Week 4, Lecture 10: Limits of Functions

1. The Punctured Neighborhood & Formal Definition

Let's set the scene with a classic mathematical trap. Consider the rational function $f(x) = \frac{x^2 - 4}{x - 2}$. On the surface, it looks like a standard, well-behaved algebraic curve. But if you attempt to plug in exactly $x = 2$, disaster strikes. The numerator evaluates to $0$, and the denominator evaluates to $0$, producing the dreaded undefined expression $\frac{0}{0}$. The machine completely seizes up. At exactly $x = 2$, the function has no output, no coordinate, and no physical reality. It is a microscopic hole in the fabric of the graph.

However, if we step just a fraction of a hair away from that danger zone, the machine behaves beautifully. If you plug in $x = 1.999$ (approaching from the left), the machine smoothly outputs $3.999$. If you plug in $x = 2.001$ (approaching from the right), it outputs $4.001$. It is incredibly obvious that both sides of this fractured curve are aiming directly at a target height of $L = 4$. The trajectory is undeniable. But how do we prove this rigorously when the actual point $f(2)$ does not exist to anchor our $\epsilon\text{-}\delta$ box?

This is where we must upgrade the $\epsilon\text{-}\delta$ game. In Lecture 9, our input fence was defined by $|x - c| < \delta$, meaning the distance between our test input $x$ and the target $c$ had to be less than delta. That old rule legally allowed $x$ to equal $c$ (a distance of exactly zero). To investigate limits safely, we must actively forbid the machine from ever touching the broken point. We do this by declaring that the distance between $x$ and $c$ must be strictly greater than zero.

Connecting the Shorthand to the Game

When you see the compact notation $\lim_{x \to c} f(x) = L$, it is easy to read it simply as "when $x$ gets close to $c$, $f(x)$ gets close to $L$." But mathematically, that shorthand is actually a compressed trophy of victory from the $\epsilon\text{-}\delta$ game.

Writing $\lim_{x \to c} f(x) = L$ is a binding declaration that the defender has a universal winning strategy. The $L$ in the equation is the center of the adversary's $\epsilon$-corridor (the target height), and the $c$ is the center of your $\delta$-window (the target input). Saying the limit equals $L$ means that no matter how aggressively the adversary shrinks the output error $\epsilon$ around $L$, you can always find a tight enough input window $\delta$ around $c$ (without ever actually touching $c$) to force the function safely inside the adversary's boundaries.

The Punctured Upgrade: Continuity vs. Limits

To understand why limits are the ultimate tool for handling mathematical holes, let's compare exactly how this new limit definition upgrades the continuity rules you learned in Lecture 9. The secret lies entirely in the left side of the implication (the input constraint). By adding a single zero to the inequality—demanding that $0 < |x - c|$—we fundamentally change the rules of the game.

2. Interactive Lab: Resolving Indeterminate Forms

How do we actually find the target height $L$ when direct substitution gives $\frac{0}{0}$ or an infinite oscillation? Because the punctured neighborhood guarantees $x \neq c$, we can replace the broken function with an identical "twin" function that has the hole patched. Step through the three core resolution architectures below.

3. One-Sided Limits & The Rendezvous Criterion

Why do we need one-sided limits? Because many physical systems and mathematical functions do not behave the same way from all directions. Think of a thermostat kicking on at exactly 20 degrees, or a tax bracket jumping to a new rate at exactly $80,000. These systems have a hard cutoff where the rules fundamentally change depending on which side of the boundary you are standing on.

On a one-dimensional number line, there are two distinct directions from which you can walk toward a target point $x = c$: you can approach from the negative side (through numbers smaller than $c$), or you can approach from the positive side (through numbers larger than $c$). By splitting the $\epsilon$-$\delta$ game in half, we can lock onto the trajectory of a function even if it suffers a massive structural jump at $c$.

4. Limit Laws & Algebraic Surgery ($\frac{0}{0}$)

Once we know that two individual limits $\lim_{x \to c} f(x) = L$ and $\lim_{x \to c} g(x) = M$ exist as finite real numbers, we do not need to play the $\epsilon\text{-}\delta$ game from scratch every time we combine them. The Algebraic Limit Laws guarantee that limits respect standard arithmetic:

The "Lego Brick" Theory of Limits

If we had to play the rigorous $\epsilon\text{-}\delta$ game every single time we encountered a new function, calculus would grind to a halt. The Algebraic Limit Laws are our escape hatch. They guarantee that once you have proven the limits of basic, fundamental building blocks (like $f(x) = x$ or $g(x) = c$), you can snap them together like Lego bricks to build massive, complicated functions without ever having to touch an $\epsilon$ or $\delta$ again.

Let's unpack what these laws actually allow us to do on the factory floor:

• The Sum & Difference Law (Divide and Conquer): If you have a massive polynomial like $x^3 + 4x^2 - 7$, you do not need to evaluate the entire monstrous machine at once. This law gives you the legal right to split the limit into three separate, tiny limits. You evaluate the $x^3$ machine, the $4x^2$ machine, and the $-7$ machine individually, and then just add their final output targets together.
• The Constant Multiple Law (Volume Knobs): If a function is being multiplied by a constant number (like $5x^2$), that constant is just a volume knob. It doesn't change the underlying structure of the $x^2$ machine. This law lets you completely pull the constant out of the limit, evaluate the core machine, and then multiply the final answer by $5$.
• The Product Law (Independent Assembly Lines): If two machines are feeding into each other via multiplication, you can find the target output of the first machine, find the target output of the second machine, and simply multiply those targets. This is how we know the limit of $x^2$ is just the limit of $x$ multiplied by itself.
• The Quotient Law (The Danger Zone): Division is beautifully cooperative—until it isn't. You are perfectly allowed to split a fraction and take the limit of the top and bottom separately. However, this is the only law with a massive warning label: the bottom limit cannot equal zero.

So, what actually happens when $M = 0$ and the Quotient Law breaks? There are two distinct possibilities:

1. Case 1: Non-Zero over Zero ($\frac{L}{0}$ where $L \neq 0$): The fraction blows up toward $+\infty$ or $-\infty$ (a vertical asymptote, which we will master in Lecture 11). The finite limit Does Not Exist.
2. Case 2: Zero over Zero ($\frac{0}{0}$ Indeterminate Form): Both numerator and denominator vanish at $x = c$. This signals a hidden common factor! We perform algebraic surgery to cancel the zero-producing factor before applying the Quotient Law.

5. Trapping Wild Curves: The Squeeze Theorem

In Section 4, we solved $\frac{0}{0}$ limits using algebraic surgery—factoring polynomials or multiplying by conjugates until the problem term canceled out. Those tools work wonderfully when a function is built from clean algebraic pieces.

But what happens when you encounter a function that is so chaotic near $x = c$ that algebra has nothing to factor, and the Limit Laws are illegal to use?

When the "Lego Brick" Laws Break Down

Consider the following limit as $x$ approaches $0$:

At first glance, your instinct might be to use the Product Law from Section 4 to split this into two separate limits: $\left(\lim_{x \to 0} x^2\right) \cdot \left(\lim_{x \to 0} \cos\left(\frac{20}{x}\right)\right)$, and say that "$0 \text{ times anything is } 0$." Stop! That is a mathematical trap.

Remember the strict legal fine print on the Product Law: you are only allowed to split a limit apart if both individual limits exist as finite numbers. Look at what happens to the two pieces of our function as $x$ gets microscopic ($x = 0.1, 0.001, 0.00001$):

• Piece 1: The Wild Oscillator $\cos\left(\frac{20}{x}\right)$ Has No Limit: When $x$ gets tiny, dividing $20$ by $x$ produces a massive number ($200$, $20{,}000$, $2{,}000{,}000$). Feeding an exploding angle into cosine forces the wave to bounce wildly back and forth between $-1$ and $+1$ infinitely many times on its way to $x = 0$. Because it never settles on a single value, $\lim_{x \to 0} \cos\left(\frac{20}{x}\right)$ Does Not Exist (DNE). Therefore, the Product Law is completely off-limits!
• Piece 2: The Crushing Multiplier $x^2$ Shrinks the Wave: Even though the cosine wave is oscillating at an infinite frenzy, its output is being multiplied by $x^2$. When $x = 0.01$, $x^2 = 0.0001$, which means the wave is only allowed to bounce between $-0.0001$ and $+0.0001$. As $x \to 0$, the $x^2$ term acts like a hydraulic press, crushing the height of those wild oscillations down to zero!

How do we prove that the answer is $0$ without using the Product Law? Instead of trying to track the chaotic function directly, we build a cage around it.

Decoding the Three Rules of the Squeeze

To use the Squeeze Theorem successfully on an exam or in a proof, you must verify that three specific conditions are met. Let's translate what the formula above is actually demanding:

1. Rule 1: Build a Valid Floor and Ceiling ($g(x) \le f(x) \le h(x)$): You must surround your complicated mystery function $f(x)$ with a lower function $g(x)$ (the floor) and an upper function $h(x)$ (the ceiling). Notice that this inequality only needs to hold in a punctured neighborhood ($x \neq c$). Even if $f(x)$ explodes and is completely undefined at $x = c$ itself, the cage still works!
2. Rule 2: Pick Simple Guards You Can Actually Solve: The entire point of the Squeeze Theorem is to trade a terrifying function (like $x^2 \cos(20/x)$) for two easy functions (like $-x^2$ and $+x^2$) that have no wild oscillations, allowing you to evaluate their limits effortlessly.
3. Rule 3: The Floor and Ceiling Must Pinch Together ($\lim g(x) = \lim h(x) = L$): This is the most critical requirement. As $x \to c$, the floor limit and the ceiling limit must arrive at the exact same number $L$. What if your floor approaches $2$ and your ceiling approaches $5$? Then at $x = c$, the gap between the floor and ceiling is still $3$ units wide! Your middle function $f(x)$ still has room to bounce around inside that gap, so the Squeeze Theorem tells you nothing. They must pinch shut at a single point.

Why Is the Squeeze Theorem So Important for Calculus?

You might be wondering if the Squeeze Theorem is just a clever trick for taming weird oscillating waves. In reality, the Squeeze Theorem is the single mathematical key that unlocks all of trigonometric calculus.

Next week, when we learn how to calculate the derivative (the instantaneous rate of change) of $\sin(x)$ and $\cos(x)$ to model pendulums, soundwaves, and orbits, the entire proof comes down to evaluating one famous limit:

Think about what happens if you try to solve $\lim_{x \to 0} \frac{\sin(x)}{x}$ using the tools from Section 4. Plugging in $x = 0$ gives $\frac{\sin(0)}{0} = \frac{0}{0}$. It is an indeterminate hole! Worse yet, $\sin(x)$ is a transcendental wave, not a polynomial—you cannot factor an $x$ out of $\sin(x)$, and there are no square roots to rationalize. Algebra hits a dead end.

Summary: Inspecting the Unreachable

By introducing the punctured neighborhood ($0 < |x - c| < \delta$), we liberated our analysis from requiring $f(c)$ to exist. Whether a curve has a missing hole from a $\frac{0}{0}$ division or oscillates rapidly near a point, limits allow us to lock onto the exact destination of the trajectory.

In Lecture 11, we will push this concept to the ultimate horizon: investigating what happens when a function blows up to $+\infty$ at a vertical asymptote, or marches out to $x \to \pm\infty$ along a horizontal asymptote.