quietly.
Concept
The diagonal argument feels tied to decimal digits, but the real engine is more general. It compares a set with its collection of subsets.
Cantor's theorem says the same phenomenon happens for every set, finite or infinite: a set can never be as large as the family of all its subsets.
\[ \text{for every set } X: \quad \lvert X \rvert < \lvert \mathcal{P}(X) \rvert \]
Concept
A subset of X is the same data as a rule that answers, for each element, in or out. That rule is its characteristic function.
\[ \chi_S : X \to \{0,1\}, \qquad \chi_S(x) = 1 \iff x \in S \]
So the power set is exactly the set of all such in-or-out vectors indexed by X. This is why subsets behave like infinite binary strings, and why a diagonal can flip them.
Intuition
Imagine trying to pair up each element of X with a subset of X, hoping to hit every subset. Element x guards one subset, its assigned partner.
Now define a rebel subset: it contains exactly those elements that are missing from their own assigned subset. This rebel disagrees with every element's subset about that very element.
So the rebel subset is assigned to no element. This is the diagonal again, dressed as membership rather than digits.
Step zero
Discussion prompt
Cantor's theorem: no set surjects onto its power set — before any calculation: what is the plan? Name the moves in order, in plain English, without doing the arithmetic.
Hint: It starts with: Build the diagonal subset
Answer:
Worked example
Let X be any set and suppose f assigns to each element a subset of X. We show f cannot be onto the power set.
\[ f : X \to \mathcal{P}(X) \]
Build the diagonal subset
Why: Collect exactly those elements that are not members of the subset f assigns to them. This is the membership-flip of the diagonal.
\[ D = \{\, x \in X : x \notin f(x) \,\} \]
Suppose D is hit, say D equals f of a
Why: If f were onto, then D, being a subset of X, equals f(a) for some element a. Ask the decisive question: is a in D?
\[ \text{assume } D = f(a) \text{ for some } a \in X \]
Derive the contradiction both ways
Why: If a is in D, then by the definition of D it is not in f(a); but f(a) is D, so a is not in D. If a is not in D, then it satisfies D's membership rule, so it is in D. Either way a paradox.
\[ a \in D \iff a \notin f(a) \iff a \notin D \]
Conclude with the strict inequality
Why: So no f is onto: the power set is strictly bigger. The map sending x to its singleton is an injection, giving less-than-or-equal, and no surjection upgrades it to strictly less.
\[ x \mapsto \{x\} \;\text{injects } X \text{ into } \mathcal{P}(X) \]
Verify the contradiction used only the definition of D
Why: The biconditional a is in D exactly when a is not in D is a genuine logical contradiction, derived purely from the rule defining D and the assumption D equals f(a). No hidden assumptions. Hence the theorem holds for every set X.
Picture it
Animation
Shows: Each line of the worked example "Cantor's theorem: no set surjects onto its power set", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: The biconditional a is in D exactly when a is not in D is a genuine logical contradiction, derived purely from the rule defining D and the assumption D equals f(a). No hidden assumptions. Hence the theorem holds for every set X.
Concept
The set of elements not belonging to their own assigned subset is the same self-reference that powers Russell's paradox: the set of all sets that do not contain themselves.
In naive set theory that self-reference is a genuine contradiction that sinks the theory. In Cantor's theorem it is a controlled tool: it does not blow up mathematics, it proves a set is smaller than its power set.
\[ D = \{\, x : x \notin f(x) \,\} \quad \text{versus} \quad R = \{\, x : x \notin x \,\} \]
Trap
Objection: D is a perfectly good subset of X, and we assumed f is onto the power set. So there must be some element a with f of a equal to D. No contradiction yet, just place a wherever is convenient.
The objector accepts the existence of such an a and expects to decide membership of a freely.
\[ D = f(a) \;\text{for some } a, \quad \text{choose } a \in D \text{ or } a \notin D \text{ at will} \]
That element a cannot be placed at all. Its membership in D is pinned by D's own rule, and the rule contradicts itself for this specific a.
Follow the forced biconditional
Why: Because D equals f(a), the defining rule of D says a is in D exactly when a is not in f(a), which is exactly when a is not in D. That equivalence between a statement and its negation is impossible. So no such a exists and f is not onto.
\[ a \in D \iff a \notin f(a) = D \]
Translation
\( D = f(a) \;\text{for some } a, \quad \text{choose } a \in D \text{ or } a \notin D \text{ at will} \)
Draw it
Translate both ways. First write the expression above as a sentence with no symbols in it at all. Then cover it, and write your sentence back as notation. If the two versions disagree, the disagreement is the thing to fix.
Picture it
Figure (svg): A rising staircase of three boxes labeled with increasing infinite cardinals: the naturals, then the power set of the naturals, then the power set of that, each strictly larger than the one before.
Discussion prompt
Read the picture before the words. What is this showing, and what is the one thing it is built to make obvious? Commit to an answer, then read on.
Hint: Name the parts, then say what changes between them — and if nothing changes, say what is being held still.
Answer:
Cantor's theorem never stops. Apply it to the naturals, then to that power set, then again, and the sizes strictly increase at every step.
Concept
Cantor's theorem never stops. Apply it to the naturals, then to that power set, then again, and the sizes strictly increase at every step.
\[ \lvert \mathbb{N} \rvert < \lvert \mathcal{P}(\mathbb{N}) \rvert < \lvert \mathcal{P}(\mathcal{P}(\mathbb{N})) \rvert < \cdots \]
Figure (svg): A rising staircase of three boxes labeled with increasing infinite cardinals: the naturals, then the power set of the naturals, then the power set of that, each strictly larger than the one before.
There is no largest infinity. The sizes form an unbounded hierarchy, one strictly above the last, without end.
Prediction
Predict first
How many subsets does X have, and how do the sizes compare?
Answer it in your own words, now, with nothing to choose from. The options are on the next slide — and picking the right one off a list is an easier skill than producing it.
Correct: 16 subsets, and no function from X onto its power set exists, so X is strictly smaller.
Why: A set of size four has two to the fourth, which is sixteen, subsets. Cantor's theorem holds for every set, so there is no surjection from the four-element X onto its sixteen-element power set, and four is strictly less than sixteen.
Check
Let X have exactly four elements. Cantor's theorem applies to finite sets too.
Check your understanding
How many subsets does X have, and how do the sizes compare?
Answer: A
Why: A set of size four has two to the fourth, which is sixteen, subsets. Cantor's theorem holds for every set, so there is no surjection from the four-element X onto its sixteen-element power set, and four is strictly less than sixteen.
Concept
A cardinal number is the size of a set, with two sets sharing a cardinal exactly when a bijection links them. The smallest infinite cardinal is written aleph-null.
\[ \lvert \mathbb{N} \rvert = \aleph_0 \]
aleph-null — The cardinality of the natural numbers, the smallest infinite cardinal. Every countably infinite set has this cardinality.
Concept
Subsets of the naturals are the same as in-or-out choices, one binary choice per natural number. Counting all such choice-vectors gives an exponential.
\[ \lvert \mathcal{P}(\mathbb{N}) \rvert = \lvert \{0,1\}^{\mathbb{N}} \rvert = 2^{\aleph_0} \]
This exponent notation is literal cardinal arithmetic: two options, raised to the number of independent slots.
Intuition
Picture an infinite row of switches, one per natural number, each either on or off. A setting of all the switches is a subset: the on switches are the members.
The set of all such infinite settings is the same object as an infinite binary string, and the same object as a real number written in binary. That coincidence is the next theorem.
Hypothesis
Predict first
The power set of the naturals is the same size as the reals is about to be worked. State your hypothesis first: which rule or definition decides this one, and what is the first move it forces? Then watch whether the example agrees with you.
Correct: Inject binary strings into the reals
Why: Send a binary sequence to the base-three number whose k-th digit is 0 or 2 according to the bit. Using only digits 0 and 2 avoids all carrying and repeating-tail clashes, so distinct sequences give distinct reals.
A hypothesis you wrote down is falsifiable; a vague sense of how it will go is not. If the example opens somewhere else, that gap is the thing worth chasing.
Worked example
We prove the continuum equals two to the aleph-null by two injections and Schroeder-Bernstein.
\[ \lvert \mathcal{P}(\mathbb{N}) \rvert = \lvert \{0,1\}^{\mathbb{N}} \rvert \quad\text{and}\quad \lvert \{0,1\}^{\mathbb{N}} \rvert = \lvert \mathbb{R} \rvert \]
Inject binary strings into the reals
Why: Send a binary sequence to the base-three number whose k-th digit is 0 or 2 according to the bit. Using only digits 0 and 2 avoids all carrying and repeating-tail clashes, so distinct sequences give distinct reals.
\[ b \mapsto \sum_{k=1}^{\infty} \frac{2 b_k}{3^{k}} \in [0,1] \]
Inject the reals into binary strings
Why: Each real in the unit interval has a binary expansion; fix the convention that forbids a tail of ones, making the expansion unique. Reading off the bits gives an injection into binary sequences.
Apply Schroeder-Bernstein
Why: An injection each way yields a bijection, so the binary strings and the reals have the same cardinality, and the power set of the naturals joins them.
Verify both maps are injective
Why: The base-three map is injective because two different bit sequences differ at some position k, forcing different base-three digits with no possible carry from digits 0 and 2. The binary-expansion map is injective once the repeating-ones tail is banned. Both hold, so the continuum equals two to the aleph-null.
\[ \lvert \mathbb{R} \rvert = 2^{\aleph_0} \]
Picture it
Animation
Shows: Each line of the worked example "The power set of the naturals is the same size as the reals", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: The base-three map is injective because two different bit sequences differ at some position k, forcing different base-three digits with no possible carry from digits 0 and 2. The binary-expansion map is injective once the repeating-ones tail is banned. Both hold, so the continuum equals two to the aleph-null.
Concept
This common size of the reals, the interval, the irrationals, and the power set of the naturals gets its own name: the cardinality of the continuum.
\[ \mathfrak{c} = \lvert \mathbb{R} \rvert = 2^{\aleph_0} \]
So there are at least two clearly different infinities in everyday mathematics: countable, and the continuum strictly above it.
Concept
Cardinal sum is the size of a disjoint union; cardinal product is the size of a Cartesian product; cardinal exponent is the size of a function space.
\[ \lvert A \rvert + \lvert B \rvert = \lvert A \sqcup B \rvert, \quad \lvert A \rvert \cdot \lvert B \rvert = \lvert A \times B \rvert, \quad \lvert B \rvert^{\lvert A \rvert} = \lvert B^{A} \rvert \]
For finite sets these recover ordinary arithmetic. For infinite cardinals they behave very differently, as we will see.
Concept
With an infinite cardinal in play, adding or multiplying by something no larger changes nothing. The larger cardinal simply swallows the smaller.
\[ \aleph_0 + \aleph_0 = \aleph_0, \qquad \aleph_0 \cdot \aleph_0 = \aleph_0 \]
These are the countable-union and pairing bijections from the previous deck, restated as arithmetic. The same absorption happens at the continuum.
Ranking
Put in order
Put the moves of The continuum absorbs the countable and itself into the order they have to happen.
Why: These are the moves of the worked example in the order it makes them, and each one is set up by the one before it. Two disjoint open intervals each have size the continuum, and their union is an interval, which again has size the continuum.
Worked example
We show the continuum plus the countable is still the continuum, and the continuum plus itself is still the continuum.
\[ \aleph_0 + \mathfrak{c} = \mathfrak{c}, \qquad \mathfrak{c} + \mathfrak{c} = \mathfrak{c} \]
Split the line into two intervals
Why: Two disjoint open intervals each have size the continuum, and their union is an interval, which again has size the continuum. So the continuum plus the continuum is the continuum.
\[ \lvert (0,1) \sqcup (1,2) \rvert = \lvert (0,2) \rvert = \mathfrak{c} \]
Squeeze the mixed sum
Why: The continuum is at most the countable plus the continuum, which is at most the continuum plus the continuum, which equals the continuum. Trapped between the continuum and itself.
\[ \mathfrak{c} \le \aleph_0 + \mathfrak{c} \le \mathfrak{c} + \mathfrak{c} = \mathfrak{c} \]
Verify by Schroeder-Bernstein
Why: The chain of less-than-or-equal relations with equal endpoints forces equality throughout, by Schroeder-Bernstein. So adjoining a countable set to the reals does not change the cardinality.
\[ \aleph_0 + \mathfrak{c} = \mathfrak{c} \]
Picture it
Animation
Shows: Each line of the worked example "The continuum absorbs the countable and itself", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: The chain of less-than-or-equal relations with equal endpoints forces equality throughout, by Schroeder-Bernstein. So adjoining a countable set to the reals does not change the cardinality.
Trap
Since the countable plus the countable equals the countable, subtract the countable from both sides and conclude the countable equals zero.
\[ \aleph_0 + \aleph_0 = \aleph_0 \;\Rightarrow\; \aleph_0 = 0 \;? \]
Cardinal addition is not cancellative, and cardinal subtraction is not even defined. Absorption destroys the information you would need to cancel.
Exhibit the failure of cancellation
Why: Adding one to the countable gives the countable, and adding zero to the countable gives the countable, so the two sums are equal while the summands one and zero are not. Cancellation would falsely conclude one equals zero.
\[ \aleph_0 + 1 = \aleph_0 = \aleph_0 + 0, \quad \text{yet } 1 \neq 0 \]
Step zero
Discussion prompt
The plane has exactly as many points as the line — before any calculation: what is the plan? Name the moves in order, in plain English, without doing the arithmetic.
Hint: It starts with: Interleave the two coordinates digit by digit
Answer:
Worked example
We build an injection from the unit square into the unit interval by interleaving decimal digits, then conclude the plane and the line are equinumerous.
Interleave the two coordinates digit by digit
Why: Given a point with coordinates having digit strings a and b, weave them into one number by alternating their digits. Different points differ in some coordinate digit, hence in some woven digit.
\[ (0.a_1 a_2 \ldots,\; 0.b_1 b_2 \ldots) \;\mapsto\; 0.a_1 b_1 a_2 b_2 \ldots \]
Get an injection the other way for free
Why: The line injects into the plane by sitting on the horizontal axis, sending a real to the point with that first coordinate and zero second coordinate.
\[ x \mapsto (x, 0) \]
Apply Schroeder-Bernstein
Why: Injections both ways, using a fixed no-nines-tail convention to keep the interleaving map injective, give a bijection between the square and the interval, hence between the plane and the line.
Verify by de-interleaving
Why: Reading off the odd-position digits recovers a, and the even-position digits recover b, so the woven number determines the original point uniquely; the map is injective. Therefore the plane and the line have the same cardinality, the continuum.
\[ \lvert \mathbb{R}^2 \rvert = \lvert \mathbb{R} \rvert = \mathfrak{c} \]
Picture it
Animation
Shows: Each line of the worked example "The plane has exactly as many points as the line", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: Reading off the odd-position digits recovers a, and the even-position digits recover b, so the woven number determines the original point uniquely; the map is injective. Therefore the plane and the line have the same cardinality, the continuum.
Fill the middle
Fill in the blanks
From Trap: the plane must be bigger than the line — finish the line. Write what belongs on the right of the equals sign before you look.
\lvert \mathbb\mathfrak{c} \quad \text{for every } n \ge 1^n \rvert = ___
Why: Producing the right-hand side unprompted is the difference between recognising this line and being able to use it. Dimension is a linear-algebra and topology invariant about coordinates and continuity.
Trap
The plane is two-dimensional and the line is one-dimensional, so the plane obviously has more points. A whole extra degree of freedom must add cardinality.
\[ \lvert \mathbb{R}^2 \rvert > \lvert \mathbb{R} \rvert \;? \]
Dimension is not cardinality. The digit-interleaving bijection matches the plane to the line point for point, so they have exactly the same size.
Separate the two notions
Why: Dimension is a linear-algebra and topology invariant about coordinates and continuity. Cardinality only counts points and ignores structure. A bijection can scramble dimension while preserving count, which is exactly what interleaving does.
\[ \lvert \mathbb{R}^n \rvert = \mathfrak{c} \quad \text{for every } n \ge 1 \]
Check
Add a countable cardinal to the continuum.
Check your understanding
What is aleph-null plus the continuum?
Answer: A
Why: Adjoining a countable set to a set of size continuum does not change the cardinality: the continuum is squeezed between itself and continuum-plus-continuum, both equal to the continuum, so the sum is the continuum.
Concept
We have two infinities in hand: the countable, and the continuum strictly above it. A natural question: is there a cardinality strictly between them?
\[ \aleph_0 \;<\; ? \;<\; \mathfrak{c} \]
In words: is there a set of reals too big to be listed, yet too small to match all the reals? This innocent-sounding question turns out to be the deepest in the deck.
Concept
The infinite cardinals are themselves well ordered, so there is a definite smallest uncountable cardinal, the immediate successor of the countable one.
\[ \aleph_1 = \text{the least cardinal strictly greater than } \aleph_0 \]
By definition nothing sits strictly between the countable and aleph-one. The open question is whether aleph-one is the continuum, or lands strictly below it.
Concept
Cantor conjectured the simplest possible answer: there is no cardinality strictly between the countable and the continuum. Equivalently, the continuum is the very next cardinal.
\[ \text{CH}: \quad \mathfrak{c} = 2^{\aleph_0} = \aleph_1 \]
Continuum Hypothesis — The statement that every infinite set of real numbers is either countable or of the same cardinality as all the reals. There is no intermediate size.
Matching
Match the pairs
Match each term to the definition this lesson gave it — not the one you would guess from the word.
Why: These are the working definitions of uncountable, aleph-null, Continuum Hypothesis as Cardinality II: Uncountability & the Continuum uses them. Pairing them correctly is the test of whether you could state each one with the slide switched off.
Concept
The same guess can be made at every level of the tower: each power set is the immediate next cardinal after the set it came from.
\[ \text{GCH}: \quad 2^{\aleph_\alpha} = \aleph_{\alpha+1} \quad \text{for every ordinal } \alpha \]
For this deck we only need the first instance, ordinary CH, but the generalized form shows it is a structural guess about the whole hierarchy.
Intuition
Some statements can be neither proved nor refuted from a set of axioms. The axioms simply do not decide them, the way the group axioms do not decide whether two elements commute.
For such a statement there are two equally legitimate mathematical worlds: one where it holds, and one where it fails, both fully consistent with the axioms.
This is not ignorance to be cured by cleverness. It is a genuine fork the axioms leave open.
Concept
The standard axioms of set theory are called ZFC. Two landmark results together show they cannot settle the Continuum Hypothesis.
Goedel built a model of ZFC in which CH is true, so ZFC cannot disprove it. Cohen, inventing the method of forcing, built a model in which CH is false, so ZFC cannot prove it either.
\[ \text{ZFC} \nvdash \text{CH} \qquad \text{and} \qquad \text{ZFC} \nvdash \neg\text{CH} \]
So CH is independent of ZFC: undecided by the axioms, not merely unproven by us.
Trap
Surely CH is like any hard conjecture. Some future genius will finally prove it, or find a counterexample, from the usual axioms of mathematics.
The view treats independence as a temporary state of ignorance, awaiting a clever enough argument.
No proof or disproof from ZFC can exist, and that itself is a theorem. Independence is a permanent, proven fact, not a knowledge gap.
Distinguish unproven from independent
Why: Goedel and Cohen exhibited models of ZFC on both sides, so a ZFC proof of either CH or its negation would contradict the existence of the opposite model. Settling CH requires new axioms beyond ZFC, not just harder work within it.
Two truths and a lie
Sort into buckets
Some of these hold up and some are the exact mistakes this lesson is built to prevent. Sort them.
Commit first
Predict first
What does it mean that the Continuum Hypothesis is independent of ZFC?
Commit to an answer, then rate it — certain, fairly sure, or guessing — and write the rating down before you turn the page.
Correct: Neither CH nor its negation can be proved from ZFC; there are models of ZFC of each kind.
Why: Independence means ZFC settles neither way: Goedel gave a model where CH holds, so it cannot be refuted, and Cohen gave a model where CH fails, so it cannot be proved. Both statements are consistent with ZFC.
The rating matters as much as the answer: confident-and-wrong is the combination that survives revision, because nothing about it feels like it needs revisiting.
Check
Pin down the precise meaning of the claim that CH is independent of ZFC.
Check your understanding
What does it mean that the Continuum Hypothesis is independent of ZFC?
Answer: A
Why: Independence means ZFC settles neither way: Goedel gave a model where CH holds, so it cannot be refuted, and Cohen gave a model where CH fails, so it cannot be proved. Both statements are consistent with ZFC.
Concept
A program in any fixed language is a finite string over a finite alphabet of symbols. Finite strings can be listed by length and then alphabetically.
\[ \lvert \{ \text{finite strings over a finite alphabet} \} \rvert = \aleph_0 \]
So there are only countably many programs in existence, ever. This modest fact collides violently with the uncountability of the reals.
Intuition
In principle, list all one-symbol programs, then all two-symbol programs, then three, and so on. Every program appears at some finite stage.
That is an enumeration of all programs, so the set of programs is countable, exactly like the naturals. There is no infinite binary explosion here, just a long list.
Estimation
Predict first
Call a real computable if some program prints its digits, one after another, on demand. We show almost all reals fail this.
Commit before you compute: what does Most real numbers are uncomputable come out to? A rough magnitude and the right form is enough — the point is to have something concrete to be wrong about.
Correct: Verify by the counting rule
Why: A prediction you can defend turns the computation into a check rather than a leap of faith — and an answer that contradicts it is caught on the spot. Countable computable reals inside uncountably many reals force uncountably many uncomputable reals, by the same union argument used for the irrationals.
Worked example
Call a real computable if some program prints its digits, one after another, on demand. We show almost all reals fail this.
Bound the computable reals by the programs
Why: Each computable real is produced by at least one program, so assigning each such real one program that computes it injects the computable reals into the countable set of programs.
\[ \{\text{computable reals}\} \hookrightarrow \{\text{programs}\} \]
Conclude the computable reals are countable
Why: A set that injects into a countable set is countable. So there are only countably many computable reals.
Subtract from the uncountable whole
Why: The reals are uncountable, and the computable ones are only countable. Removing a countable set from an uncountable one leaves an uncountable remainder.
\[ \lvert \mathbb{R} \rvert = \mathfrak{c}, \quad \lvert \{\text{computable reals}\} \rvert = \aleph_0 \]
Verify by the counting rule
Why: Countable computable reals inside uncountably many reals force uncountably many uncomputable reals, by the same union argument used for the irrationals. So almost every real can never be computed by any program.
\[ \lvert \{\text{uncomputable reals}\} \rvert = \mathfrak{c} \]
Picture it
Animation
Shows: Each line of the worked example "Most real numbers are uncomputable", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: Countable computable reals inside uncountably many reals force uncountably many uncomputable reals, by the same union argument used for the irrationals. So almost every real can never be computed by any program.
Concept
The same count applies to functions from the naturals to the naturals. There are only countably many programs, but the function space is uncountable.
\[ \lvert \mathbb{N}^{\mathbb{N}} \rvert = \mathfrak{c} > \aleph_0 = \lvert \{\text{programs}\} \rvert \]
So most functions the naturals could take to the naturals are computed by no algorithm at all. Computability is the rare exception, not the rule.
Concept
Uncomputability by counting is abstract; the halting problem makes it concrete and specific. Can one program decide whether another program eventually stops?
Suppose a universal halting decider existed: a program that, given the code of any program and an input, always halts and correctly answers whether that program halts on that input.
\[ H(p, x) = \begin{cases} \text{halts} & \text{if program } p \text{ halts on input } x \\ \text{loops} & \text{otherwise} \end{cases} \]
Step zero
Discussion prompt
The halting problem is the diagonal argument again — before any calculation: what is the plan? Name the moves in order, in plain English, without doing the arithmetic.
Hint: It starts with: Construct the contrarian program D
Answer:
Worked example
Assume the decider H exists. We build a single program that H cannot answer correctly about itself, exactly the diagonal move.
Construct the contrarian program D
Why: D takes a program p, asks H whether p halts on its own code, and then does the opposite: it loops when H predicts halting, and halts when H predicts looping.
def D(p):
if H(p, p) == "halts":
while True:
pass # loop forever
else:
return # halt nowFeed D its own code
Why: Now run D on the code of D itself. This is the diagonal element: the program applied to its own index.
\[ \text{consider } D(D) \]
Chase the contradiction both ways
Why: If D of D halts, then H predicted halting, so by D's own code it loops forever, not halts. If D of D loops, then H predicted looping, so D halts. Either branch contradicts itself.
\[ D(D) \text{ halts} \iff H(D,D) = \text{halts} \iff D(D) \text{ loops} \]
Verify no branch survives
Why: The assumption that H always answers correctly leads to D of D halting exactly when it does not halt, a genuine contradiction. Therefore no such H exists: the halting problem is undecidable. Same recipe as the diagonal on the reals.
Picture it
Animation
Shows: Each line of the worked example "The halting problem is the diagonal argument again", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: The assumption that H always answers correctly leads to D of D halting exactly when it does not halt, a genuine contradiction. Therefore no such H exists: the halting problem is undecidable. Same recipe as the diagonal on the reals.
Concept
The uncountability of the reals, Cantor's theorem, and the undecidability of halting are the same proof wearing three costumes.
Each assumes a complete listing: an enumeration of reals, a surjection onto a power set, or a total halting decider. Each then builds a diagonal object that differs from every entry, or answers opposite to the decider on its own index.
\[ \text{list of reals} \;\leftrightarrow\; f : X \to \mathcal{P}(X) \;\leftrightarrow\; H(p,x) \]
Diagonalization is a single idea that reaches from set theory to the limits of computation.
Concept
We have leaned on one theorem repeatedly to conclude two sets have equal size. It deserves a clean statement.
Schroeder-Bernstein theorem — If there is an injection from A into B and also an injection from B into A, then there is a bijection between A and B. Two one-way size comparisons that agree force equality of cardinality.
\[ \lvert A \rvert \le \lvert B \rvert \;\text{and}\; \lvert B \rvert \le \lvert A \rvert \;\Longrightarrow\; \lvert A \rvert = \lvert B \rvert \]
It lets us prove two sets are equinumerous without ever exhibiting a single explicit bijection, which is exactly why the interval-versus-line and plane-versus-line arguments went through so smoothly.
Intuition
Picture an injection from A into B and one from B into A. Following them alternately traces chains of elements, back and forth between the two sets.
Schroeder-Bernstein untangles those chains, matching each element of A with exactly one partner in B. The two loose one-way maps zip together into a perfect pairing.
Explain it
Discussion prompt
Explain Two injections zip into a bijection to a student a year behind you. No notation, no jargon they have not met — and it still has to be true.
Hint: If your explanation needs a symbol they have never seen, you are describing the notation rather than the idea.
Answer:
Picture an injection from A into B and one from B into A. Following them alternately traces chains of elements, back and forth between the two sets.
Ranking
Put in order
Put the moves of Infinite binary sequences are uncountable, directly into the order they have to happen.
Why: These are the moves of the worked example in the order it makes them, and each one is set up by the one before it. Suppose all infinite binary sequences could be enumerated.
Worked example
The diagonal argument is even cleaner on bit sequences than on decimals, with no nines-and-zeros loophole to dodge. This is Cantor's theorem for the naturals in disguise.
Assume a complete list of sequences
Why: Suppose all infinite binary sequences could be enumerated. Then each sequence is some row, and the k-th bit of row n is well defined.
\[ s_1, s_2, s_3, \ldots \quad \text{with bits } (s_n)_k \in \{0,1\} \]
Flip the diagonal bit
Why: Define a new sequence t whose n-th bit is the opposite of the n-th bit of row n. Flipping is the binary version of changing a digit.
\[ t_n = 1 - (s_n)_n \]
Locate the disagreement
Why: By construction t differs from row n in position n, for every n. So t is none of the listed sequences, yet it is a legitimate infinite binary sequence.
Verify the flip guarantees difference at position n
Why: Since t of n is one minus the n-th bit of row n, it is never equal to that bit, so t and row n differ at position n and are different sequences. The list omits t, so no list is complete and the sequences are uncountable.
\[ t_n \neq (s_n)_n \;\Rightarrow\; t \neq s_n \quad \text{for every } n \]
Picture it
Animation
Shows: Each line of the worked example "Infinite binary sequences are uncountable, directly", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: Since t of n is one minus the n-th bit of row n, it is never equal to that bit, so t and row n differ at position n and are different sequences. The list omits t, so no list is complete and the sequences are uncountable.
Prediction
Predict first
What are the first three bits of the flipped diagonal sequence t?
Answer it in your own words, now, with nothing to choose from. The options are on the next slide — and picking the right one off a list is an easier skill than producing it.
Correct: 0 1 0
Why: The diagonal bits are 1, 0, 1 taken from position n of row n. Flipping each gives 0, 1, 0, which differs from row one at bit one, row two at bit two, and row three at bit three.
Check
Apply the flipping rule to the first three rows of an assumed list of binary sequences.
\[ \begin{aligned} s_1 &= \mathbf{1}\,0\,1\ldots \\ s_2 &= 0\,\mathbf{0}\,1\ldots \\ s_3 &= 1\,1\,\mathbf{1}\ldots \end{aligned} \]
Check your understanding
What are the first three bits of the flipped diagonal sequence t?
Answer: A
Why: The diagonal bits are 1, 0, 1 taken from position n of row n. Flipping each gives 0, 1, 0, which differs from row one at bit one, row two at bit two, and row three at bit three.
Concept
With countability behind us, this deck completes the size theory of infinite sets: there is a strict, unbounded hierarchy, and the continuum is genuinely larger than the countable.
The tools built here, bijections, injections both ways, and Schroeder-Bernstein, are the same structure-preserving maps and comparisons that the algebra and analysis units will lean on next.
Analogy
Discussion prompt
Explain Where this sits in the course by analogy to something with no Foundations of Higher Mathematics in it at all — a queue, a recipe, a map, a bank balance, whatever fits. Then say where your analogy breaks.
Hint: An analogy that never breaks is not an analogy, it is the same idea wearing a hat. Find the seam — that is the part that is actually new.
Answer:
With countability behind us, this deck completes the size theory of infinite sets: there is a strict, unbounded hierarchy, and the continuum is genuinely larger than the countable.
Elimination
Eliminate the wrong options
Why must some real numbers be uncomputable?
3 of these 4 are wrong. Strike them one at a time, and say what rules each one out before you strike the next. The survivor is the answer.
Survives elimination: A
Why: Programs are finite strings, hence countable, and each computable real needs a program, so the computable reals are countable. The reals are uncountable, so an uncountable remainder is computed by no program at all.
Check
Reason from the two cardinalities you now know.
Check your understanding
Why must some real numbers be uncomputable?
Answer: A
Why: Programs are finite strings, hence countable, and each computable real needs a program, so the computable reals are countable. The reals are uncountable, so an uncountable remainder is computed by no program at all.
Concept
Four results anchor this deck, each proved by the same diagonal idea or its consequences.
\[ \lvert \mathbb{R} \rvert > \aleph_0, \qquad \lvert X \rvert < \lvert \mathcal{P}(X) \rvert, \qquad \mathfrak{c} = 2^{\aleph_0}, \qquad \text{ZFC} \nvdash \text{CH} \]
And one slogan: diagonalize a supposed complete list, and it defeats itself.
Counterexample
Discussion prompt
Four results anchor this deck, each proved by the same diagonal idea or its consequences.
That is stated as though it always holds. Do one of two things: produce a case where it fails, or say precisely what rules such a case out. "It just does" is not on the menu.
Hint: Hunt at the extremes first — zero, one, negative, empty, equal. If every extreme survives, the reason they survive is the proof.
Connect it up
Draw it
One page, no notation unless you need it: draw how these connect — The diagonalization recipe · Countable, recalled · A list is a function from the naturals · The target: the open interval · Reals as infinite decimals. Put an arrow wherever one of them is what makes another possible, and label the arrow with why.
Recap
You can run Cantor's diagonal argument to prove the reals are uncountable, dodging the nines-and-zeros loophole, and extend it to show the irrationals and the transcendentals are uncountable too.
You can prove Cantor's theorem that every set is strictly smaller than its power set, recognize its Russell-paradox core, and read off the unbounded tower of infinities.
You can do cardinal arithmetic with absorption, identify the continuum as two to the aleph-null, state the Continuum Hypothesis, and say precisely why its independence from ZFC is a theorem rather than an open problem.
And you can see the punchline for computation: countably many programs against uncountably many reals means most reals and most functions are uncomputable, with the halting problem falling to the very same diagonal argument.
| Result | How it is proved |
|---|---|
| The reals are uncountable | Diagonal on an assumed list |
| A set is smaller than its power set | Diagonal set D of non-self-members |
| The continuum equals two to the aleph-null | Two injections and Schroeder-Bernstein |
| CH is independent of ZFC | Goedel and Cohen models |
| Halting is undecidable | Diagonal on an assumed decider |
Want this taught 1-on-1? Alexander tutors Foundations of Higher Mathematics — $55/session, free consultation.