Chapter 01
Rigorous

Limits and Continuity

00 · Symbol Glossary

$\mathbb{R}$Blackboard R

The set of all real numbers — every number on the number line: negative, zero, positive, decimals, irrationals like π\pi. The double-struck style distinguishes it from an ordinary letter RR. Every function and limit in this chapter operates on real numbers.

$f(x)$f of x — function

A rule that assigns exactly one output to each input xx. Read "f of x." The letter ff names the function; xx is the input variable. f(3)f(3) means "evaluate ff at x=3x=3."

$\lim_{x \to a} f(x)$limit of f of x as x approaches a

The value that f(x)f(x) gets arbitrarily close to as xx gets arbitrarily close to aa — without necessarily ever equaling aa. This is the central object of the chapter. Read aloud: "the limit as x approaches a of f of x."

$\to$approaches / tends to

An arrow meaning "gets arbitrarily close to, without necessarily reaching." xax \to a means xx moves toward aa from either or both sides, never required to land exactly on aa.

$x \to a^+$x approaches a from the right

The right-hand limit: xx approaches aa only through values greater than aa. The superscript ++ marks "from above."

$x \to a^-$x approaches a from the left

The left-hand limit: xx approaches aa only through values less than aa. The superscript - marks "from below."

$\varepsilon$epsilon — target tolerance

Greek lowercase epsilon. In the formal definition of a limit, ε\varepsilon is an arbitrarily small positive number that measures how close f(x)f(x) must be to the limit value. Chosen first, by whoever is testing the limit.

$\delta$delta — input tolerance

Greek lowercase delta. A positive number measuring how close xx must be to aa to guarantee f(x)f(x) lands within ε\varepsilon of the limit. Chosen second, in response to ε\varepsilon, and typically depends on it.

$\forall,\ \exists$"for all", "there exists"

Quantifiers borrowed from formal logic. ε>0\forall\,\varepsilon>0 reads "for every positive ε\varepsilon, no matter how small." δ>0\exists\,\delta>0 reads "there exists some positive δ\delta." Together they build the precise limit definition in Section 02.

$\lvert x-a \rvert$absolute value of x minus a — distance

The distance between xx and aa on the number line, always non-negative. xa<δ\lvert x-a\rvert < \delta means "xx is within δ\delta of aa."

$\infty,\ -\infty$infinity, negative infinity

Not real numbers — symbols describing unbounded growth. limxaf(x)=\lim_{x\to a}f(x)=\infty means f(x)f(x) grows without bound as xax\to a; it does not mean the limit exists as a number. xx\to\infty means xx grows without bound.

$\text{DNE}$does not exist

Shorthand written when a limit fails to settle on a single finite value — the left- and right-hand limits disagree, or the function oscillates, or it grows without bound.


01 · The Idea of a Limit

Suppose f(x)=x21x1f(x) = \dfrac{x^2-1}{x-1}. Plug in x=1x=1 directly and you get 00\dfrac{0}{0} — undefined. But look at what happens as xx gets close to 11 without touching it: at x=0.9x=0.9, f(0.9)=1.9f(0.9)=1.9; at x=0.99x=0.99, f(0.99)=1.99f(0.99)=1.99; at x=1.01x=1.01, f(1.01)=2.01f(1.01)=2.01. The outputs are closing in on 22 from both sides, even though f(1)f(1) itself is undefined.

That closing-in value is the limit. A limit describes the trend of a function near a point, independent of whether the function is even defined at that point.

Definition — Limit (informal)

We write

limxaf(x)=L\lim_{x \to a} f(x) = L

if f(x)f(x) gets arbitrarily close to the number LL whenever xx gets sufficiently close to aa, from either side, with xax \neq a.

LL — the limit value, a single real number (when the limit exists).

xax \neq a — the behavior at aa is irrelevant; only the behavior near aa matters.

Example — Evaluating a limit by direct substitution

For a well-behaved function like f(x)=3x+2f(x) = 3x + 2, the limit as x4x \to 4 is found by simply plugging in:

limx4(3x+2)=3(4)+2=14\lim_{x \to 4} (3x+2) = 3(4)+2 = 14

The 1212 came from 3×43\times4; the final 1414 from 12+212+2. This works because ff is a polynomial — polynomials have no gaps, jumps, or holes anywhere, so the limit always equals the direct evaluation.

❌ Direct substitution gives 0/0

For f(x)=x21x1f(x) = \dfrac{x^2-1}{x-1} and a=1a=1: substituting gives 12111=00\dfrac{1^2-1}{1-1} = \dfrac{0}{0}.

Why it breaks: 00\frac{0}{0} is an indeterminate form — it does not mean the limit is 00, or 11, or undefined outright. It means direct substitution gives no information and a different method is required.

Correct approach: factor first. x21=(x1)(x+1)x^2-1=(x-1)(x+1), so for x1x\neq1: (x1)(x+1)x1=x+1\dfrac{(x-1)(x+1)}{x-1} = x+1. Since the limit only cares about xx near 11, not at 11, the cancellation is valid, and limx1(x+1)=2\lim_{x\to1}(x+1) = 2 — matching the numerical trend from above.

Consequence: whenever substitution yields 00\frac{0}{0}, algebraic simplification (factoring, rationalizing) must be tried before concluding anything.


02 · The Formal ε\varepsilonδ\delta Definition

The informal phrase "gets arbitrarily close" needs a precise meaning to be usable in proofs. The formal definition pins it down with two tolerances: how close the output must be (ε\varepsilon), and how close the input must be to guarantee it (δ\delta).

Definition — Limit (formal, epsilon-delta)

limxaf(x)=L\lim_{x\to a} f(x) = L means:

ε>0, δ>0 such that 0<xa<δ    f(x)L<ε\forall\, \varepsilon > 0,\ \exists\, \delta > 0 \text{ such that } 0 < \lvert x-a\rvert < \delta \implies \lvert f(x)-L\rvert < \varepsilon

Read left to right: "for every tolerance ε>0\varepsilon>0 someone hands you, you can find a tolerance δ>0\delta>0 such that whenever xx is within δ\delta of aa (but not equal to aa), f(x)f(x) is guaranteed to land within ε\varepsilon of LL."

0<xa0 < \lvert x-a \rvert — strictly positive, meaning xax \neq a; the point aa itself is excluded.

Step-by-step — Prove $\lim_{x\to3}(2x-1)=5$ using epsilon-delta
1
Start from the target inequality: we need f(x)L<ε\lvert f(x)-L\rvert < \varepsilon, i.e. (2x1)5<ε\lvert (2x-1)-5\rvert < \varepsilon.
2
Simplify the expression inside the bars: (2x1)5=2x6=2(x3)(2x-1)-5 = 2x-6 = 2(x-3), so the condition becomes 2(x3)<ε\lvert 2(x-3)\rvert < \varepsilon, i.e. 2x3<ε2\lvert x-3\rvert < \varepsilon.
3
Isolate x3\lvert x-3\rvert: divide both sides by 22: x3<ε/2\lvert x-3\rvert < \varepsilon/2.
4
Choose δ\delta: set δ=ε/2\delta = \varepsilon/2. This choice guarantees that whenever 0<x3<δ0<\lvert x-3\rvert<\delta, reversing steps 3–2 gives f(x)5<ε\lvert f(x)-5\rvert<\varepsilon.
5
Numeric check: if ε=0.1\varepsilon = 0.1, then δ=0.05\delta = 0.05. Take x=3.03x = 3.03 (within δ\delta of 33, since 3.033=0.03<0.05\lvert3.03-3\rvert=0.03<0.05). Then f(3.03)=2(3.03)1=5.06f(3.03) = 2(3.03)-1 = 5.06, and 5.065=0.06<0.1\lvert5.06-5\rvert=0.06<0.1. ✓
Why bother with epsilon-delta if the answer is obvious?

For a straight line, the limit is visually obvious. The formal definition matters because it is the only tool precise enough to handle functions where intuition fails — oscillating functions, piecewise functions, and rigorous proofs of the limit laws used constantly from here on.


03 · One-Sided Limits and Existence

A limit is a single statement about behavior from both directions at once. Sometimes the trend from the left disagrees with the trend from the right — in that case, the two-sided limit does not exist, even though each one-sided limit exists individually.

Definition — One-Sided and Two-Sided Limits

limxaf(x)=L1\lim_{x\to a^-}f(x)=L_1 — the value f(x)f(x) approaches as xx approaches aa using only values less than aa.

limxa+f(x)=L2\lim_{x\to a^+}f(x)=L_2 — the value f(x)f(x) approaches using only values greater than aa.

limxaf(x) exists    limxaf(x)=limxa+f(x)\lim_{x\to a} f(x) \text{ exists} \iff \lim_{x\to a^-}f(x) = \lim_{x\to a^+}f(x)

The two-sided limit exists if and only if both one-sided limits exist and agree.

Example — Step function with a jump

Let f(x)={1x<03x0f(x) = \begin{cases} 1 & x < 0 \\ 3 & x \geq 0 \end{cases}.

From the left: limx0f(x)=1\lim_{x\to0^-}f(x) = 1 (values just below 00 all give output 11).

From the right: limx0+f(x)=3\lim_{x\to0^+}f(x) = 3 (values just above 00 all give output 33).

Since 131 \neq 3, limx0f(x)\lim_{x\to0}f(x) does not exist. This is a jump discontinuity — revisited in Section 05.

❌ Assuming a limit exists because f(a) is defined

For f(x)={x2x210x=2f(x) = \begin{cases} x^2 & x \neq 2 \\ 10 & x = 2 \end{cases}, a student might reason: "f(2)=10f(2)=10 is defined, so the limit at 22 is 1010."

Why it breaks: the limit only cares about values near 22, not the value assigned at 22. As x2x\to2 from either side, f(x)=x24f(x)=x^2 \to 4, since the definition f(x)=10f(x)=10 only kicks in exactly at x=2x=2, a single point that the approaching values never touch.

Correct value: limx2f(x)=4f(2)=10\lim_{x\to2}f(x) = 4 \neq f(2) = 10.

Consequence: the limit and the function value at a point are two separate quantities. They coincide for continuous functions (Section 05) but can disagree elsewhere, as here.


04 · Limit Laws and Computing Limits

Once a few basic limits are known, complicated limits can be built up algebraically rather than re-proving epsilon-delta every time.

Definition — Limit Laws

Suppose limxaf(x)=L\lim_{x\to a}f(x)=L and limxag(x)=M\lim_{x\to a}g(x)=M, both finite. Then:

limxa[f(x)±g(x)]=L±Mlimxa[cf(x)]=cL\lim_{x\to a}\big[f(x)\pm g(x)\big] = L \pm M \qquad \lim_{x\to a}\big[c\cdot f(x)\big] = cL
limxa[f(x)g(x)]=LMlimxaf(x)g(x)=LM  (M0)\lim_{x\to a}\big[f(x)g(x)\big] = LM \qquad \lim_{x\to a}\frac{f(x)}{g(x)} = \frac{L}{M} \ \ (M\neq0)

cc — any constant (real number), not depending on xx.

Step-by-step — Evaluate $\lim_{x\to2}\dfrac{x^3-4x}{x+1}$
1
Check the denominator's limit first: limx2(x+1)=2+1=30\lim_{x\to2}(x+1) = 2+1=3 \neq 0, so the quotient law applies directly — no factoring needed.
2
Evaluate the numerator's limit: limx2(x34x)=234(2)=88=0\lim_{x\to2}(x^3-4x) = 2^3 - 4(2) = 8-8=0. The 88 comes from 23=2×2×22^3=2\times2\times2; the second 88 from 4×24\times2.
3
Apply the quotient law: 03=0\dfrac{0}{3}=0.
4
Conclude: limx2x34xx+1=0\lim_{x\to2}\dfrac{x^3-4x}{x+1} = 0. Note this is different from the earlier 00\frac00 case — here the numerator is 00 but the denominator is not, which is a perfectly valid limit of 00, not an indeterminate form.

05 · Continuity

A function is continuous at a point if its graph has no hole, jump, or break there — you could trace it through that point without lifting a pen. This intuition is formalized by tying together the limit and the function value.

Definition — Continuity at a Point

ff is continuous at x=ax=a if all three conditions hold:

1. f(a) is defined2. limxaf(x) exists3. limxaf(x)=f(a)\text{1. } f(a) \text{ is defined} \qquad \text{2. } \lim_{x\to a}f(x) \text{ exists} \qquad \text{3. } \lim_{x\to a}f(x) = f(a)

If ff is continuous at every point of an interval, it is continuous on that interval.

Example — Verifying continuity

f(x)=x2+1f(x) = x^2 + 1 at a=3a = 3: f(3)=10f(3) = 10 (defined). limx3(x2+1)=32+1=10\lim_{x\to3}(x^2+1) = 3^2+1=10 (exists). Both equal 1010. All three conditions hold — continuous at x=3x=3.

Three flavors of discontinuity

Removable: the limit exists but disagrees with (or is missing from) f(a)f(a) — a single-point hole, as in Section 03's x2x^2 example. Jump: the one-sided limits exist but disagree, as in the step function above. Infinite: f(x)±f(x)\to\pm\infty near aa, typically from a zero denominator — covered next.

❌ Infinite discontinuity at a vertical asymptote

For f(x)=1x2f(x) = \dfrac{1}{x-2} at a=2a=2: f(2)f(2) is undefined (division by 00). Checking the trend: as x2+x\to2^+, x2x-2 is a tiny positive number, so f(x)+f(x)\to+\infty. As x2x\to2^-, x2x-2 is a tiny negative number, so f(x)f(x)\to-\infty.

Why it breaks: the one-sided trends do not approach a finite number — they grow without bound in opposite directions. Neither limx2f(x)\lim_{x\to2^-}f(x) nor limx2+f(x)\lim_{x\to2^+}f(x) is a real number, so condition 2 of continuity fails outright, and the two-sided limit does not exist.

Consequence: ff has an infinite discontinuity at x=2x=2, marked graphically by a vertical asymptote — the curve never crosses the vertical line x=2x=2 and shoots off toward ±\pm\infty on either side of it.


06 · The Intermediate Value Theorem

Continuity has a powerful consequence: a continuous function cannot skip over values. If it starts below a target height and ends above it, it must cross that height somewhere in between.

Definition — Intermediate Value Theorem (IVT)

If ff is continuous on the closed interval [a,b][a,b], and NN is any number between f(a)f(a) and f(b)f(b), then there exists at least one c[a,b]c \in [a,b] such that f(c)=Nf(c) = N.

[a,b][a,b] — the closed interval from aa to bb, including both endpoints.

Example — Locating a root with the IVT

Let f(x)=x3x1f(x) = x^3 - x - 1. This is a polynomial, so it is continuous everywhere. Compute f(1)=111=1f(1) = 1-1-1=-1 and f(2)=821=5f(2) = 8-2-1=5.

Since ff is continuous on [1,2][1,2] and 00 lies between 1-1 and 55, the IVT guarantees some c[1,2]c \in [1,2] with f(c)=0f(c)=0 — a root of ff exists in that interval, even without solving for it exactly. This is the theoretical justification behind numerical root-finding methods used throughout applied mathematics.

❌ Applying the IVT to a discontinuous function

Let f(x)=1x1f(x) = \dfrac{1}{x-1} on [0,2][0,2], with f(0)=1f(0)=-1 and f(2)=1f(2)=1. A student might claim: "00 is between 1-1 and 11, so by the IVT some c[0,2]c\in[0,2] gives f(c)=0f(c)=0."

Why it breaks: ff is not continuous on [0,2][0,2] — it has an infinite discontinuity at x=1x=1 (Section 05), so the hypothesis of the IVT is not satisfied.

Correct conclusion: in fact 1x1=0\frac{1}{x-1}=0 has no solution at all — the numerator 11 can never equal 00. The IVT's guarantee simply does not apply here.

Consequence: always verify continuity on the entire interval before invoking the IVT — checking only the endpoints is not sufficient.


07 · Exercises

EXERCISE 1.1

Try direct substitution first. If it gives 00\frac{0}{0}, factor the numerator and cancel the common factor with the denominator.

Direct substitution gives 32933=00\dfrac{3^2-9}{3-3}=\dfrac{0}{0}, indeterminate. Factor: x29=(x3)(x+3)x^2-9=(x-3)(x+3), so for x3x\neq3: (x3)(x+3)x3=x+3\dfrac{(x-3)(x+3)}{x-3}=x+3. Then limx3(x+3)=3+3=6\lim_{x\to3}(x+3)=3+3=6.

Evaluate limx3x29x3\lim_{x\to3}\dfrac{x^2-9}{x-3}. Show the indeterminate form and the factoring step explicitly.

EXERCISE 1.2

Compute the left-hand and right-hand limits separately by plugging into each piece of the definition, then compare.

f(x)={2x+1x<15x=1x2+2x>1f(x)=\begin{cases}2x+1 & x<1\\5 & x=1\\x^2+2 & x>1\end{cases}. Left: limx1(2x+1)=2(1)+1=3\lim_{x\to1^-}(2x+1)=2(1)+1=3. Right: limx1+(x2+2)=12+2=3\lim_{x\to1^+}(x^2+2)=1^2+2=3. Both equal 33, so limx1f(x)=3\lim_{x\to1}f(x)=3. But f(1)=53f(1)=5\neq3, so the limit exists yet ff is not continuous at x=1x=1 (a removable discontinuity).

For f(x)={2x+1x<15x=1x2+2x>1f(x)=\begin{cases}2x+1 & x<1\\5 & x=1\\x^2+2 & x>1\end{cases}, find limx1f(x)\lim_{x\to1}f(x) and determine whether ff is continuous at x=1x=1.

EXERCISE 1.3

Use the formal definition: simplify f(x)L<ε\lvert f(x)-L\rvert < \varepsilon down to an inequality in xa\lvert x-a\rvert, then read off δ\delta in terms of ε\varepsilon.

Need (4x3)5<ε    4x8<ε    4x2<ε    x2<ε/4\lvert(4x-3)-5\rvert<\varepsilon \iff \lvert4x-8\rvert<\varepsilon \iff 4\lvert x-2\rvert<\varepsilon \iff \lvert x-2\rvert<\varepsilon/4. Choose δ=ε/4\delta=\varepsilon/4. Check with ε=0.08\varepsilon=0.08: δ=0.02\delta=0.02. At x=2.01x=2.01: f(2.01)=4(2.01)3=5.04f(2.01)=4(2.01)-3=5.04, and 5.045=0.04<0.08\lvert5.04-5\rvert=0.04<0.08. ✓

Using the epsilon-delta definition, find δ\delta in terms of ε\varepsilon to prove limx2(4x3)=5\lim_{x\to2}(4x-3)=5. Verify numerically with ε=0.08\varepsilon=0.08.

EXERCISE 1.4

Identify what kind of expression sits in the denominator near the point in question, and check the one-sided trends separately.

f(x)=x+3x29=x+3(x3)(x+3)f(x)=\dfrac{x+3}{x^2-9}=\dfrac{x+3}{(x-3)(x+3)}. At x=3x=-3: for x3x\neq-3, this simplifies to 1x3\dfrac{1}{x-3}, which is defined and finite at x=3x=-3 (value 14-\frac14) — so x=3x=-3 is a removable discontinuity (a hole), not infinite. At x=3x=3: the simplified form 1x3\frac{1}{x-3} blows up as x3x\to3, giving ++\infty from the right and -\infty from the left — an infinite discontinuity.

For f(x)=x+3x29f(x)=\dfrac{x+3}{x^2-9}, classify the discontinuities at x=3x=-3 and at x=3x=3 as removable, jump, or infinite. Justify each with the appropriate limit computation.

EXERCISE 1.5

Confirm ff is continuous on the whole interval, evaluate both endpoints, and check whether 00 lies between them.

f(x)=x33x+1f(x)=x^3-3x+1 is a polynomial, hence continuous everywhere, including on [0,2][0,2]. f(0)=00+1=1f(0)=0-0+1=1. f(2)=86+1=3f(2)=8-6+1=3. Since 00 is not between 11 and 33, the IVT as stated does not directly guarantee a root on [0,2][0,2] using these endpoints — check f(1)=13+1=1f(1)=1-3+1=-1 instead: f(0)=1>0f(0)=1>0 and f(1)=1<0f(1)=-1<0, and 00 lies between 11 and 1-1, so by the IVT there is a root in (0,1)(0,1).

Show that f(x)=x33x+1f(x)=x^3-3x+1 has a root somewhere in the interval (0,1)(0,1) using the Intermediate Value Theorem.

EXERCISE 1.6

A quoted price feed that "jumps" between two levels with no intermediate value is the discrete analogue of a function with a jump discontinuity.

Model the fee schedule as f(x)={0.10x0x<10,0000.08xx10,000f(x)=\begin{cases}0.10x & 0\le x<10{,}000\\0.08x & x\ge10{,}000\end{cases} where xx is trade size. Left-hand limit at x=10,000x=10{,}000: 0.10(10,000)=1,0000.10(10{,}000)=1{,}000. Right-hand limit: 0.08(10,000)=8000.08(10{,}000)=800. Since 1,0008001{,}000\neq800, the two one-sided limits disagree — the fee function has a jump discontinuity at x=10,000x=10{,}000: a trade of 9,9999{,}999 costs more in fees than a trade of 10,00010{,}000, a discontinuous (and exploitable) jump in the schedule.

A trading fee schedule charges 10%10\% on trade sizes under 10,00010{,}000 dollars and 8%8\% on trade sizes at or above 10,00010{,}000 dollars. Model this as a piecewise function of trade size xx, and show it has a jump discontinuity at x=10,000x=10{,}000 by computing both one-sided limits there.


08 · Chapter Summary

ConceptFormula / Rule
Limit (informal)f(x)Lf(x)\to L as xax\to a, xax\neq a
Limit (formal)ε>0δ>0:0<xa<δ    f(x)L<ε\forall\varepsilon>0\,\exists\delta>0: 0<\lvert x-a\rvert<\delta\implies\lvert f(x)-L\rvert<\varepsilon
One-sided limitslimxaf(x)\lim_{x\to a}f(x) exists     \iff left limit = right limit
Limit lawsSum, product, quotient limits combine from known limits
Indeterminate form00\frac{0}{0}: substitution fails; factor or simplify first
Continuity at aaf(a)f(a) defined, limit exists, and they're equal
Removable discontinuityLimit exists but f(a)\neq f(a) (or f(a)f(a) undefined)
Jump discontinuityOne-sided limits exist but disagree
Infinite discontinuityOne-sided limit(s) are ±\pm\infty; vertical asymptote
IVTContinuous on [a,b][a,b]     \implies hits every value between f(a)f(a) and f(b)f(b)

Next: Chapter 02 — Derivatives: Definition and Rules, where the limit machinery built here is used to define the instantaneous rate of change.