4.1.9 · D4Calculus I — Limits & Derivatives

Exercises — Epsilon-delta definition of a limit — formal proofs

2,138 words10 min readBack to topic

Before we start, two words we will lean on constantly:

Figure — Epsilon-delta definition of a limit — formal proofs

The picture shows both distances at once: a -wide window around on the horizontal axis, and an -tall band around on the vertical axis. Winning means every inside the window has its inside the band.


Level 1 — Recognition

You only need to read the definition and the picture correctly.

Exercise 1.1

State, symbol for symbol, what means. Then say in one sentence what the strict inequality contributes.

Recall Solution 1.1

The part excludes itself: a limit describes the approach to , and must not care whether is even defined at .

Exercise 1.2

An adversary picks . You must reply with a . Who chose first, and may your depend on the adversary's ?

Recall Solution 1.2

The adversary chooses first; you reply with second. Your may depend on (that is the whole point — a smaller ring demands a smaller wobble). Your may not depend on , because is chosen after you commit to .

Exercise 1.3

For the constant function , prove . (Hint: what is the output distance here, for any ?)

Recall Solution 1.3

The output distance is , which is for every , regardless of . Proof. Let . Choose (any positive number works). If , then .


Level 2 — Application

Plug into the recipe on functions where the "factor" is a constant — no bounding needed yet.

Exercise 2.1

Prove and report the in terms of .

Recall Solution 2.1

Scratchwork. Output distance: The factor is the constant . To force we need . Take . Proof. Let , choose . If then

Exercise 2.2

Prove . (Watch the sign inside the absolute value.)

Recall Solution 2.2

Scratchwork. . The absolute value swallows the minus sign; the factor is . Take . Proof. Let , choose . If then

Exercise 2.3

Prove , giving as a multiple of .

Recall Solution 2.3

. The factor is . To make we need . Take . Proof. Let , choose . If then Note the factor was less than 1, so came out bigger than — a gentle function tolerates a wide window.


Level 3 — Analysis

Now the factor depends on — you must Bound it with a preliminary restriction and take a .

Exercise 3.1

Prove .

Recall Solution 3.1

Scratchwork. . The factor is not constant — bound it first. Impose the preliminary , so , hence and . Then Need , i.e. . Enforce both with . Proof. Let , . If : since , ; since ,

Exercise 3.2

Prove .

Recall Solution 3.2

Scratchwork. , so . Preliminary gives , so and . Then . Need . Take . Proof. Let , . If : and

Exercise 3.3

Prove . (Here the factor sits in a denominator — you must bound it away from zero.)

Recall Solution 3.3

Scratchwork. The dangerous piece is : if could reach it would blow up. So we must keep away from . Preliminary gives . Since on this interval, is positive, so ; hence . Then Need , i.e. . Take . Proof. Let , . If : then , so and ; therefore

Figure — Epsilon-delta definition of a limit — formal proofs

The figure shows why the preliminary "" matters for : it fences into the safe zone , where is tame; without the fence, near would make the output distance explode.


Level 4 — Synthesis

Combine ideas: build a proof from a limit law, or prove a limit is false, or handle a one-sided approach.

Exercise 4.1

Prove , even though has no limit at .

Recall Solution 4.1

Key fact: for every , so whenever . Thus The wild oscillating factor is bounded by the constant — no preliminary restriction needed. Take . Proof. Let , choose . If then This is the squeeze idea living inside an proof; see Limit Laws (sum, product, quotient).

Exercise 4.2

Prove that does not exist. (Use the negation of the definition.)

Recall Solution 4.2

Here for and for . Negation: show some defeats every and every candidate . Take . Fix any candidate and any . Construct two explicit points inside the window: let and . Both satisfy , and , . If both were within of , then by the triangle inequality a strict contradiction ( is false). So at least one of lies from . Since we produced such points for every and every , the limit does not exist. Why ? The two output values are apart; a band of half-width (full width ) cannot contain both.

Exercise 4.3

Using the definition of a one-sided limit, prove . (One-sided means: replace with ; see One-sided limits.)

Recall Solution 4.3

For , output distance is . We want , which (squaring, valid since both sides ) is . Take . Proof. Let , choose . If then . Note the non-linear link: to shrink the output by a factor, the input window must shrink by its square.


Level 5 — Mastery

Prove general statements, choose 's cleverly, and chain proofs.

Exercise 5.1

Suppose and . Prove the sum law directly from the definition.

Recall Solution 5.1

Idea: we must control the output distance . Split it with the triangle inequality: Each piece can be forced below , so the sum stays below . Splitting in half is the trick. Proof. Let . Since , there is with . Since , there is with . Choose . If , then both conditions hold, so This is exactly the sum rule in Limit Laws (sum, product, quotient).

Exercise 5.2

Prove .

Recall Solution 5.2

Scratchwork. Factor: , so . Preliminary gives . On this interval is increasing and positive; at it equals , so . Then . Need . Take . Proof. Let , . If : since , ; since ,

Exercise 5.3

The Definition of the derivative of at is . Prove, with (in the variable , target ), that this limit is .

Recall Solution 5.3

Simplify first (for ): So the output distance is . The factor is the constant . Take . Proof. Let , choose . If , then the quotient equals (defined since ), and Hence , matching at .


Active recall

What proves ?
, since the output distance is .
For , what works?
— bound then match .
For , why bound from below?
The factor is ; it blows up near , so we need , giving .
For , what ?
, using .
For the one-sided , what ?
, since .
In the sum-law proof, why split into ?
So exactly.
For , what ?
, bounding on .

Connections

  • Limit Laws (sum, product, quotient) — Exercises 4.1 and 5.1 are these laws proved raw.
  • One-sided limits — Exercise 4.3 uses the one-sided window .
  • Definition of the derivative — Exercise 5.3 evaluates a difference quotient by .
  • Continuity — every convergent example here is also a proof that the function is continuous at that point.
  • Uniform continuity — notice each above depended on ; making one serve all is the next level.
  • Sequences and their limits — the discrete cousin, with replacing .