This deck presents the three faces of continuity - epsilon-delta, sequential, and the topological definition by preimages of open sets - and shows how continuity carries topological properties across, through the Intermediate Value and Extreme Value theorems. It closes with a synthesis that names the arc running from logic through algebra to topology by its three recurring threads: structure-preserving maps, quotients by congruences and equivalences, and completeness and compactness. It targets the classic traps: quantifier-order errors in epsilon-delta, the pen-lifting cartoon of continuity, confusing preimage-open with image-open, and misapplying the IVT or the EVT outside their required domains.
Subject: Foundations of Higher Mathematics · 103 slides · symbolic lesson
Open the interactive version of this deck · Homework for this lesson
Objectives
This is the finale. By the end of this deck you will be able to:
1. State and use all three equivalent definitions of continuity: epsilon-delta, sequential, and preimage-of-open.
2. Run a clean epsilon-delta continuity proof and diagnose the quantifier-order mistakes that wreck it.
3. Explain how continuity carries connectedness and compactness across, giving the Intermediate Value and Extreme Value theorems.
4. Name the three threads that run through the entire course and place every unit on the logic-to-topology arc.
Warm-up
Discussion prompt
Before we open Continuity & the Structural Synthesis: without looking back, what was the main idea of Topology of R: Open, Closed & Compact, 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:
That deck builds the topology of the real line from its order. It covers open sets as wiggle room and closed sets as complements that contain their limit points, the line and the empty set as the clopen pair, the union and intersection laws and why an infinite intersection of open sets can fail to be open, compactness via open covers together with the Heine-Borel theorem, sequential compactness, and the connectedness of intervals. It targets the misconceptions that closed means not open, that an infinite intersection of open sets is open, that boundedness alone forces compactness, and that closed means finite or bounded.
Concept
Informally, a function is continuous when small changes in the input cause only small changes in the output. That picture is good fuel, but you cannot prove theorems with it.
We need a statement precise enough to settle the hard cases: functions that wiggle infinitely fast, functions defined by cases, functions on strange domains. The naive picture cannot decide those.
continuity (informal) — The idea that outputs stay close when inputs stay close. Made precise by the epsilon-delta condition below, which quantifies exactly how close is close enough.
Counterexample
Discussion prompt
Informally, a function is continuous when small changes in the input cause only small changes in the output. That picture is good fuel, but you cannot prove theorems with it.
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 need a statement precise enough to settle the hard cases: functions that wiggle infinitely fast, functions defined by cases, functions on strange domains. The naive picture cannot decide those.
Concept
Fix a function and a point a in its domain. Continuity at a is a promise about control: you name how close the output must land, and I must produce how close the input has to be.
Here is the definition in symbols. Read it slowly, left to right:
\[ \forall \varepsilon > 0 \ \; \exists \delta > 0 \ \; \forall x \; \bigl( |x - a| < \delta \ \Rightarrow\ |f(x) - f(a)| < \varepsilon \bigr) \]
continuous at a — For every positive output tolerance epsilon there is a positive input tolerance delta so that every input within delta of a is sent within epsilon of f(a).
Analogy
Discussion prompt
Explain The epsilon-delta definition at a point 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:
Fix a function and a point a in its domain. Continuity at a is a promise about control: you name how close the output must land, and I must produce how close the input has to be.
Picture it
Figure (svg): A curve through the point a; a horizontal epsilon band brackets f(a) and a vertical delta window brackets a, with the curve staying inside the band over the window.
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:
Think of it as a two-player game. The challenger picks a tiny target band around the value f(a) and dares you to stay inside it.
Intuition
Think of it as a two-player game. The challenger picks a tiny target band around the value f(a) and dares you to stay inside it.
Figure (svg): A curve through the point a; a horizontal epsilon band brackets f(a) and a vertical delta window brackets a, with the curve staying inside the band over the window.
You respond by shrinking the input window around a until the whole curve over that window lands inside the band. If you can answer every challenge, the function is continuous at a.
The order is the entire game: the challenger moves first with epsilon, then you answer with delta. Your delta is allowed to depend on the challenge.
Explain it
Discussion prompt
Explain Continuity as a challenge-response game 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:
Think of it as a two-player game. The challenger picks a tiny target band around the value f(a) and dares you to stay inside it.
Concept
This is exactly the quantifier machinery from the predicate-logic deck. Continuity at a point is a for-all, there-exists, for-all sentence.
\[ \underbrace{\forall \varepsilon}_{\text{challenger}} \ \underbrace{\exists \delta}_{\text{you}} \ \underbrace{\forall x}_{\text{every input}} \ \bigl( |x-a| < \delta \Rightarrow |f(x)-f(a)| < \varepsilon \bigr) \]
The dependency runs left to right: delta sits inside the scope of epsilon, so delta may be chosen using epsilon. Reversing those two quantifiers changes the meaning completely, as it always does.
Concept
A function is continuous on a set when it is continuous at every point of that set. Point continuity is local; set continuity is that local property holding everywhere at once.
\[ f \text{ continuous on } S \iff \forall a \in S,\ f \text{ continuous at } a \]
Keep the distinction sharp: the delta in the point definition may vary from point to point. When one delta works for all points at once, we get the stronger notion of uniform continuity, which we reach later.
Ranking
Put in order
Put the moves of Worked example: the squaring function is continuous at 3 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. Factor the difference of squares so the quantity we can control, the input gap, appears explicitly as a factor.
Worked example
Prove that the function which squares its input is continuous at the point three.
\[ f(x) = x^2, \qquad a = 3 \]
Write the output gap as a product
Why: Factor the difference of squares so the quantity we can control, the input gap, appears explicitly as a factor.
\( |x^2 - 9| = |x-3|\,|x+3| \)
Tame the wild factor by pre-restricting delta
Why: If we agree to keep the input within 1 of 3, then x lies between 2 and 4, so the factor with the sum is bounded above by 7.
\( |x-3| < 1 \ \Rightarrow\ 2 < x < 4 \ \Rightarrow\ |x+3| < 7 \)
Force the product below epsilon
Why: Now the output gap is at most 7 times the input gap, so making the input gap smaller than epsilon over 7 finishes the job.
\( |x^2 - 9| < 7\,|x-3| \)
Choose delta as the smaller of the two demands
Why: We needed both the pre-restriction and the epsilon-over-7 bound, so take the minimum of the two so both hold at once.
\( \delta = \min\!\left(1,\ \tfrac{\varepsilon}{7}\right) \)
Verify the choice meets the definition
Why: Given epsilon, this delta forces both bounds, so the output gap is under 7 times epsilon over 7, which is epsilon. The definition is satisfied and the answer is the boxed delta rule.
\[ |x-3| < \delta \ \Rightarrow\ |x^2 - 9| < 7\cdot\tfrac{\varepsilon}{7} = \varepsilon \]
Picture it
Animation
Shows: Each line of the worked example "the squaring function is continuous at 3", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: Given epsilon, this delta forces both bounds, so the output gap is under 7 times epsilon over 7, which is epsilon. The definition is satisfied and the answer is the boxed delta rule.
Trap
Fixing one delta once and for all, before seeing the challenge, for instance always taking delta to be a tenth.
\( \exists \delta\ \forall \varepsilon\ \forall x\ \bigl(\dots\bigr) \)
This flipped statement says one input window works for every output tolerance. For the squaring function at 3 it fails: with delta fixed, pick epsilon smaller than the output spread across that window and the bound breaks.
Let epsilon move first, then build delta from it.
\( \forall \varepsilon\ \exists \delta\ \forall x\ \bigl(\dots\bigr) \)
Delta is a function of epsilon and of the point. For smaller epsilon you simply take a smaller delta. That is precisely why the minimum rule from the last slide works: shrink delta as the challenge shrinks.
Concept
There is a second, equivalent way to say continuity, phrased entirely in terms of sequences. It is often the easiest tool for proving a function is not continuous.
\[ x_n \to a \ \Longrightarrow\ f(x_n) \to f(a) \]
sequential continuity at a — Whenever a sequence of inputs converges to a, the sequence of outputs converges to f(a). The function commutes with the operation of taking limits.
Definition probe
Sort into buckets
Every line below is part of the definition of continuity (informal) or of sequential continuity at a — one or the other, never both. Put each where it belongs.
Intuition
A sequence converging to a is a way of sneaking up on the point along a chosen path. Sequential continuity says every path of approach gives outputs that home in on the same value, f(a).
To break continuity you only need one bad path: a single sequence approaching a whose outputs refuse to approach f(a). That is why sequences are the demolition tool of choice.
Concept
For functions on the real line the two definitions agree exactly. This is a theorem, not a coincidence, and the forward direction is a short epsilon-delta argument.
\[ (\text{epsilon-delta continuous at } a) \iff (\text{sequentially continuous at } a) \]
The reverse direction quietly builds a witnessing sequence by choosing one point from each shrinking window, a countable choice. Keep that dependency in mind; it echoes the choice-flavored steps from the cardinality decks.
Step zero
Discussion prompt
Worked example: the Dirichlet function is continuous nowhere — before any calculation: what is the plan? Name the moves in order, in plain English, without doing the arithmetic.
Hint: It starts with: Take any point a and split into two cases
Answer:
Worked example
Show that the function returning one on rationals and zero on irrationals is continuous at no point at all.
\[ D(x) = \begin{cases} 1 & x \in \mathbb{Q} \\ 0 & x \notin \mathbb{Q} \end{cases} \]
Take any point a and split into two cases
Why: Every real number is either rational or irrational; we attack each case with a deliberately bad approaching sequence.
If a is irrational, approach it by rationals
Why: The rationals are dense, so a sequence of rationals converges to a. Every one of their D-values equals one, so the outputs converge to one, not to zero.
\( q_n \to a,\ q_n \in \mathbb{Q} \ \Rightarrow\ D(q_n) = 1 \to 1 \ne 0 = D(a) \)
If a is rational, approach it by irrationals
Why: The irrationals are also dense, so a sequence of them converges to a with D-values all zero, converging to zero, not to one.
\( t_n \to a,\ t_n \notin \mathbb{Q} \ \Rightarrow\ D(t_n) = 0 \to 0 \ne 1 = D(a) \)
Verify the failure at every point
Why: In both cases we exhibited an input sequence converging to a whose output sequence misses D(a). Sequential continuity fails at every a, so D is continuous nowhere, the required conclusion.
Picture it
Animation
Shows: Each line of the worked example "the Dirichlet function is continuous nowhere", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: In both cases we exhibited an input sequence converging to a whose output sequence misses D(a). Sequential continuity fails at every a, so D is continuous nowhere, the required conclusion.
Trap
Believing that continuous means you can draw the whole graph without lifting your pen.
This cartoon is misleading. The wildly oscillating curve below packs infinitely many wiggles near zero; the pen test gives no honest verdict there, yet the epsilon-delta condition does.
\( g(x) = \sin\!\left(\tfrac{1}{x}\right) \)
Continuity is the epsilon-delta and sequential condition, full stop. The pen is a picture of the smooth case, not the definition.
Trust the quantifiers, not the drawing. The definition decides hard cases the pen cannot even attempt to sketch.
Concept
The deepest definition of continuity uses only open sets, so recall from the topology deck what open means on the real line.
\[ U \text{ open} \iff \forall x \in U\ \exists r > 0\ \ (x-r,\, x+r) \subseteq U \]
Open sets are exactly the ones where every point has a little breathing room entirely inside the set. Arbitrary unions of open sets are open; finite intersections of open sets are open.
Concept
Here is the definition that will outlive this course. It never mentions epsilon, delta, or distance; it speaks only of open sets.
\[ f \text{ continuous} \iff \text{for every open } U,\ f^{-1}(U) \text{ is open} \]
In words: the preimage of every open set is open. On the real line this is provably the same as the epsilon-delta definition, but it makes sense in any topological space, which is why it is the one that generalizes.
continuity (topological) — A function is continuous when the preimage of every open set in the codomain is an open set in the domain. No metric or distance is needed.
Intuition
Why the preimage and not the image? Because continuity is about not tearing the domain apart, and that is a statement about pulling open sets back, not pushing them forward.
The image of an open set under a continuous function need not be open. The squaring map sends the open interval around zero to a set that includes its endpoint zero, which is not open. Continuity survives; image-openness does not.
\( f(x) = x^2:\quad f\bigl((-1,1)\bigr) = [0,1) \)
Step zero
Discussion prompt
Worked example: doubling is continuous, via preimages — before any calculation: what is the plan? Name the moves in order, in plain English, without doing the arithmetic.
Hint: It starts with: Take an arbitrary open interval in the codomain
Answer:
Worked example
Prove the doubling map is continuous using only the topological definition.
\[ f(x) = 2x \]
Take an arbitrary open interval in the codomain
Why: Open sets on the line are unions of open intervals, so it is enough to pull back a single open interval.
\( (a, b) \subseteq \mathbb{R} \)
Compute its preimage
Why: An output lands in the interval exactly when the input lands in the halved interval; solve the two inequalities.
\( f^{-1}\bigl((a,b)\bigr) = \{x : a < 2x < b\} = \left(\tfrac{a}{2},\, \tfrac{b}{2}\right) \)
Observe the preimage is open
Why: The halved interval is itself an open interval, hence an open set.
Lift to arbitrary open sets
Why: Any open set is a union of open intervals, preimage commutes with unions, and a union of open sets is open.
\( f^{-1}\!\left(\bigcup_i I_i\right) = \bigcup_i f^{-1}(I_i) \)
Verify against the definition
Why: The preimage of every open set is open, so by the topological definition the doubling map is continuous. Check: the preimage of the interval from a to b is the interval from a-halves to b-halves, which is open.
Picture it
Animation
Shows: Each line of the worked example "doubling is continuous, via preimages", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: The preimage of every open set is open, so by the topological definition the doubling map is continuous. Check: the preimage of the interval from a to b is the interval from a-halves to b-halves, which is open.
Trap
Stating the definition as: the image of every open set is open.
That is a different, and generally false, property. The squaring map is continuous, yet it sends the open interval from minus one to one onto the set from zero up to but not including one, which is not open.
\( f\bigl((-1,1)\bigr) = [0,1)\ \text{not open} \)
The definition pulls open sets back: the preimage of every open set is open.
Preimage, not image. That same squaring map satisfies this condition, which is exactly why it is continuous.
\( f^{-1}\bigl((a,b)\bigr)\ \text{open for every } (a,b) \)
Break the constraint
Discussion prompt
The rule this trap just fixed:
The definition pulls open sets back: the preimage of every open set is open.
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
If the inner function is continuous at a point and the outer function is continuous at the image of that point, then applying one after the other is continuous. This is what makes continuous maps a well-behaved category.
\[ f \text{ cont. at } a,\ \ g \text{ cont. at } f(a) \ \Rightarrow\ g \circ f \text{ cont. at } a \]
In the topological language the proof is one line: the preimage of an open set under the composite is the preimage of a preimage, and each step keeps sets open.
Hypothesis
Predict first
Worked example: composition is continuous, by sequences 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: Take any input sequence converging to a
Why: Sequential continuity is a statement about all such sequences, so start with an arbitrary one.
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
Prove the composition of two continuous functions is continuous at a point, using the sequential definition.
\[ h = g \circ f \]
Take any input sequence converging to a
Why: Sequential continuity is a statement about all such sequences, so start with an arbitrary one.
\( x_n \to a \)
Apply continuity of the inner function
Why: Because f is continuous at a it commutes with this limit, sending the sequence to one converging to the value at a.
\( f(x_n) \to f(a) \)
Apply continuity of the outer function at that image
Why: The sequence of images converges to the image of a, and g is continuous there, so g commutes with that limit.
\( g(f(x_n)) \to g(f(a)) \)
Verify the composite meets the definition
Why: We showed every sequence to a yields a composite sequence to the composite value. That is exactly sequential continuity of the composite at a. Check: each arrow used continuity at the correct point, a for f and the image of a for g.
\( h(x_n) \to h(a) \)
Picture it
Animation
Shows: Each line of the worked example "composition is continuous, by sequences", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: We showed every sequence to a yields a composite sequence to the composite value. That is exactly sequential continuity of the composite at a. Check: each arrow used continuity at the correct point, a for f and the image of a for g.
Concept
Continuous functions are closed under the pointwise operations. Sums, differences, and products of continuous functions are continuous; quotients are continuous wherever the denominator is nonzero.
\[ f, g \text{ continuous} \ \Rightarrow\ f+g,\ f-g,\ fg \text{ continuous};\quad \tfrac{f}{g} \text{ continuous where } g \ne 0 \]
The proofs all reduce to the algebra of limits from the sequences deck. This single fact is why every polynomial, and every rational function on its domain, is continuous with no further work.
Ranking
Put in order
Put the moves of Worked example: the sum of continuous functions is continuous 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. Once more we verify sequential continuity, so begin with an arbitrary sequence approaching a.
Worked example
Prove that the sum of two continuous functions is continuous at a point, using sequences and the limit laws.
Take an input sequence converging to a
Why: Once more we verify sequential continuity, so begin with an arbitrary sequence approaching a.
\( x_n \to a \)
Use continuity of each summand
Why: Both functions are continuous at a, so each output sequence converges to its value at a.
\( f(x_n) \to f(a),\qquad g(x_n) \to g(a) \)
Add the two limits
Why: The limit of a sum is the sum of the limits, a law proved in the sequences deck.
\( f(x_n) + g(x_n) \to f(a) + g(a) \)
Verify the definition for the sum
Why: The output of the sum along any sequence to a converges to its value at a, so the sum is sequentially continuous at a. Check: the step used only the additive limit law, which requires both limits to exist, and they do.
Picture it
Animation
Shows: Each line of the worked example "the sum of continuous functions is continuous", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: The output of the sum along any sequence to a converges to its value at a, so the sum is sequentially continuous at a. Check: the step used only the additive limit law, which requires both limits to exist, and they do.
Concept
Recall from the topology deck that an interval cannot be split into two nonempty pieces that are each open and disjoint. This unbreakable quality is called connectedness.
\[ \text{interval } I = A \cup B,\ A,B \text{ open, disjoint, nonempty} \ \Rightarrow\ \text{impossible} \]
Connectedness is a purely topological property, defined without any mention of a formula. The next theorem says continuous functions cannot destroy it.
Concept
The Intermediate Value Theorem is the precise statement that a continuous function on a closed interval hits every value between any two values it attains.
\[ f \text{ continuous on } [a,b],\ \ f(a) < y < f(b) \ \Rightarrow\ \exists c \in (a,b):\ f(c) = y \]
It is an existence theorem: it promises a solution exists, without telling you where. That is exactly enough to prove that equations have roots.
Intuition
Why is the theorem true? Because a continuous function sends a connected set to a connected set. The image of the interval is again an unbreakable piece of the line.
A connected subset of the line is itself an interval, and an interval containing both endpoint values must contain everything between them. The missing value cannot be skipped without tearing the image apart, which continuity forbids.
\( f \text{ continuous},\ I \text{ connected} \ \Rightarrow\ f(I) \text{ connected} \)
Step zero
Discussion prompt
Worked example: a root of a cubic — before any calculation: what is the plan? Name the moves in order, in plain English, without doing the arithmetic.
Hint: It starts with: Confirm continuity on the closed interval
Answer:
Worked example
Show the cubic below has a real root strictly between one and two.
\[ p(x) = x^3 - x - 1 \]
Confirm continuity on the closed interval
Why: Polynomials are continuous everywhere by the arithmetic of continuous functions, so this cubic is continuous on the interval from one to two.
Evaluate at the left endpoint
Why: Compute the value at one to find the sign there.
\( p(1) = 1 - 1 - 1 = -1 < 0 \)
Evaluate at the right endpoint
Why: Compute the value at two to find the opposite sign.
\( p(2) = 8 - 2 - 1 = 5 > 0 \)
Invoke the Intermediate Value Theorem
Why: Zero lies between the two endpoint values, and the function is continuous, so some interior point maps to zero.
\( \exists c \in (1,2):\ p(c) = 0 \)
Verify the hypotheses actually held
Why: Check: the cubic is continuous on the closed interval and the endpoint values have opposite signs, negative then positive, so the theorem genuinely applies and a root exists in the open interval from one to two.
Picture it
Animation
Shows: Each line of the worked example "a root of a cubic", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: Check: the cubic is continuous on the closed interval and the endpoint values have opposite signs, negative then positive, so the theorem genuinely applies and a root exists in the open interval from one to two.
Trap
Applying the theorem to a function with a jump, or on a domain that is not an interval, and still expecting a guaranteed value in between.
The step function that is minus one below zero and plus one from zero onward leaps straight past the value zero; it never equals zero even though it takes a negative and a positive value. The continuity hypothesis is doing real work.
\( s(x) = \begin{cases} -1 & x < 0 \\ 1 & x \ge 0 \end{cases} \)
Check both hypotheses before invoking the theorem: the function must be continuous, and the domain must be a genuine interval, that is, connected.
With continuity on a true interval, connectedness is preserved and no intermediate value can be skipped. Drop either hypothesis and the conclusion collapses.
Concept
Recall the other great topological property from the previous deck: compactness, defined by the finite-subcover condition and pinned down on the line by Heine and Borel.
\[ K \subseteq \mathbb{R} \text{ compact} \iff K \text{ closed and bounded} \]
Compactness is finiteness control: any cover of the set by open sets can be trimmed to finitely many. Completeness of the reals is exactly what makes this equivalence true.
Concept
The Extreme Value Theorem says a continuous function on a compact interval actually attains a highest and a lowest value; the maximum and minimum are reached, not merely approached.
\[ f \text{ continuous on } [a,b] \ \Rightarrow\ \exists\, x_{\max}, x_{\min} \in [a,b]:\ f(x_{\min}) \le f(x) \le f(x_{\max}) \]
This is the theorem behind optimization existence: on a closed bounded interval, a best value is guaranteed to exist somewhere.
Intuition
The mechanism mirrors the last theorem. A continuous function sends a compact set to a compact set, so the image is again closed and bounded.
A closed bounded set of reals contains its supremum and infimum. Those attained extremes are the maximum and minimum values, so the function must reach them.
\( f \text{ continuous},\ K \text{ compact} \ \Rightarrow\ f(K) \text{ compact} \)
Step zero
Discussion prompt
Worked example: an open interval can lose its maximum — before any calculation: what is the plan? Name the moves in order, in plain English, without doing the arithmetic.
Hint: It starts with: Confirm the function is continuous and bounded
Answer:
Worked example
Show that on the open interval from zero to one, the identity function is continuous and bounded yet has no maximum value.
\[ f(x) = x,\quad x \in (0,1) \]
Confirm the function is continuous and bounded
Why: The identity is continuous, and on this interval its values stay strictly between zero and one, so it is bounded above by one.
\( 0 < f(x) < 1 \)
Show the value one is never attained
Why: For any point in the interval, that point is strictly less than one, so no input produces the output one.
\( c \in (0,1) \Rightarrow f(c) = c < 1 \)
Beat every candidate maximizer
Why: Given any proposed maximizer, the midpoint between it and one still lies in the interval and has a strictly larger value, so no maximizer exists.
\( \tfrac{c+1}{2} \in (0,1),\qquad f\!\left(\tfrac{c+1}{2}\right) = \tfrac{c+1}{2} > c \)
Verify why the theorem does not apply
Why: Check: the supremum is one but it is never attained. There is no conflict with the Extreme Value Theorem because the domain, the open interval, is bounded but not closed, hence not compact.
Picture it
Animation
Shows: Each line of the worked example "an open interval can lose its maximum", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: Check: the supremum is one but it is never attained. There is no conflict with the Extreme Value Theorem because the domain, the open interval, is bounded but not closed, hence not compact.
Trap
Assuming any continuous function on an interval attains a maximum.
On the open interval from zero to one the identity has supremum one but no maximum; the reciprocal is even worse, continuous there but unbounded, with no supremum at all.
\( \tfrac{1}{x} \text{ on } (0,1):\ \text{continuous, unbounded} \)
The Extreme Value Theorem requires a compact domain: a closed and bounded interval. Only then is attainment guaranteed.
Close the interval to make it compact and the maximum reappears. On the closed interval from zero to one the identity attains its maximum, the value one, at the right endpoint.
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.
Concept
There is a stronger cousin of continuity in which a single input tolerance works everywhere at once, with no dependence on the point.
\[ \forall \varepsilon > 0\ \exists \delta > 0\ \forall x, y\ \bigl( |x-y| < \delta \Rightarrow |f(x)-f(y)| < \varepsilon \bigr) \]
uniform continuity — For every output tolerance there is one input tolerance delta that works simultaneously at every pair of nearby points, chosen before any point is named.
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 continuity (informal), continuous at a, sequential continuity at a, continuity (topological), uniform continuity as Continuity & the Structural Synthesis uses them. Pairing them correctly is the test of whether you could state each one with the slide switched off.
Intuition
The difference is quantifier placement, the theme of this whole course. Ordinary continuity lets delta depend on the point; uniform continuity demands one delta good across the entire domain at once.
\[ \text{continuous: } \forall x\, \forall \varepsilon\, \exists \delta \qquad \text{uniform: } \forall \varepsilon\, \exists \delta\, \forall x \]
Pulling the point quantifier inside the delta is the entire content of the strengthening. Where the function steepens without bound, no single delta can keep up, and uniformity fails.
Ranking
Put in order
Put the moves of Worked example: squaring is not uniformly continuous on the line 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. Aim to defeat uniformity for a single fixed output tolerance, say two; we will beat any proposed delta.
Worked example
Prove the squaring function fails to be uniformly continuous on the whole real line, even though it is continuous at every point.
\[ f(x) = x^2 \text{ on } \mathbb{R} \]
Fix a challenge no delta can meet
Why: Aim to defeat uniformity for a single fixed output tolerance, say two; we will beat any proposed delta.
\( \varepsilon = 2 \)
Build two points a shrinking distance apart
Why: Take a large index and a nearby partner; their gap goes to zero as the index grows, so eventually it drops under any delta.
\( x_n = n,\quad y_n = n + \tfrac{1}{n},\quad |x_n - y_n| = \tfrac{1}{n} \to 0 \)
Compute the output gap
Why: Expand the squares; the cross term does not shrink, it stays above two no matter how large the index is.
\( |y_n^2 - x_n^2| = 2 + \tfrac{1}{n^2} > 2 \)
Verify uniformity fails
Why: Check: for the output tolerance two and any candidate delta, choose the index so that one over it is below delta; then the inputs are within delta but the outputs differ by more than two. No single delta works, so the squaring map is not uniformly continuous on the line.
Picture it
Animation
Shows: Each line of the worked example "squaring is not uniformly continuous on the line", appearing one at a time.
The same working the example does, in the order a tutor would write it.
Takeaway: Check: for the output tolerance two and any candidate delta, choose the index so that one over it is below delta; then the inputs are within delta but the outputs differ by more than two. No single delta works, so the squaring map is not uniformly continuous on the line.
Concept
The pathology in the last example needed an unbounded domain. On a compact domain it disappears entirely: continuity automatically upgrades to uniform continuity.
\[ f \text{ continuous on compact } K \ \Rightarrow\ f \text{ uniformly continuous on } K \]
This is the Heine-Cantor theorem, and its engine is again compactness: finitely many local deltas merge into one global delta. Order and completeness underwrite yet another analysis result.
Pattern
1. Start from the output gap and factor out the input gap
Why: Write the difference of outputs so the quantity you control, the distance from x to a, appears as a factor you can shrink.
2. Bound every remaining factor by pre-restricting delta
Why: Agree that delta is at most one, say, to trap the other factors inside a fixed numeric bound near the point.
3. Solve for delta and take the minimum
Why: Set delta to the smaller of the pre-restriction and the value that forces the product below the tolerance, so both requirements hold at once.
4. Verify by substituting the chosen delta
Why: Confirm inputs within delta really do land within the tolerance; this closing check is mandatory in a symbolic proof.
Explain it to yourself
Discussion prompt
In Pattern: choosing your tool this move is made:
To transport a topological property, use the Intermediate or Extreme Value theorem
Why is that legal? Name the rule or definition it rests on before you read on.
Hint: If you can only say "because that is what you do", the rule is the thing to go and find.
Answer:
Reach for the intermediate value on a connected domain when you need existence of a value; reach for the extreme value on a compact domain when you need an attained maximum or minimum.
Pattern
To prove continuity of a formula, use epsilon-delta or the arithmetic of continuous functions
Why: Polynomials and rational functions are continuous for free; for a raw formula, run the epsilon-delta recipe.
To disprove continuity, use a single bad sequence
Why: One input sequence to a whose outputs miss the value at a kills continuity outright; this is the sequential demolition tool.
To transport a topological property, use the Intermediate or Extreme Value theorem
Why: Reach for the intermediate value on a connected domain when you need existence of a value; reach for the extreme value on a compact domain when you need an attained maximum or minimum.
To generalize beyond the line, use preimages of open sets
Why: The topological definition is the one that survives into metric and topological spaces, so state continuity that way when the setting is abstract.
Real world
Discussion prompt
Outside this lesson: where does Continuity & the Structural Synthesis 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 Pattern: choosing your tool 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 presents the three faces of continuity - epsilon-delta, sequential, and the topological definition by preimages of open sets - and shows how continuity carries topological properties across, through the Intermediate Value and Extreme Value theorems. It closes with a synthesis that names the arc running from logic through algebra to topology by its three recurring threads: structure-preserving maps, quotients by congruences and equivalences, and completeness and compactness. It targets the classic traps: quantifier-order errors in epsilon-delta, the pen-lifting cartoon of continuity, confusing preimage-open with image-open, and misapplying the IVT or the EVT outside their required domains.
Elimination
Eliminate the wrong options
Which delta does the rule give for this tolerance?
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 tolerance divided by seven is 0.1 over 7, which is one seventieth, about 0.0143. That is smaller than the cap of one, so the minimum is one seventieth. Substituting confirms the output stays within 0.1.
Check
Use the rule from the worked example: delta is the minimum of one and the tolerance divided by seven.
\( f(x) = x^2,\ a = 3,\quad \delta = \min\!\left(1, \tfrac{\varepsilon}{7}\right),\quad \varepsilon = 0.1 \)
Check your understanding
Which delta does the rule give for this tolerance?
Answer: A
Why: The tolerance divided by seven is 0.1 over 7, which is one seventieth, about 0.0143. That is smaller than the cap of one, so the minimum is one seventieth. Substituting confirms the output stays within 0.1.
Prediction
Predict first
Why is the Dirichlet function discontinuous at every irrational point a?
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: A sequence of rationals converges to a with all outputs one, so the outputs converge to one, not to the value zero at a.
Why: At an irrational a the value is zero, but density of the rationals gives a sequence of rationals converging to a whose outputs are all one, so the output sequence converges to one and misses zero, breaking sequential continuity.
Check
Recall the function that is one on rationals and zero on irrationals, and consider an irrational point a.
Check your understanding
Why is the Dirichlet function discontinuous at every irrational point a?
Answer: A
Why: At an irrational a the value is zero, but density of the rationals gives a sequence of rationals converging to a whose outputs are all one, so the output sequence converges to one and misses zero, breaking sequential continuity.
Check
Pick the statement that defines continuity in the language of open sets alone.
Check your understanding
Which condition is the correct topological definition of a continuous function?
Answer: A
Why: Continuity is defined by pulling open sets back: for every open set in the codomain, its preimage is open in the domain. This is the definition that generalizes to any topological space.
Commit first
Predict first
Which fact lets the Intermediate Value Theorem guarantee a root between one and two?
Commit to an answer, then rate it — certain, fairly sure, or guessing — and write the rating down before you turn the page.
Correct: The cubic is continuous on the interval and its endpoint values have opposite signs.
Why: The theorem needs continuity on the closed interval plus a sign change between the endpoints; here the value is negative at one and positive at two, bracketing zero, so a root exists strictly inside.
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
Return to the cubic that changes sign between one and two.
\( p(x) = x^3 - x - 1 \)
Check your understanding
Which fact lets the Intermediate Value Theorem guarantee a root between one and two?
Answer: A
Why: The theorem needs continuity on the closed interval plus a sign change between the endpoints; here the value is negative at one and positive at two, bracketing zero, so a root exists strictly inside.
Check
Consider a continuous function on the open interval from zero to one.
Check your understanding
Why can such a function fail to attain a maximum value?
Answer: A
Why: The Extreme Value Theorem requires a compact domain, meaning closed and bounded. The open interval is bounded but not closed, so attainment is not guaranteed and the supremum can go unreached.
Prediction
Predict first
Why is the squaring function not uniformly continuous on the whole real line?
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: For a fixed input gap the output gap grows without bound as the points move out, so no single delta works everywhere.
Why: Take points a fixed small distance apart far out on the line; their squared outputs differ by more than two however small the gap, so a single delta cannot serve every point at once, defeating uniform continuity.
Check
Recall that the squaring function is continuous at every point but consider it on the whole line.
Check your understanding
Why is the squaring function not uniformly continuous on the whole real line?
Answer: A
Why: Take points a fixed small distance apart far out on the line; their squared outputs differ by more than two however small the gap, so a single delta cannot serve every point at once, defeating uniform continuity.
Concept
You have reached the end. Step back: the thirty decks were not thirty topics but one idea seen from many angles, and continuity is where the angles visibly converge.
Three threads ran through everything: structure-preserving maps, quotients by congruences and equivalences, and the completeness-to-compactness bridge from order to topology. We name each in turn.
Picture it
Figure (svg): Five boxes left to right labeled Logic, Sets and Functions, Algebra, Linear Algebra, and Topology, showing the progression of the course.
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:
Here is the arc, five movements, each feeding the next.
Concept
Here is the arc, five movements, each feeding the next.
Figure (svg): Five boxes left to right labeled Logic, Sets and Functions, Algebra, Linear Algebra, and Topology, showing the progression of the course.
Logic gave us proof and quantifiers. Sets and functions gave the raw material and the maps. Algebra abstracted operations. Linear algebra specialized to vector spaces. Analysis and topology built the continuum, and continuity ties this last movement back to the first.
Concept
The first thread is the structure-preserving map, the morphism. In every setting the interesting functions are the ones that respect the structure at hand.
\[ \varphi(a * b) = \varphi(a) \star \varphi(b) \]
A group homomorphism respects multiplication. A linear map respects addition and scaling. A continuous map respects the open sets. Same slogan, different structure: the map commutes with the operation that defines the objects.
Concept
The morphisms of each unit line up in a single table. The right column is what each kind of map is required to preserve.
| Setting | Objects | The map preserves |
|---|---|---|
| Sets | sets | nothing extra: any function |
| Groups | groups | the product: homomorphism |
| Rings | rings | sum and product |
| Vector spaces | vector spaces | sums and scalar multiples: linear map |
| Topology | spaces | open sets: continuous map |
Read down the last column and you are reading the definition of each field. The objects change; the demand that maps preserve structure never does.
Comparison
Comparison matrix
From One table of morphisms: refill the Objects column from what you know. The rest of the table is as it appeared.
| Setting | Objects | The map preserves |
|---|---|---|
| Sets | sets | nothing extra: any function |
| Groups | groups | the product: homomorphism |
| Rings | rings | sum and product |
| Vector spaces | vector spaces | sums and scalar multiples: linear map |
| Topology | spaces | open sets: continuous map |
Intuition
The mental image is a translator who preserves meaning. Whether you combine first and then translate, or translate first and then combine, you land in the same place.
That commuting square is the beating heart of abstract mathematics. Injections, surjections, bijections, isomorphisms, kernels, images: every one of these words was defined for maps and then reused verbatim in each unit.
Concept
Because morphisms preserve structure, they preserve invariants: quantities that stay the same under the maps we allow. Invariants are how we prove two objects are genuinely different.
| Map type | An invariant it respects |
|---|---|
| Bijection of sets | cardinality |
| Group isomorphism | element orders |
| Linear isomorphism | dimension and rank |
| Homeomorphism | connectedness and compactness |
The Intermediate and Extreme Value theorems are exactly invariance statements: continuous maps carry connectedness and compactness across, and those carried properties become existence theorems.
Trade off
Comparison matrix
From Morphisms preserve invariants: every row here is a choice with a cost. Fill the An invariant it respects column, then say which row you would actually pick and what you give up for it.
| Map type | An invariant it respects |
|---|---|
| Bijection of sets | cardinality |
| Group isomorphism | element orders |
| Linear isomorphism | dimension and rank |
| Homeomorphism | connectedness and compactness |
Concept
The second thread is the quotient: you declare certain elements to be the same by an equivalence relation compatible with the structure, then work with the classes as new objects.
\[ X \longrightarrow X/\!\sim,\qquad x \longmapsto [x] \]
When the equivalence is compatible with the operations, it is called a congruence, and the operations descend to well-defined operations on the classes. That single move built most of the objects in this course.
Explain it
Discussion prompt
Explain Thread two: quotients by congruences 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:
The second thread is the quotient: you declare certain elements to be the same by an equivalence relation compatible with the structure, then work with the classes as new objects.
Concept
| You quotient | By | And get |
|---|---|---|
| the integers | congruence mod n | the ring of integers mod n |
| a group | a normal subgroup | the quotient group |
| a ring | an ideal | the quotient ring |
| a vector space | a subspace | the quotient space |
| the reals | differ by an integer | the circle |
Each row is the same construction. The condition that makes the quotient legal, normality for groups, an ideal for rings, a subspace for vector spaces, is exactly the compatibility that turns an equivalence into a congruence.
Comparison
Comparison matrix
From One table of quotients: refill the By column from what you know. The rest of the table is as it appeared.
| You quotient | By | And get |
|---|---|---|
| the integers | congruence mod n | the ring of integers mod n |
| a group | a normal subgroup | the quotient group |
| a ring | an ideal | the quotient ring |
| a vector space | a subspace | the quotient space |
| the reals | differ by an integer | the circle |
Intuition
A quotient is a chosen forgetting. You decide to stop distinguishing elements that a congruence calls equal, and the resulting coarser world is often simpler and more powerful.
Congruence mod twelve forgets everything about an hour except its clock position. That forgetting is not a loss; it is exactly the structure of a clock. Every quotient in the course is a structured forgetting of this kind.
Analogy
Discussion prompt
Explain A quotient is deliberate forgetting 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:
A quotient is a chosen forgetting. You decide to stop distinguishing elements that a congruence calls equal, and the resulting coarser world is often simpler and more powerful.
Concept
The first two threads meet in one theorem that appeared in three costumes. A structure-preserving map has a kernel, the kernel is exactly the right thing to quotient by, and the quotient is isomorphic to the image.
\[ X / \ker \varphi \ \cong \ \operatorname{im} \varphi \]
For groups this is the First Isomorphism Theorem; for rings the same statement with ideals; for vector spaces it is the rank-nullity theorem in disguise. Map, then quotient by what the map collapses, and you recover the image exactly.
Concept
The third thread is the bridge from order to topology, built on completeness. The least-upper-bound axiom is the single property that separates the reals from the rationals.
\[ \text{every nonempty } S \subseteq \mathbb{R} \text{ bounded above has a supremum} \]
From completeness flowed the monotone convergence theorem, Bolzano-Weierstrass, the equivalence of Cauchy and convergent, and finally Heine-Borel, which is what makes compactness on the line mean closed and bounded.
Counterexample
Discussion prompt
The third thread is the bridge from order to topology, built on completeness. The least-upper-bound axiom is the single property that separates the reals from the rationals.
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.
Concept
Completeness is the hidden engine under every analysis theorem in this unit. Remove it and the machinery seizes: the rationals are an ordered field where sequences that should converge do not, and where closed bounded sets are not compact.
| Property | Fails over the rationals because |
|---|---|
| Cauchy implies convergent | a Cauchy sequence can approach an absent limit |
| Bolzano-Weierstrass | a bounded sequence can have no rational limit point |
| Heine-Borel compactness | closed bounded sets still have gaps |
| Intermediate Value Theorem | a sign-changing function can skip zero at a gap |
Completeness closes the gaps, and closing the gaps is what lets continuity carry topological properties across. Order becomes topology through completeness.
Comparison
Comparison matrix
From Completeness underwrites all of analysis: refill the Fails over the rationals because column from what you know. The rest of the table is as it appeared.
| Property | Fails over the rationals because |
|---|---|
| Cauchy implies convergent | a Cauchy sequence can approach an absent limit |
| Bolzano-Weierstrass | a bounded sequence can have no rational limit point |
| Heine-Borel compactness | closed bounded sets still have gaps |
| Intermediate Value Theorem | a sign-changing function can skip zero at a gap |
Picture it
Figure (svg): A triangle with vertices labeled Logic, Algebra, and Topology, indicating the mutual bridges between them.
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:
Now name the grand synthesis. The three great subjects are not merely neighbors; they are often the same structures wearing different clothes.
Concept
Now name the grand synthesis. The three great subjects are not merely neighbors; they are often the same structures wearing different clothes.
Figure (svg): A triangle with vertices labeled Logic, Algebra, and Topology, indicating the mutual bridges between them.
Logic and algebra meet in Boolean algebra: the laws of and, or, not are exactly the laws of a Boolean ring, and the Lindenbaum construction is a quotient of formulas by logical equivalence. Algebra and topology meet in continuity and in the compactness that both propositional logic and the real line enjoy.
Concept
Two deep bridges bookend the course. Curry and Howard identified propositions with types and proofs with programs: conjunction is a product, disjunction a sum, implication a function type, and a proof is a term inhabiting that type.
\[ \text{propositions} \leftrightarrow \text{types},\qquad \text{proofs} \leftrightarrow \text{programs} \]
Stone duality closes the loop on the other side, matching Boolean algebras with certain compact spaces, so that logic, algebra, and topology become the same data viewed three ways. That is the synthesis this course was built to reveal.
Elimination
Eliminate the wrong options
Which trio names the three threads that run through the entire course?
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 course was organized around morphisms that preserve structure, quotients that identify elements by a compatible equivalence, and the completeness that bridges order to topology through compactness.
Check
One last question, about the shape of the whole course.
Check your understanding
Which trio names the three threads that run through the entire course?
Answer: A
Why: The course was organized around morphisms that preserve structure, quotients that identify elements by a compatible equivalence, and the completeness that bridges order to topology through compactness.
Concept
So what was structure all along? It was whatever a chosen class of maps is asked to preserve. Choose the maps and you have chosen the subject; the objects are simply what those maps act on.
Where this leads next: metric and topological spaces generalize the line, category theory promotes the morphism and quotient patterns to first-class objects, and Galois theory fuses groups with fields to answer questions no single unit could. You now have the vocabulary to read all three.
Connect it up
Draw it
One page, no notation unless you need it: draw how these connect — Pattern: the epsilon-delta continuity recipe · Pattern: choosing your tool · Why the naive picture is not enough · The epsilon-delta definition at a point · Continuity as a challenge-response game. Put an arrow wherever one of them is what makes another possible, and label the arrow with why.
Recap
You can now state continuity three equivalent ways, run and debug an epsilon-delta proof, and use the Intermediate and Extreme Value theorems to turn topological properties into existence results.
More than that, you can see the whole course as one structure: structure-preserving maps, quotients by congruences and equivalences, and completeness and compactness as the bridge from order to topology. Logic, algebra, and topology are three views of the same mathematics.
| Thread | One-line summary |
|---|---|
| Structure-preserving maps | functions that commute with the defining operation: morphisms, invariants, isomorphism |
| Quotients by congruences | identify elements a compatible equivalence calls equal: kernel plus First Isomorphism |
| Completeness and compactness | the least-upper-bound axiom closes the gaps and turns order into topology |
Want this taught 1-on-1? Alexander tutors Foundations of Higher Mathematics — $55/session, free consultation.