Calculus › Section F · Logarithmic, exponential, and other transcendental functions › Topic F.6

Indeterminate forms and L'Hôpital's rule

Some limits resist every algebraic trick of Section B. L'Hôpital's rule resolves the stubborn 0/0 and ∞/∞ forms by comparing how fast top and bottom move, using derivatives.

The last holdouts

By now the limit toolbox is deep: substitution, factoring, conjugates, squeezing. Yet forms like \(\lim_{x \to 0}\frac{e^x - 1}{x}\) survive them all. The diagnosis from B.3 stands: \(0/0\) records a tug-of-war between two quantities racing to zero, and its outcome depends on how fast each is moving at the crucial point. Rates are derivatives. That observation is L'Hôpital's rule in one breath: the ratio of the functions approaches the ratio of their rates. The rule below is the precise statement, with the conditions that keep it honest.

The rule

Suppose \(f\) and \(g\) are differentiable on an open interval around \(a\), except possibly at \(a\) itself, with \(g'(x) \ne 0\) there, and that \(\frac{f}{g}\) approaches an indeterminate form \(0/0\) or \(\pm\infty/\pm\infty\) as \(x \to a\). Then, provided the new limit exists or is infinite,

\lim_{x \to a} \frac{f(x)}{g(x)} = \lim_{x \to a} \frac{f'(x)}{g'(x)}

The same result holds for one-sided limits and for \(x \to \pm\infty\). The machinery replaces a stalemate with a comparison of speeds, and if the new form is still indeterminate, apply the rule again, checking the form each time.

It is not the quotient rule

One line of insurance against the classic blunder: L'Hôpital differentiates the numerator and denominator separately, each as its own function. There is no \(g^2\) underneath and no subtraction in the top; the quotient rule has no business here. The rule's plausibility argument is one line of local linearisation: near \(a\), \(f(x) \approx f'(a)(x - a)\) when \(f(a) = 0\), likewise \(g\), and the factors \((x - a)\) cancel, leaving the ratio of slopes. That is the geometry of the rule, though the complete proof lives a level above this course.

Worked example: e^x minus one over x

Compute \(\lim_{x \to 0} \frac{e^x - 1}{x}\). Substitution lands on \(0/0\), and factoring gets nowhere since the exponential will not factor. Check the form, then apply the rule: differentiate top and bottom separately to get \(\frac{e^x}{1}\), which at \(x = 0\) is \(1\). So the limit is \(1\). A numeric sense-check confirms: \(e^{0.001} \approx 1.0010005\), so the ratio at \(0.001\) is already \(1.0005\). Note the bonus content: this limit is exactly the derivative of \(e^x\) at \(0\), and its value \(1\) is the slope at \((0, 1)\) from F.4.

Worked example: trig limits, once and twice

The rule also recovers trig limits. For \(\lim_{x \to 0} \frac{\sin 3x}{x}\), the form is \(0/0\); differentiating gives \(\frac{3\cos 3x}{1}\), which tends to \(3\). Sometimes one application is not enough. For \(\lim_{x \to 0} \frac{1 - \cos 2x}{x^2}\), the form is \(0/0\), and the rule gives \(\frac{2\sin 2x}{2x}\), still \(0/0\). Apply it again: \(\frac{4\cos 2x}{2} \to \frac{4}{2} = 2\). A numeric check at \(x = 0.01\): \(\frac{1 - \cos 0.02}{0.0001} \approx 1.9999\). These agree with the special limit \(\frac{\sin x}{x} \to 1\) from B.3, which the rule reproduces in one step.

Worked example: powers lose to exponentials

Compute \(\lim_{x \to \infty} \frac{x^2}{e^x}\). The form is \(\infty/\infty\), so the rule applies: \(\frac{2x}{e^x}\). Still \(\infty/\infty\): apply it again, to get \(\frac{2}{e^x} \to 0\). The limit is \(0\). The slogan this earns is worth remembering for modelling: exponential growth beats polynomial growth, every time, whatever the powers. One discipline to keep: after each application, re-check that the new form is genuinely indeterminate before applying the rule again, because applying it to a decidable form produces confident, wrong answers.

Worked example: a 0·∞ rewrite

Compute \(\lim_{x \to 0^+} x \ln x\). Direct reading gives \(0 \cdot (-\infty)\), which is indeterminate since the two tendencies fight. L'Hôpital needs a quotient, so rewrite \(x\ln x = \frac{\ln x}{1/x}\), now an \(\infty/\infty\) form. Differentiate top and bottom: the top gives \(\frac{1}{x}\), the bottom \(-\frac{1}{x^2}\), and the ratio simplifies to \(-x\), which tends to \(0\). So \(x\ln x \to 0\): the log's plunge loses to the linear squeeze. The lesson generalises: products \(0 \cdot \infty\), differences \(\infty - \infty\) (combine them into one fraction), and power forms must all first be rewritten into a genuine \(0/0\) or \(\infty/\infty\) quotient.

Power forms: logarithm first

Forms like \(1^{\infty}\), \(0^0\) and \(\infty^0\) yield to one manoeuvre: if the limit sought is \(L\), take the logarithm of the expression, evaluate that limit, and exponentiate back. The showcase example is \(\lim_{x \to \infty}\left(1 + \frac{1}{x}\right)^x\). Its log is \(x\ln\left(1 + \frac{1}{x}\right)\), which rewrites as \(\frac{\ln(1 + 1/x)}{1/x}\), a \(0/0\) form. Differentiating gives \(\frac{1}{1 + 1/x}\), tending to \(1\), so the log of the answer is \(1\) and the answer is \(e^1 = e\). Observe: this is the limit underneath continuous compounding from F.5, now computed by pure calculus.

The cautionary counterexample

The rule's first word is a form-check, and ignoring it is instructive. Consider \(\frac{x + 1}{x}\) as \(x \to 0\): top to \(1\), bottom to \(0\), not indeterminate, this is B.5's unbounded blow-up. A student applying L'Hôpital automatically gets \(\frac{1}{1} = 1\), a firm, definite, wrong answer. Whenever the form is not indeterminate, the rule is simply void. Similarly, if the new limit fails to exist, the rule says nothing about the original. The form check is not pedantry but the licence of the method, and writing "form is \(0/0\)" on your first line is good practice as well as good marks.

Common mistakes to avoid

The recurring failures: applying the rule to a form that is not indeterminate, as the counterexample showed; differentiating the quotient as a quotient instead of top and bottom separately; forgetting to re-check the form between repeated applications; applying L'Hôpital to \(0 \cdot \infty\) or \(1^{\infty}\) without first rewriting as a quotient or taking the logarithm; and inferring anything when the derived limit fails to exist.

Key takeaways

For \(0/0\) or \(\infty/\infty\), with differentiability near the point and \(g' \ne 0\) there, L'Hôpital's rule equates \(\lim \frac{f}{g}\) with \(\lim\frac{f'}{g'}\), provided the latter exists. It is two separate derivatives, never the quotient rule. Rewrite other indeterminate forms, products to quotients, power forms via the logarithm, and re-check the form before each repeat. The trophies: \(\frac{e^x - 1}{x} \to 1\), \(x^2/e^x \to 0\), \(x\ln x \to 0\), and \((1 + 1/x)^x \to e\). Next, F.7 brings the inverse trig functions, whose derivatives turn out to be algebraic.

Flashcards

Flashcard

Indeterminate forms

0/0, ∞/∞, 0·∞, ∞−∞, 1^∞, 0⁰, ∞⁰

The form alone never decides the value.

Test your understanding

Question 1 of 5

What is the limit of (e^x − 1)/x as x approaches 0?

Question 2 of 5

What is the limit of x²/e^x as x approaches infinity?

Question 3 of 5

What is the limit of x ln x as x approaches 0 from the right?

Question 4 of 5

Which limit is a legitimate target for L'Hôpital's rule?

Question 5 of 5

What is the limit of (1 + 1/x)^x as x approaches infinity?

Standards: 4

← Previous topicNext topic →