Calculus › Section B · Limits › Topic B.2

Finding limits from graphs and tables

Before we learn to compute limits with algebra, we learn to see them. Graphs and tables show where a function is heading — and just as importantly, they reveal when it is heading nowhere at all.

Where we are

Topic B.1 asked for the slope of a tangent line and the speed of a ball at one instant, questions that demanded to know what a quantity approaches, not what it equals. That is a limit. This topic gives you two instruments for detecting one: the graph, which shows the whole behaviour at a glance, and the table, which zooms in numerically. Neither instrument is proof — the next topic supplies that — but both build the instinct you need: predicting a limit, and recognising at once the three situations where no limit exists. Learn to read these pictures and numbers well, and the algebra later will feel like it is confirming what you already suspected.

The idea, stated plainly

We write \(\lim_{x \to a} f(x) = L\) and say: the values of \(f(x)\) can be made as close as we like to \(L\), simply by taking \(x\) sufficiently close to \(a\), on either side, but not equal to \(a\). Three consequences matter more than the wording. First, the limit is about approach, not arrival: \(x\) never lands on \(a\). Second, \(f(a)\) itself is irrelevant; the function may be undefined there, or defined as something else entirely, and the limit stands regardless. Third, both sides must agree: the left approach and the right approach have to be heading to the same \(L\).

Reading a limit from a graph

To read \(\lim_{x \to a} f(x)\) from a graph, put a finger on the curve well to the left of \(a\) and trace it rightward, watching the height. Where is the height going? Now do the same from the right side, tracing leftward. If both traces are heading to the same height \(L\), the limit is \(L\). It does not matter whether the curve actually has a point at that height, or a point at \(x = a\) at all. A hole, a dot plotted elsewhere, even a total absence of the graph at \(x = a\): none of these move the limit, because the limit only ever asked where the neighbourhood was heading.

Worked example: a graph with a hole

Suppose the graph of \(f\) is the straight line \(y = x + 1\), except that at \(x = 2\) the point is missing, and a separate dot is drawn at \((2, 5)\). So \(f(2) = 5\), but what is \(\lim_{x \to 2} f(x)\)? Trace in from the left: heights climb toward \(3\). Trace in from the right: heights fall toward \(3\). Both sides agree on \(3\), so the limit is \(3\), and the plotted dot at height \(5\) is irrelevant to it. The value and the limit answer different questions: \(f(2)\) says where the point is; the limit says where the graph is heading.

Reading a limit from a table

A table is a microscope. To estimate \(\lim_{x \to a} f(x)\), compute \(f(x)\) at values creeping toward \(a\): \(a - 0.1,\ a - 0.01,\ a - 0.001\) from the left, and the matching values from the right. If the outputs from both directions are settling on the same number, that number is your estimate. The discipline is in the details: always sample both sides, shrink the step steadily rather than once, and state your conclusion as an estimate. A table can strongly suggest a value, but it inspects only the points you chose, and a function can misbehave exactly where you happened not to look.

Worked example: a cyclist's speed from a table

A cyclist's position is \(s(t) = t^2 + 3t\) metres after \(t\) seconds. Estimate the instantaneous speed at \(t = 2\) using average speeds over shrinking intervals. From the right: over \([2, 2.1]\), \(\frac{10.71 - 10}{0.1} = 7.1\) m/s; over \([2, 2.01]\), \(\frac{10.0701 - 10}{0.01} = 7.01\) m/s. From the left: over \([1.9, 2]\), \(\frac{10 - 9.31}{0.1} = 6.9\) m/s; over \([1.99, 2]\), \(6.99\) m/s. Both sides close in on \(7\), so the speed at \(t = 2\) is about \(7\) metres per second. Notice what just happened: this is the instantaneous velocity problem from B.1, done with a table.

When limits fail: the two sides disagree

The first way a limit fails is a split verdict. If the left approach heads to one number and the right approach to another, no single \(L\) exists, and we say the limit does not exist. Step functions do this at every step, and piecewise functions do it wherever their pieces do not meet. A phone plan that bills every started minute as a full minute jumps at each whole minute: approach 3 minutes from below and you are heading for a 3-minute charge; from above, a 4-minute charge. At the instant \(x = 3\) itself, the function has one value, but the limit does not exist.

When limits fail: blow-ups and wiggles

Two more failure modes. Unbounded behaviour: near \(x = 0\), \(f(x) = \frac{1}{x^2}\) grows without bound, shooting past every height you can name. No finite \(L\) can play the target, so the limit does not exist; the vertical asymptote tells the story, and topic B.5 studies it properly. Oscillation: \(f(x) = \sin\frac{1}{x}\) wiggles between \(-1\) and \(1\) infinitely often in any neighbourhood of \(0\), never settling. Again, no limit. Three signatures, then: sides that disagree, outputs that explode, outputs that refuse to settle. Spot any of the three and the answer is 'no limit there'.

Why tables and graphs are not proof

Both instruments share one weakness: they sample finitely many points. A function could agree with your table to ten decimal places and still leap sideways between the points you checked, or wiggle at a scale your graphing window cannot show. This is why the subject needs the algebra of the next topic, and why, deeper down, mathematicians built the epsilon-delta definition: name any output tolerance \(\varepsilon\), however tiny, and the definition demands an input distance \(\delta\) that guarantees it. You do not need to compute with that yet; just know the machinery exists, and that it, not your graphing calculator, is the final judge.

Worked example: spotting failure from numbers

Consider \(f(x) = \dfrac{|x|}{x}\) near \(x = 0\). The function is undefined at \(0\), but that is not the story. For any positive \(x\), \(|x| = x\), so \(f(x) = 1\): the table from the right reads \(1, 1, 1\). For any negative \(x\), \(|x| = -x\), so \(f(x) = -1\): the table from the left reads \(-1, -1, -1\). Each side settles nicely, but on different numbers, so \(\lim_{x \to 0} f(x)\) does not exist. The lesson: sample both sides, always. A one-sided table here would have seemed perfect and been completely wrong about the whole limit.

Common mistakes to avoid

The recurring errors: assuming the limit must equal \(f(a)\), when the two answer different questions and either can exist without the other; concluding 'no limit' merely because \(f(a)\) is undefined, when a hole still has a limit; sampling only one side, which misses every disagreement at a jump; trusting one small step instead of shrinking it repeatedly; and confusing oscillation, which stays bounded, with unbounded growth. Classify the failure correctly, because later topics handle each one with different tools.

Key takeaways

The limit of \(f\) at \(a\) is the height the outputs approach from both sides, and \(f(a)\) itself is irrelevant. From a graph, trace in from the left and the right and see whether both aim at the same height. From a table, shrink the step on both sides and watch the outputs settle. Both instruments produce evidence, never proof. Limits fail in three recognisable ways: the two sides disagree, the outputs grow without bound at a vertical asymptote, or the outputs oscillate without settling. Learn to spot all three, and bring your suspicions to the algebra of the next topic for confirmation.

Flashcards

Flashcard

lim (x → a) f(x) = L

f(x) approaches L as x approaches a

Approach, not arrival: x never equals a.

Test your understanding

Question 1 of 5

f is undefined at x = 4, but its graph approaches height 7 from both sides. What is the limit as x approaches 4?

Question 2 of 5

A table gives f(2.9) = 4.8, f(2.99) = 4.98, f(2.999) = 4.998 and f(3.1) = 5.2, f(3.01) = 5.02, f(3.001) = 5.002. What is the best estimate of the limit as x approaches 3?

Question 3 of 5

What happens to f(x) = (x − 2)/|x − 2| as x approaches 2?

Question 4 of 5

Your table shows f(x) settling on 2.5000 at x = 0.9, 0.99, 0.999 and at x = 1.1, 1.01, 1.001. What can you conclude?

Question 5 of 5

What happens to sin(1/x) as x approaches 0?

Standards: 1

← Previous topicNext topic →