Calculus › Section D · Applications of derivatives › Topic D.1

Extreme values on an interval

Finding the biggest or smallest value of a function is the oldest applied question in calculus. This topic pins down where extremes can hide, and gives a four-step method that never misses them on a closed interval.

Why we hunt extremes

Businesses want maximum profit at minimum cost, engineers want the strongest part from the least material, drivers and data packets want the shortest route. Every one of these is asking for an extreme value of some function. Before calculus the search was guesswork: try lots of inputs and hope the best one was among them. Derivatives turn the hunt into a short list of suspects. In this topic we learn exactly what kinds of suspects exist, and a method that finds the absolute largest and smallest values of a continuous function on a closed interval without fail.

Absolute versus local

Two flavours of extreme deserve separate names. An absolute maximum of \(f\) on an interval is a value \(f(c)\) that is at least as large as every other value of \(f\) there; an absolute minimum is at least as small as every other value. A local maximum is a hilltop: \(f(c)\) is at least as large as every value on some open interval around \(c\), whether or not it wins globally. A function can enjoy many local extrema, one per bump, but at most one absolute maximum value, though that value may be attained at several points. Because a local extremum needs an open interval around it, an endpoint is never a local extremum, but it can still be an absolute one.

The guarantee: the Extreme Value Theorem

Here is the anchor of the topic. If \(f\) is continuous on a closed interval \([a, b]\), then \(f\) has both an absolute maximum and an absolute minimum on \([a, b]\). Both hypotheses earn their place. Drop 'closed': \(\frac{1}{x}\) on \((0, 1]\) grows without bound near the missing endpoint \(0\), so no maximum exists. Drop 'continuous': \(f(x) = x\) on \([0, 1)\) with \(f(1) = 0\) creeps towards height 1 but never reaches it, so it has no maximum. When both conditions hold, the graph is an unbroken stretch over a finite strip, and intuition is right: somewhere it must top out and somewhere it must bottom out.

Critical numbers and Fermat's theorem

Where can a local extremum live? At a smooth peak or valley the tangent is horizontal, so the derivative is \(0\). A corner or cusp can also be an extreme, and there the derivative fails to exist. Both cases are swept into one definition: a critical number is a domain point \(c\) where \(f'(c) = 0\) or \(f'(c)\) does not exist. Fermat's theorem states the smooth half precisely: if \(f\) has a local extremum at \(c\) and is differentiable there, then \(f'(c) = 0\). So the search space is small: interior extrema can only occur at critical numbers.

Warning: critical does not mean extreme

Fermat's theorem is one-directional, and the counterexample to memorise is \(f(x) = x^3\) at \(x = 0\). There \(f'(0) = 0\), the tangent is perfectly level, yet the curve passes straight through, rising before and after: no peak, no valley, just a pause. A level tangent or a corner tells you the derivative did something interesting; whether the function turned around is a separate question, answered by the tests of topics D.3 and D.4. For now the discipline is: collect critical numbers as candidates only, never as verdicts.

The closed-interval method

To find the absolute extrema of a continuous \(f\) on \([a, b]\): first, differentiate. Second, find every critical number inside the interval. Third, evaluate \(f\) at each critical number and at both endpoints. Fourth, compare the output values: the largest is the absolute maximum, and the smallest is the absolute minimum. Nothing else can win: an interior winner would be a local extremum, hence a critical number, and the endpoints are already on the list. Four steps, complete coverage. Most of the marks lost on this method are lost by skipping the endpoints.

Worked example: hunting on a closed interval

Find the absolute extrema of \(f(x) = x^3 - 12x + 5\) on \([0, 3]\). Differentiate: \(f'(x) = 3x^2 - 12 = 3(x^2 - 4)\), zero at \(x = \pm 2\). Only \(x = 2\) lies in the interval, so the candidate list is \(0, 2, 3\). Evaluate: \(f(0) = 5\), \(f(2) = 8 - 24 + 5 = -11\), \(f(3) = 27 - 36 + 5 = -4\). Compare: the absolute maximum is \(5\), attained at the endpoint \(x = 0\); the absolute minimum is \(-11\), at the critical number \(x = 2\). Note how the endpoint won the maximum; that happens constantly and is exactly why the method insists on them.

Worked example: a corner wins a minimum

Find the absolute extrema of \(f(x) = |x - 1|\) on \([0, 3]\). The V-shape points at \(x = 1\), where the derivative does not exist, so \(1\) is a critical number. Endpoints: \(f(0) = 1\) and \(f(3) = 2\). At the corner: \(f(1) = 0\). Comparing the three values, the absolute minimum is \(0\) at \(x = 1\), the sharp point exactly where no derivative exists, and the absolute maximum is \(2\) at the endpoint \(x = 3\). Two lessons in one example: non-differentiable points must be included as candidates, and again an endpoint takes the maximum.

What can go wrong without the hypotheses

The Extreme Value Theorem is a guarantee, and outside its hypotheses that guarantee lapses. On an open interval a function can approach a height forever without reaching it: \(f(x) = x^2\) on \((0, 1)\) takes values that get as close as you like to 0 and to 1 but never reach either, so it has no absolute maximum or minimum. A discontinuous function on a closed interval can jump past its own best value. This matters in practice because real models often live on open domains, such as prices \(p > 0\) or times \(t > 0\). There, extrema need a separate argument, for example from the shape of the graph or from a single critical number, a theme Section D returns to.

Common mistakes to avoid

The classics: forgetting to evaluate the endpoints, the single most common way to lose the actual answer; reporting the location \(c\) when the question asks for the maximum value \(f(c)\), or the reverse, so read the question twice; assuming every point with \(f'(c) = 0\) is a maximum or minimum, with \(x^3\) at \(0\) as the standing counterexample; ignoring points where the derivative fails to exist, which are critical numbers too, provided they lie in the domain; and applying the method on an open interval, where no theorem has promised you anything. Candidates first, verdicts after comparison.

Key takeaways

Extremes come in two flavours: absolute over the whole domain and local over a neighbourhood. The Extreme Value Theorem guarantees that a continuous function on a closed interval has both an absolute maximum and an absolute minimum. Interior extremes can only occur at critical numbers, where \(f'\) is zero or undefined, but not every critical number is an extreme. For a continuous function on \([a, b]\), the closed-interval method (differentiate, collect critical numbers, evaluate those and the endpoints, compare) never misses. Keep values and locations distinct in your answer, and outside closed, continuous territory, argue from the shape of the function instead.

Flashcards

Flashcard

Absolute maximum

The largest value on the whole domain

One value, possibly at several points.

Test your understanding

Question 1 of 5

What are the critical numbers of f(x) = x³ − 12x + 5?

Question 2 of 5

What is the absolute maximum value of f(x) = x³ − 12x + 5 on [0, 3]?

Question 3 of 5

Which conditions guarantee a function has an absolute maximum and minimum on an interval?

Question 4 of 5

The function f(x) = x³ satisfies f′(0) = 0. What is true at x = 0?

Question 5 of 5

What is the absolute minimum of f(x) = |x − 1| on the interval [0, 3]?

Standards: 5

← Previous topicNext topic →