Skip to content
Topos Uranos

Areas and the average value of a function

Summary

This lesson opens the applications of the integral with the magnitude for which it was defined: area. The region between two graphs is enclosed between rectangles contained in it and rectangles that cover it, like the region under a graph; the measures of those unions are differences of lower and upper sums, and the requirements for area from the lesson on the area problem leave a single possible value, the integral of the difference, which is adopted as the definition. If the graphs cross, the area is the integral of the absolute value of the difference, and it is computed by splitting the interval at the crossing points; if the region is better described with the abscissas as functions of the ordinates, the same argument, with the axes interchanged, gives the integral with respect to yy. A region swept by a polar curve is enclosed between inscribed and circumscribed circular sectors, whose areas are those from the lesson on angles and their measure, and its area is half the integral of the square of the radius. Finally, the average value of an integrable function, the quotient of its integral by the length of the interval, is the limit of the arithmetic means of its values on uniform partitions, by Darboux’s theorem, and a continuous function attains it at some interior point.

Learning objectives

  1. Prove, from the requirements for area, that the area between two graphs is the integral of the difference, and compute it when the graphs cross or when it is convenient to describe the region with xx as a function of yy.
  2. Prove the formula for area in polar coordinates with inscribed and circumscribed sectors, and apply it to regions bounded by polar curves.
  3. Define the average value of an integrable function, and prove that it is the limit of the averages of its values and that a continuous function attains it.
  4. Distinguish in each measure what is a definition and what is a theorem, and recognize which hypotheses cannot be omitted.

The area between two graphs

The lesson on the Riemann integral defined the area of the region under the graph of a nonnegative integrable function after proving that it is the only value compatible with the requirements for area from the lesson on the area problem: the rectangle requirement, monotonicity and additivity. This unit always uses the same method to measure: settle what reasonable requirements the magnitude has, enclose the object between simple pieces whose measure is known, recognize in those measures lower and upper sums of an integral, and conclude with the integrability criterion or with Darboux’s theorem, from the lesson on the Riemann integral. The region between two graphs is enclosed between rectangles, and no new requirement is needed.

TheoremThe area between two graphs is the integral of the difference

Let ff and gg be integrable functions on [a,b]\icc{a}{b}, with a<ba < b and g(x)≤f(x)g(x) \leq f(x) for every xx in [a,b]\icc{a}{b}, and let DD be the region between the graphs of gg and of ff, the set of the points (x,y)(x, y) with a≤x≤ba \leq x \leq b and g(x)≤y≤f(x)g(x) \leq y \leq f(x). If DD has an area AA that meets the requirements for area, then:

A=∫ab(f(x)−g(x)) dxA = \int_a^b \left(f(x) - g(x)\right) \, \mathrm{d}x

On each subinterval of a partition, a rectangle between the largest possible value of gg and the smallest of ff lies inside the region, and another between the smallest of gg and the largest of ff covers it. The figure shows the region with the rectangles contained in it for four subintervals; then the covering ones, added; then both for eight subintervals; and for sixteen, with the strips that separate them already narrow.

Proof

  1. A≥∑Mk′<mk(mk−Mk′) Δxk≥∑k=1n(mk−Mk′) Δxk=L(f,P)−U(g,P)A \geq \sum_{M'_k < m_k} (m_k - M'_k) \, \Delta x_k \geq \sum_{k=1}^{n} (m_k - M'_k) \, \Delta x_k = \resaltar{L(f, P) - U(g, P)}

    Let P={x0,…,xn}P = \{x_0, \ldots, x_n\} be a partition of [a,b]\icc{a}{b}. Integrable functions are bounded, and by the completeness axiom, from the lesson on number systems, on each subinterval there exist the infimum mkm_k and the supremum MkM_k of ff, and the infimum mk′m'_k and the supremum Mk′M'_k of gg. If Mk′<mkM'_k < m_k, the rectangle with base [xk−1,xk]\icc{x_{k - 1}}{x_k} lying between the heights Mk′M'_k and mkm_k is contained in DD: for xx in the subinterval and yy between those heights, g(x)≤Mk′≤y≤mk≤f(x)g(x) \leq M'_k \leq y \leq m_k \leq f(x). Those rectangles only touch along vertical sides; by the rectangle requirement and additivity, extended by induction to several pieces, their union has as its area the sum of their areas, and by monotonicity that sum does not exceed AA. The terms with Mk′≥mkM'_k \geq m_k, which have no rectangle, are not positive; if no subinterval has a rectangle, the bound holds because A≥0A \geq 0.

  2. L(f,P)−U(g,P)≤A≤U(f,P)−L(g,P)\dato{enc}{L(f, P) - U(g, P) \leq A \leq \resaltar{U(f, P) - L(g, P)}}

    For xx in subinterval kk, mk′≤g(x)≤f(x)≤Mkm'_k \leq g(x) \leq f(x) \leq M_k: the points of DD over that subinterval lie in the rectangle with base [xk−1,xk]\icc{x_{k - 1}}{x_k} lying between the heights mk′m'_k and MkM_k. The region DD is contained in the union of those rectangles, which only touch along vertical sides, and the same requirements give the upper bound.

  3. (U(f,P)−L(g,P))−(L(f,P)−U(g,P))=(U(f,P)−L(f,P))+(U(g,P)−L(g,P))<ε\left(U(f, P) - L(g, P)\right) - \left(L(f, P) - U(g, P)\right) = \left(U(f, P) - L(f, P)\right) + \left(U(g, P) - L(g, P)\right) < \resaltar{\varepsilon}

    Let ε>0\varepsilon > 0. Riemann’s integrability criterion, from the lesson on the Riemann integral, gives a partition P1P_1 with U(f,P1)−L(f,P1)<ε2U(f, P_1) - L(f, P_1) < \frac{\varepsilon}{2} and another P2P_2 with U(g,P2)−L(g,P2)<ε2U(g, P_2) - L(g, P_2) < \frac{\varepsilon}{2}, which depend on ε\varepsilon; by the refinement lemma, from the same lesson, their union PP satisfies both inequalities. By the proposition on the order of the lower and upper integrals, L(f,P)≤∫abf≤U(f,P)L(f, P) \leq \int_a^b f \leq U(f, P), and likewise with gg; on subtracting, and by the linearity of the integral, from the lesson on the properties of the definite integral, ∫ab(f−g)\int_a^b (f - g) lies in the same interval as AA, whose length is:

  4. (∀ε>0)(∣A−∫ab(f−g)∣<ε)→A=∫ab(f(x)−g(x)) dx(\forall \varepsilon > 0)\left(\abs{A - \int_a^b (f - g)} < \varepsilon\right) \rightarrow \resaltar{A = \int_a^b \left(f(x) - g(x)\right) \, \mathrm{d}x}

    Two numbers in the same interval are at most its length apart, and ε\varepsilon is arbitrary. If the distance between AA and the integral were positive, taking it as ε\varepsilon it would turn out smaller than itself; therefore, it is 00.

DefinitionArea between two graphs

Let ff and gg be integrable functions on [a,b]\icc{a}{b}, with a<ba < b and g(x)≤f(x)g(x) \leq f(x) for every xx in [a,b]\icc{a}{b}. The area of the region between the graphs of gg and of ff is ∫ab(f−g)\int_a^b (f - g).

With g=0g = 0 this is the definition of the area under a graph, from the lesson on the Riemann integral. The difference f−gf - g is not negative and is integrable, by linearity, so ∫ab(f−g)\int_a^b (f - g) is also the area under the graph of f−gf - g: each vertical line segment of the region, from g(x)g(x) to f(x)f(x), is the vertical translate of the line segment from 00 to f(x)−g(x)f(x) - g(x). Likewise, if cc is a constant with g(x)+c≥0g(x) + c \geq 0 on [a,b]\icc{a}{b}, which exists because gg is bounded, linearity gives ∫ab(f−g)=∫ab(f+c)−∫ab(g+c)\int_a^b (f - g) = \int_a^b (f + c) - \int_a^b (g + c): translated upward, the region is what remains of the region under f+cf + c after removing the region under g+cg + c. The requirements for area do not include invariance under translations, and that is why the theorem does not use it: it encloses the region directly between rectangles.

If the graphs cross, neither of the two functions stays always on top, and the region between them is bounded, at each abscissa, by the smaller and the larger of the two values.

CorollaryArea between two graphs that cross

Let ff and gg be integrable functions on [a,b]\icc{a}{b}, with a<ba < b. The area of the region of the points (x,y)(x, y) with a≤x≤ba \leq x \leq b and yy between g(x)g(x) and f(x)f(x) is ∫ab∣f−g∣\int_a^b \abs{f - g}. If moreover mm is a natural number, a=c0<c1<⋯<cm=ba = c_0 < c_1 < \cdots < c_m = b, and on each interval [cj−1,cj]\icc{c_{j - 1}}{c_j} the difference f−gf - g does not change sign, then:

∫ab∣f(x)−g(x)∣ dx=∑j=1m∣∫cj−1cj(f(x)−g(x)) dx∣\int_a^b \abs{f(x) - g(x)} \, \mathrm{d}x = \sum_{j=1}^{m} \abs{\int_{c_{j - 1}}^{c_j} \left(f(x) - g(x)\right) \, \mathrm{d}x}

The absolute values go outside each partial integral, not outside the sum: where gg is on top, the integral of f−gf - g is negative, and added without more ado it would subtract area instead of adding it.

Proof

  1. max⁡(f,g)−min⁡(f,g)=∣f−g∣→∫ab(max⁡(f,g)−min⁡(f,g))=∫ab∣f−g∣\max(f, g) - \min(f, g) = \abs{f - g} \rightarrow \resaltar{\int_a^b \left(\max(f, g) - \min(f, g)\right) = \int_a^b \abs{f - g}}

    By the corollary on the maximum and minimum of two integrable functions, from the lesson on the properties of the definite integral, max⁡(f,g)\max(f, g) and min⁡(f,g)\min(f, g) are integrable, and the region is the one between their graphs. For each xx, if f(x)≥g(x)f(x) \geq g(x) the difference between the larger and the smaller is f(x)−g(x)f(x) - g(x), and otherwise g(x)−f(x)g(x) - f(x); in both cases it is ∣f(x)−g(x)∣\abs{f(x) - g(x)}, and the definition of the area between two graphs gives the area.

  2. ∫cj−1cj∣f−g∣=∣∫cj−1cj(f−g)∣→∫ab∣f−g∣=∑j=1m∣∫cj−1cj(f−g)∣\int_{c_{j - 1}}^{c_j} \abs{f - g} = \abs{\int_{c_{j - 1}}^{c_j} (f - g)} \rightarrow \resaltar{\int_a^b \abs{f - g} = \sum_{j=1}^{m} \abs{\int_{c_{j - 1}}^{c_j} (f - g)}}

    By the corollary on integrability on the subintervals and additivity with respect to the interval, from the same lesson, applied m−1m - 1 times, the integral is the sum of the integrals over the intervals [cj−1,cj]\icc{c_{j - 1}}{c_j}. On one where f−g≥0f - g \geq 0, ∣f−g∣=f−g\abs{f - g} = f - g, and its integral is not negative, by monotonicity; where f−g≤0f - g \leq 0, ∣f−g∣=−(f−g)\abs{f - g} = -(f - g), and the integral of f−gf - g is not positive. In both cases, the integral of ∣f−g∣\abs{f - g} is the absolute value of that of f−gf - g.

The solved problem “A difference with zero integral” shows a region whose area is 88 even though the integral of the difference over the whole interval is 00. Some regions are described more naturally by horizontal strips: bounded on the left and on the right by graphs of functions of yy. The requirements for area treat both axes alike, because the rectangle requirement holds for every rectangle with sides parallel to them, and the theorem carries over without changes.

PropositionRegions described with xx as a function of yy

Let uu and vv be integrable functions on [c,d]\icc{c}{d}, with c<dc < d and u(y)≤v(y)u(y) \leq v(y) for every yy in [c,d]\icc{c}{d}, and let DD be the set of the points (x,y)(x, y) with c≤y≤dc \leq y \leq d and u(y)≤x≤v(y)u(y) \leq x \leq v(y). If DD has an area AA that meets the requirements for area, then:

A=∫cd(v(y)−u(y)) dyA = \int_c^d \left(v(y) - u(y)\right) \, \mathrm{d}y

The rectangles are now stacked over the subintervals of an interval of the axis of ordinates.

Proof

  1. L(v,Q)−U(u,Q)≤A≤U(v,Q)−L(u,Q)L(v, Q) - U(u, Q) \leq A \leq \resaltar{U(v, Q) - L(u, Q)}

    Let Q={y0,…,yn}Q = \{y_0, \ldots, y_n\} be a partition of [c,d]\icc{c}{d}, and on each subinterval let mkm_k and MkM_k be the infimum and the supremum of vv, and mk′m'_k and Mk′M'_k those of uu. If Mk′<mkM'_k < m_k, the rectangle with height [yk−1,yk]\icc{y_{k - 1}}{y_k} lying between the abscissas Mk′M'_k and mkm_k is contained in DD, and the rectangles lying between the abscissas mk′m'_k and MkM_k cover DD; those of each family only touch along horizontal sides. The first two moves of the proof of the theorem, with the roles of the axes interchanged, give the bounds.

  2. A=∫cd(v(y)−u(y)) dy\resaltar{A = \int_c^d \left(v(y) - u(y)\right) \, \mathrm{d}y}

    The last two moves of that proof apply without changes, with vv and uu in place of ff and gg, and with partitions of [c,d]\icc{c}{d}.

Under the same hypotheses, the area of such a region is defined as that integral. A region may admit both descriptions; if it has an area that meets the requirements, the theorem and the proposition give the same number. The solved problem “Better by the ordinates” checks this in one case, and the proposed problem “The area under a function and that under its inverse” proves it for the regions bounded by the graph of a continuous, strictly increasing function. That the two definitions agree for every region that admits both descriptions is a fact this course takes for granted: it is proved in later courses, with the integral in several variables. Unbounded regions, such as the one under the graph of a continuous positive function on [a,∞)\ico{a}{\infty}, fall outside these definitions, because the Riemann integral only applies to closed and bounded intervals; the lesson on improper integrals will measure some of them as limits of areas of bounded regions.