Calculus › Section B · Limits › Topic B.3
Evaluating limits with algebra
Graphs and tables suggest what a limit might be, but algebra proves it. A short toolkit, built on a handful of limit laws, turns most limits into exact calculations.
From guessing to certainty
In topic B.2 you estimated limits from graphs and tables, and you saw that such estimates are evidence, not proof: a table never checks every value near a point. This topic replaces guessing with machinery. Recall what \(\lim_{x \to a} f(x) = L\) says: the outputs can be made as close to \(L\) as we like by taking \(x\) close enough to \(a\), and the value \(f(a)\) itself is irrelevant, even undefined. Because the limit ignores the point itself, we are allowed to simplify a function, cancel factors that vanish only at \(a\), and reshape expressions, as long as the new version agrees with the old one everywhere else near \(a\). That single observation powers almost everything in this topic.
The limit laws
Suppose \(\lim_{x \to a} f(x) = L\) and \(\lim_{x \to a} g(x) = M\). Then the limit of a sum is \(L + M\), of a difference is \(L - M\), of a constant multiple is \(cL\), and of a product is \(LM\):
The quotient law gives \(\frac{L}{M}\), but only when \(M \ne 0\). Powers and roots behave as you would expect: the limit of \([f(x)]^n\) is \(L^n\), and the limit of \(\sqrt[n]{f(x)}\) is \(\sqrt[n]{L}\), provided \(L > 0\) when \(n\) is even. These laws let you break a complicated limit into small, obvious pieces, work on each piece, and reassemble the answer. The hidden condition in every law is that the individual limits must exist.
Direct substitution first
The limit laws lead to a very practical theorem. Polynomials behave perfectly everywhere: \(\lim_{x \to a} p(x) = p(a)\). So \(\lim_{x \to 3}(2x^3 - x + 1) = 2(27) - 3 + 1 = 52\). Rational functions behave wherever the denominator is not zero: \(\lim_{x \to 1}\frac{x+1}{x-2} = \frac{2}{-1} = -2\). The same holds for the other well-behaved families, such as trig functions, exponentials and logarithms, at interior points of their domains, and for compositions built from them. The strategy begins here. Always try direct substitution first; it is free, and far more often than you might expect, it simply works.
Reading the result of substitution
Substitution has three possible outcomes, and each tells you what to do next. A finite number, like \(\frac{4}{5}\) or \(-7\): you are done; that is the limit. The form \(\frac{0}{0}\): inconclusive. It is called an indeterminate form, and it means the top and bottom both shrink to nothing, so their tug-of-war is not decided yet. Do more algebra. A nonzero number over \(0\), like \(\frac{5}{0}\): not indeterminate at all. The outputs are blowing up, and the limit fails to exist in a very particular way involving vertical asymptotes, which topic B.5 treats in full. Read the outcome before you react to it.
Worked example: factor and cancel
Find \(\lim_{x \to 2}\frac{x^2 - 5x + 6}{x - 2}\). Substitution gives \(\frac{4 - 10 + 6}{0} = \frac{0}{0}\), indeterminate, so we work. The numerator factors: \(x^2 - 5x + 6 = (x - 2)(x - 3)\). For every \(x\) except \(2\) itself,
and since the limit never visits \(x = 2\), the two expressions have the same limit there. So \(\lim_{x \to 2}(x - 3) = 2 - 3 = -1\). Check numerically: at \(x = 1.999\) the original ratio is \(-1.001\), and at \(x = 2.001\) it is \(-0.999\), squeezing in on \(-1\) from both sides.
Why cancelling is honest
Cancelling feels like cheating, because \(\frac{x - 2}{x - 2}\) is genuinely undefined at \(x = 2\), not equal to \(1\). Here is the honest version: if two functions agree at every point near \(a\), except possibly at \(a\) itself, then they have the same limit as \(x \to a\). The cancelled function \(x - 3\) and the original ratio disagree only at the single point \(x = 2\), which the limit never examines. Graphically, the ratio is the line \(y = x - 3\) with one point removed: a hole. The limit asks where the line is heading, hole or no hole. This is why a cancelled factor leaves a removable gap in the graph rather than changing the trend.
Worked example: multiply by the conjugate
Find \(\lim_{x \to 0}\frac{\sqrt{x + 4} - 2}{x}\). Substitution gives \(\frac{2 - 2}{0} = \frac{0}{0}\), and nothing obvious factors. Roots call for the conjugate: multiply top and bottom by \(\sqrt{x + 4} + 2\):
The difference-of-squares identity did the work, turning the top into \(x\) and cancelling the troublemaker. Now substitute: \(\frac{1}{\sqrt{4} + 2} = \frac{1}{4}\). A numeric check agrees: at \(x = 0.001\), the original ratio is about \(0.24998\).
The squeeze theorem
Some functions misbehave near a point but are pinned between two better-behaved ones. The squeeze theorem: if \(g(x) \le f(x) \le h(x)\) for all \(x\) near \(a\), except possibly at \(a\), and both \(g\) and \(h\) approach the same limit \(L\), then \(f\) is forced to \(L\) as well. The classic target is \(x^2\cos\frac{1}{x}\) as \(x \to 0\). The cosine oscillates wildly, so substitution and factoring get nowhere. But cosine never leaves \([-1, 1]\), so \(-x^2 \le x^2\cos\frac{1}{x} \le x^2\). Both walls close to \(0\), and the function, trapped between them, approaches \(0\) too. Bounding converts chaos into certainty.
Worked example: a trigonometric limit
The squeeze theorem also proves the two special trig limits:
Both require \(x\) in radians. You will not reprove them, but you will constantly use them. Find \(\lim_{x \to 0}\frac{\sin 3x}{x}\). Substitution gives \(\frac{0}{0}\), and the 3 inside the sine spoils a direct match. Fix the ratio: multiply top and bottom by 3 to get \(\frac{\sin 3x}{x} = 3\cdot\frac{\sin 3x}{3x}\). The second factor is the known limit with \(u = 3x\), which approaches \(1\). So the limit is \(3 \cdot 1 = 3\). In general, \(\frac{\sin kx}{x} \to k\).
Common mistakes to avoid
The worst reflex is treating \(\frac{0}{0}\) as an answer: it is neither \(0\) nor evidence that a limit fails to exist, it is a request for more work. Second, cancelling terms instead of factors: in \(\frac{x^2 + 1}{x^2}\) the \(x^2\) terms do not cancel, because only whole factors cancel. Third, splitting a quotient whose bottom limit is \(0\): the quotient law is silent there, so any conclusion it seems to give is void. Fourth, sign slips with the conjugate: \((\sqrt{u} - v)(\sqrt{u} + v)\) equals \(u - v^2\), not \(u + v^2\). Last, forgetting radians: the special trig limits are false in degrees.
Key takeaways
Evaluate limits in this order. Try direct substitution; for polynomials, trig functions and rational functions away from zero denominators, the value is the limit. If substitution gives \(\frac{0}{0}\), do algebra: factor and cancel, or multiply by the conjugate for roots. If an oscillating factor is trapped between bounds, squeeze it. Know the two special trig limits, \(\frac{\sin x}{x} \to 1\) and \(\frac{1 - \cos x}{x} \to 0\), in radians, and the scaling trick \(\frac{\sin kx}{x} \to k\). A nonzero number over \(0\) is not indeterminate: it signals unbounded behaviour, the subject of topic B.5. The limit laws justify every step, provided each piece's limit exists.
Flashcards
Flashcard
Limits split over sums, differences, multiples and products
Test your understanding
Question 1 of 5
What is the limit of 2x³ − x + 1 as x approaches 3?
Question 2 of 5
Find the limit of (x² − 4)/(x − 2) as x approaches 2.
Question 3 of 5
Find the limit of (√(x + 9) − 3)/x as x approaches 0.
Question 4 of 5
What is the limit of x² cos(1/x) as x approaches 0?
Question 5 of 5
What is the limit of sin(3x)/x as x approaches 0 (x in radians)?
Standards: 1
