Skip to content
Topos Uranos

Applications of the integral: synthesis and challenges

Summary

This lesson closes the unit on the applications of the integral. It gathers into a single deductive chain what the four content lessons proved with one and the same method: laying down the reasonable requirements for a quantity, enclosing it between simple pieces whose measure is known, recognizing in those measures lower and upper sums of an integral, and passing to the limit. In this way the following are determined: the area between two graphs and that of a polar region, the average value of a function, the volume of a solid whose cross-sections have integrable area, with Cavalieri’s principle, the cone and the solids of revolution, and the length of a curve with a continuous derivative, from which arc length and curvature follow; numerical integration returns to the sums to approximate the integral itself, with error bounds proved without the Taylor polynomial. It explains in each case what is a definition, what is a theorem and what rests on what Algebra and Geometry admitted, how the unit rests on Calculus I, on Algebra and Geometry and on the preceding units, which of their promises it fulfills and what it leaves to improper integrals. It ends with an exit assessment, parallel to the entry assessment, and with ten challenges, harder than the problems of the lessons, each of which combines results from several of them.

Learning objectives

  1. Present in logical order the results of the unit, indicating on which results of Calculus I, of Algebra and Geometry and of the preceding units each one depends, and in which proofs the requirements, the cylinder condition and the iterated Cauchy theorem are reused.
  2. Distinguish, for each quantity, what is defined from what is proved, and recognize which results are consistency checks with what Algebra and Geometry admitted and which do not depend on it.
  3. Relate the unit to Calculus I, to Algebra and Geometry, to the unit on the definite integral and to the units on improper integrals and series.
  4. Solve nonroutine problems that combine several lessons: bounding a length without computing it, measuring a classical figure with integrals, deciding which numerical rule guarantees a given accuracy, and characterizing a curve by its length or by its curvature.

What the unit achieved

The unit on the definite integral began by observing that the area of a curved figure was not defined, laid down the requirements that any reasonable definition must satisfy (the area of a rectangle, monotonicity and additivity) and obtained the integral as the number that separates the lower sums from the upper sums. This unit repeated that gesture with other quantities. If a quantity satisfies its requirements and the simple pieces that enclose it from below and from above measure lower and upper sums of an integrable function, then the quantity lies between those sums for every partition; it follows, by the integrability criterion or by Darboux’s theorem, that the only possible value is the integral:

(∀P)(L(f,P)≤A≤U(f,P))→A=∫abf(x) dx(\forall P)\left(L(f, P) \leq A \leq U(f, P)\right) \rightarrow A = \int_a^b f(x) \, \mathrm{d}x

Now, that theorem is conditional: it says what value the quantity must have if it exists, not that it exists. Therefore, in each case the unit adopted the integral as the definition, and the conditional theorem was the reason for adopting it. Length was the exception: inscribed polygonal paths approximate it only from below, and the unit defined it directly as a supremum, so that the integral formula is there an unconditional theorem.

Areas and the average value

The lesson on areas and the average value of a function enclosed the region between the graphs of two integrable functions g≤fg \leq f between rectangles contained in it and rectangles that cover it, and proved that its area, if it satisfies the requirements, is ∫ab(f−g)\int_a^b (f - g), a value it adopted as the definition; if the graphs cross, the area is ∫ab∣f−g∣\int_a^b \abs{f - g}, and it is computed by splitting the interval at the crossing points. Regions described with xx as a function of yy are treated in the same way, stacking the rectangles on the axis of ordinates. A polar region under a continuous and nonnegative function rr is enclosed between circular sectors, and its area is 12∫αβr(θ)2 dθ\frac{1}{2} \int_{\alpha}^{\beta} r(\theta)^2 \, \mathrm{d}\theta; since the requirements do not give the area of a sector, the theorem demands it explicitly, with the value of the theorem on arc length and the area of a sector from Algebra and Geometry, deduced there from the admitted area of the disk. The average value of an integrable function, 1b−a∫abf\frac{1}{b - a} \int_a^b f, turned out to be the limit of the averages of its values over finer and finer partitions, and a continuous function attains it at an interior point.

Volumes

The lesson on volumes of solids laid down the requirements for volume (that of a cylinder, right or oblique, is the area of the base times the height; monotonicity and additivity) and proved that a solid whose cross-sections have integrable area A(x)A(x) and that satisfies the cylinder condition, that is, whose slices contain cylinders and are contained in cylinders with bases as close as desired to the cross-sections, can only have the volume ∫abA(x) dx\int_a^b A(x) \, \mathrm{d}x, which it adopted as the definition. When the cross-sections are nested the condition holds automatically, and Cavalieri’s principle is a corollary. With a lemma on the area of a scaled and translated figure, and with oblique cylinders, it obtained the volume of every cone whose base is a convex figure lying between two continuous graphs, 13A(h) h\frac{1}{3} A(h) \, h, and in particular that of the pyramid with a convex base. For solids of revolution it proved the formulas by disks and washers and by cylindrical shells, and that, for a continuous, nonnegative and strictly monotonic function, the two give the same number, with a summation by parts and without derivatives.

Lengths and curvature

The lesson on arc length and curvature defined the length of a plane parametric curve as the supremum of the lengths of the inscribed polygonal paths, which exists by the completeness axiom when those lengths are bounded; the triangle inequality shows that the polygonal path does not get shorter under refinement. The central theorem says that, if the coordinates have continuous derivatives, the curve is rectifiable and its length is ∫abx′(t)2+y′(t)2 dt\int_a^b \sqrt{x'(t)^2 + y'(t)^2} \, \mathrm{d}t: Lagrange’s mean value theorem gives on each subinterval a point for each coordinate, in general different ones, and the Heine-Cantor theorem bounds the difference with a Riemann sum of the speed. From it follow the length of a graph and that of a polar curve, the arc length function, whose derivative is the speed by the fundamental theorem, and the curvature as the rate at which the tangent turns per unit of length, with its formula for parametric curves, which for a graph coincides with that of Calculus I.

The error of the numerical rules

The lesson on numerical integration returned to the sums of the preceding unit to approximate the integral itself: the midpoint and trapezoidal rules integrate exactly, on each subinterval, a constant and a polynomial of degree at most one, and Simpson’s rule, on each pair of subintervals, one of degree at most two, all passing through points of the curve. Their errors were proved without the Taylor polynomial: the error on a subinterval, as a function of the half-width, is differentiated with the derivative of an integral with variable limits and compared with a power by means of a lemma that iterates Cauchy’s mean value theorem. Hence the bounds (b−a)h224B2\frac{(b - a) h^2}{24} B_2, (b−a)h212B2\frac{(b - a) h^2}{12} B_2 and (b−a)h4180B4\frac{(b - a) h^4}{180} B_4, the degree of exactness and the sign of the error, the number of subintervals that guarantees a given accuracy and the behavior of the error when the number of subintervals is doubled.

What is admitted, what is proved and what remains

The balance of the unit can be summarized as follows:

  • Admitted are the completeness axiom of Algebra and Geometry I, with the Archimedean property that is deduced from it there; the requirements for area, from the preceding unit, and the analogous ones for volume; the areas of polygons; that the different descriptions of a plane figure give it the same area, a fact proved in later courses; and, through Algebra and Geometry, the area of the disk, on which rest the area of the circular sector and, through the inequality between the sine, the arc and the tangent, the derivatives of the trigonometric functions. The results of Calculus I on continuity and the derivative are used, together with the theorems of Weierstrass, Lagrange and Cauchy and the intermediate value theorem, and those of the preceding units on the integral and the Heine-Cantor theorem.
  • Proved are the theorems that justify the definitions of the area between graphs, of the polar area and of the volume by cross-sections; the formulas by disks and by shells and their equivalence; the formula for length and that for curvature; and the error bounds of the three rules. Several promises are thus fulfilled: that of the lesson on the Riemann integral, which announced the area between two curves; that of the lesson on the properties of the definite integral, which announced the study of the average value; that of the lesson on the area problem, which announced that the sums of rectangles would be refined in order to approximate, and that of the lesson on the fundamental theorem of calculus, whose value 0.9225…0.9225\ldots has been computed with a bounded error; that of the Calculus I lesson on curvature and radius of curvature and that of the synthesis of the applications of derivatives, which announced arc length defined with the integral and curvature as the rate at which the tangent turns; and the formula for the pyramid that the lesson on the cross product and the scalar triple product admitted, proved for a convex base contained in a plane perpendicular to an axis, and for any position if one adds to the requirements that volume does not change when the solid is rotated.
  • Not proved independently, but checked to be consistent with what is admitted, are the area πa2\pi a^2 of the disk, the length 2πR2\pi R of the circle and the volumes of the circular cone and of the ball: the computations of the first two use the derivatives of the sine and the cosine, or the antiderivatives that come from them, which rest on the admitted area of the disk, and the two volumes take from the first the area of their cross-sections. The same holds for the expressions of π\pi as an integral in the lesson on arc length and in the lesson on numerical integration: they compute, with any desired accuracy, a number whose definition is still that of Algebra and Geometry.
  • Three instruments are reused instead of repeating arguments: bracketing between sums, which turns the requirements for a quantity into an integral; the cylinder condition, which gives all the volumes by cross-sections; and the iterated Cauchy theorem, which gives all the error bounds.
  • What needs another tool is left for later: unbounded regions and solids, in the unit on improper integrals; the area of a surface, volume defined with double and triple integrals, and space curves, in Calculus III. That the different definitions of area agree for every figure that admits several descriptions is admitted, as was said, and proved in later courses; that the definitions of volume by cross-sections in different directions agree is mentioned as another fact from those courses, on which nothing depends.

The following figure gathers the four segments into a single chain. Each node of the column is a result, numbered according to the logical order and joined by an arc to the earlier links that its proof uses; around it float the supports, that is, the principles and tools to which the chain resorts again and again, each one joined by a solid line to the links that rest on it and by a dotted one to the link where it is established, except those that come from Calculus I, from Algebra and Geometry or from the preceding units, which are drawn as rings. Clicking a support highlights all the links that use it, and each link refers to the lesson in which it is proved.

Supports of the chain

  • ACompleteness axiom. Axiom of Algebra and Geometry I: every nonempty set of real numbers that is bounded above has a supremum. There the Archimedean property is deduced from it; here it provides the infima and suprema on each subinterval and the length of a rectifiable curve. Admitted without proof in this unit.
  • BRequirements and bracketing between sums. The requirements for area, from the lesson on the area problem (the rectangle, monotonicity and additivity), and the analogous ones for volume, with the cylinder in place of the rectangle; and the bracketing: if a function is integrable, its integral is the only number lying between all its lower sums and all its upper sums. Admitted without proof in this unit.
  • CDefinite integral. Proved in the unit on the definite integral: Riemann’s integrability criterion, Darboux’s theorem and the sums over uniform partitions, the integrability of continuous functions, linearity, additivity with respect to the interval, monotonicity, the Cauchy-Schwarz inequality, the first mean value theorem for integrals, the fundamental theorem, Barrow’s rule, the derivative of an integral with variable limits, change of variable and integration by parts. Admitted without proof in this unit.
  • DSequences. From the unit on sequences: the Heine-Cantor theorem, according to which every function continuous on a closed and bounded interval is uniformly continuous; the limit theorems and subsequences. Admitted without proof in this unit.
  • ECalculus I. Proved in Calculus I: continuity and the differentiation rules, including that of the square root and those of the sine and the cosine; the Weierstrass theorem; the mean value theorems of Lagrange and of Cauchy and their corollaries, such as that of the zero derivative on an interval; the theorems of Bolzano and of the intermediate value and the continuity of the inverse; monotonicity from the sign of the derivative, and convexity; and the curvature of a graph. Admitted without proof in this unit.
  • FAlgebra and Geometry. Proved in Algebra and Geometry: the triangle inequality and the norm of plane vectors, the telescoping sum, polar coordinates, the trigonometric functions with the Pythagorean identity, the inequality between the sine, the arc and the tangent, and the area of the circular sector; these last two rest on the area of the disk that is admitted there, and on the inequality rest the derivatives of the sine and the cosine, from Calculus I. Admitted without proof in this unit.
  • GCylinder condition. A solid between the planes x=ax = a and x=bx = b whose cross-sections have integrable area A(x)A(x) satisfies the condition if for each ε>0\varepsilon > 0 there is a partition whose slices contain cylinders of height Δxk\Delta x_k with base of area αk\alpha_k (except when αk=0\alpha_k = 0) and are contained in cylinders of the same height with base of area βk\beta_k (both with the base in the plane x=xk−1x = x_{k - 1}), with αk≤A(x)≤βk\alpha_k \leq A(x) \leq \beta_k on each subinterval and ∑k=1n(βk−αk) Δxk<ε\sum_{k=1}^{n} (\beta_k - \alpha_k) \, \Delta x_k < \varepsilon: the way of enclosing a volume between sums. Established in: 8. Volume is the integral of the cross-sectional area.
  • HIterated Cauchy theorem. If FF and its derivatives up to order m−1m - 1 vanish at 00 and G(u)=upG(u) = u^p, with m≤pm \leq p, then F(ℓ) G(m)(ξ)=G(ℓ) F(m)(ξ)F(\ell) \, G^{(m)}(\xi) = G(\ell) \, F^{(m)}(\xi) for some ξ\xi in (0,ℓ)\ioo{0}{\ell}: the way of comparing an error with a power without the Taylor polynomial. Established in: 25. Iterated Cauchy theorem.
  1. Segment 1 Areas and the average value

    • Prove that the area between two graphs, if it satisfies the requirements, is the integral of the difference, and compute it when the graphs cross.
    • Enclose a polar region between sectors and obtain its area as half the integral of the square of the radius.
    • Characterize the average value of a function as a limit of averages of values and show that a continuous function attains it.
    1. The area between two graphs is the integral of the difference. If ff and gg are integrable on [a,b]\icc{a}{b}, with g≤fg \leq f, and the region between their graphs has an area that satisfies the requirements, that area is ∫ab(f−g)\int_a^b (f - g); this number is adopted as the definition of the area of the region.

      Rests on: B (Requirements and bracketing between sums), A (Completeness axiom), C (Definite integral). Proved in: Areas and the average value of a function.

    2. Area between two graphs that cross. If ff and gg are integrable on [a,b]\icc{a}{b}, the area of the region between their graphs is ∫ab∣f−g∣\int_a^b \abs{f - g}; if f−gf - g does not change sign on each interval of a partition a=c0<⋯<cm=ba = c_0 < \cdots < c_m = b, it is the sum of the absolute values of the integrals of f−gf - g over those intervals.

      Uses: 1. The area between two graphs is the integral of the difference. Rests on: C (Definite integral). Proved in: Areas and the average value of a function.

    3. Regions described with x as a function of y. If u≤vu \leq v are integrable on [c,d]\icc{c}{d} and the region of the points with c≤y≤dc \leq y \leq d and u(y)≤x≤v(y)u(y) \leq x \leq v(y) has an area that satisfies the requirements, that area is ∫cd(v−u)\int_c^d (v - u).

      Rests on: B (Requirements and bracketing between sums), A (Completeness axiom), C (Definite integral). Proved in: Areas and the average value of a function.

    4. The area in polar coordinates. If rr is continuous and nonnegative on [α,β]\icc{\alpha}{\beta}, with β−α≤2π\beta - \alpha \leq 2\pi, and the polar region under rr has an area that satisfies the requirements and gives each circular sector the area from Algebra and Geometry, that area is 12∫αβr(θ)2 dθ\frac{1}{2} \int_{\alpha}^{\beta} r(\theta)^2 \, \mathrm{d}\theta, which is adopted as the definition.

      Rests on: B (Requirements and bracketing between sums), F (Algebra and Geometry), E (Calculus I), C (Definite integral). Proved in: Areas and the average value of a function.

    5. The average value is the limit of the averages of the values. If ff is integrable on [a,b]\icc{a}{b}, the averages of the values f(ξk)f(\xi_k) weighted by the lengths of the subintervals tend to 1b−a∫abf\frac{1}{b - a} \int_a^b f as the norm of the partition tends to 00; in particular, the arithmetic means over uniform partitions.

      Rests on: C (Definite integral), A (Completeness axiom). Proved in: Areas and the average value of a function.

    6. Properties of the average value. The average value is linear, lies between the lower and upper bounds of the function, and that on [a,b]\icc{a}{b} is the average of those on [a,c]\icc{a}{c} and [c,b]\icc{c}{b} weighted by their lengths.

      Rests on: C (Definite integral). Proved in: Areas and the average value of a function.

    7. A continuous function attains its average value. If ff is continuous on [a,b]\icc{a}{b}, there exists ξ\xi in (a,b)\ioo{a}{b} with f(ξ)=1b−a∫abff(\xi) = \frac{1}{b - a} \int_a^b f.

      Rests on: C (Definite integral). Proved in: Areas and the average value of a function.

  2. Segment 2 Volumes

    • Lay down the requirements for volume and prove that a solid that satisfies the cylinder condition has as its volume the integral of the area of its cross-sections.
    • Deduce Cavalieri’s principle and the volume of the cone with a convex base lying between two continuous graphs, in particular that of the pyramid.
    • Compute volumes of revolution by disks, by washers and by shells, and prove that the two methods agree for a strictly monotonic function.
    1. Volume is the integral of the cross-sectional area. If a solid between the planes x=ax = a and x=bx = b has cross-sections of integrable area A(x)A(x), satisfies the cylinder condition and has a volume that satisfies the requirements, that volume is ∫abA(x) dx\int_a^b A(x) \, \mathrm{d}x; this number is adopted as the definition of volume.

      Rests on: B (Requirements and bracketing between sums), A (Completeness axiom), C (Definite integral). Proved in: Volumes of solids.

    2. Nested cross-sections. If in each interval [u,w]\icc{u}{w} contained in [a,b]\icc{a}{b}, with u<wu < w, there are two points x′x' and x′′x'' of that interval with Ex′⊆Ex⊆Ex′′E_{x'} \subseteq E_x \subseteq E_{x''} and A(x′)≤A(x)≤A(x′′)A(x') \leq A(x) \leq A(x'') for every xx in [u,w]\icc{u}{w}, the solid satisfies the cylinder condition.

      Uses: 8. Volume is the integral of the cross-sectional area. Rests on: G (Cylinder condition), C (Definite integral). Proved in: Volumes of solids.

    3. Cavalieri’s principle. If two solids satisfy the cylinder condition between the same planes and A1(x)=λ A2(x)A_1(x) = \lambda \, A_2(x) with λ>0\lambda > 0, the volume of the first is λ\lambda times that of the second; with cross-sections of equal area, the volumes are equal.

      Uses: 8. Volume is the integral of the cross-sectional area. Rests on: G (Cylinder condition), C (Definite integral). Proved in: Volumes of solids.

    4. Area of a scaled and translated figure. Scaling with ratio λ>0\lambda > 0 and translating the region between the graphs of two continuous functions yields another region of the same type, whose area is λ2\lambda^2 times the original one.

      Uses: 1. The area between two graphs is the integral of the difference. Rests on: E (Calculus I), C (Definite integral). Proved in: Volumes of solids.

    5. Volume of the cone. If the base BB of a cone is a convex figure of the plane x=hx = h, a region between two continuous graphs, and the vertex lies in the plane x=0x = 0, the cross-section at xx has area x2h2A(h)\frac{x^2}{h^2} A(h), the cone satisfies the cylinder condition and its volume is 13A(h) h\frac{1}{3} A(h) \, h; in particular, that of the pyramid.

      Uses: 11. Area of a scaled and translated figure, 8. Volume is the integral of the cross-sectional area. Rests on: G (Cylinder condition), A (Completeness axiom), C (Definite integral). Proved in: Volumes of solids.

    6. Volume by disks and washers. If 0≤g≤f0 \leq g \leq f are continuous on [a,b]\icc{a}{b}, the solid generated by revolving about the axis of abscissas the region between their graphs satisfies the cylinder condition, and its volume is π∫ab(f(x)2−g(x)2) dx\pi \int_a^b \left(f(x)^2 - g(x)^2\right) \, \mathrm{d}x.

      Uses: 1. The area between two graphs is the integral of the difference, 8. Volume is the integral of the cross-sectional area. Rests on: G (Cylinder condition), F (Algebra and Geometry), E (Calculus I), C (Definite integral). Proved in: Volumes of solids.

    7. Volume by cylindrical shells. If ff is continuous and nonnegative on [a,b]\icc{a}{b}, with 0≤a0 \leq a, and the solid generated by revolving about the axis of ordinates the region under its graph has a volume that satisfies the requirements, that volume is 2π∫abx f(x) dx2\pi \int_a^b x \, f(x) \, \mathrm{d}x.

      Rests on: B (Requirements and bracketing between sums), F (Algebra and Geometry), E (Calculus I), A (Completeness axiom), C (Definite integral). Proved in: Volumes of solids.

    8. Equivalence of washers and shells. If ff is continuous, nonnegative and strictly monotonic on [a,b]\icc{a}{b}, with 0≤a0 \leq a, the integral of the areas of the cross-sections perpendicular to the axis of ordinates, written with the inverse of ff, is equal to 2π∫abx f(x) dx2\pi \int_a^b x \, f(x) \, \mathrm{d}x: the volume by washers is that by shells. The proof uses no derivatives.

      Rests on: E (Calculus I), F (Algebra and Geometry), A (Completeness axiom), C (Definite integral). Proved in: Volumes of solids.

  3. Segment 3 Lengths and curvature

    • Define the length of a curve as a supremum of inscribed polygonal paths and prove that it does not depend on the parametrization.
    • Prove with Lagrange’s theorem and the Heine-Cantor theorem that the length of a curve with a continuous derivative is the integral of its speed.
    • Define curvature as the rate at which the tangent turns, obtain its formula and recognize in it that of Calculus I.
    1. The polygonal path does not get shorter under refinement. The length of a curve is the supremum of the lengths L(γ,P)L(\gamma, P) of its inscribed polygonal paths, if they are bounded; and if P⊆QP \subseteq Q, then L(γ,P)≤L(γ,Q)L(\gamma, P) \leq L(\gamma, Q), so that the distance between the endpoints does not exceed any polygonal path.

      Rests on: A (Completeness axiom), F (Algebra and Geometry). Proved in: Arc length and curvature.

    2. Length does not depend on the parametrization. If φ\varphi is strictly monotonic on [c,d]\icc{c}{d} with image [a,b]\icc{a}{b}, the curve γ∘φ\gamma \circ \varphi is rectifiable if and only if γ\gamma is, and then they have the same length.

      Uses: 16. The polygonal path does not get shorter under refinement. Rests on: E (Calculus I), A (Completeness axiom). Proved in: Arc length and curvature.

    3. Length of a curve with a continuous derivative. If x′x' and y′y' are continuous on [a,b]\icc{a}{b}, the curve is rectifiable, its length is ∫abx′(t)2+y′(t)2 dt\int_a^b \sqrt{x'(t)^2 + y'(t)^2} \, \mathrm{d}t, and the lengths of the inscribed polygonal paths tend to it as the norm of the partition tends to 00.

      Uses: 16. The polygonal path does not get shorter under refinement. Rests on: E (Calculus I), D (Sequences), F (Algebra and Geometry), A (Completeness axiom), C (Definite integral). Proved in: Arc length and curvature.

    4. Length of a graph. If f′f' is continuous on [a,b]\icc{a}{b}, the graph of ff is rectifiable and its length is ∫ab1+f′(x)2 dx\int_a^b \sqrt{1 + f'(x)^2} \, \mathrm{d}x.

      Uses: 18. Length of a curve with a continuous derivative. Proved in: Arc length and curvature.

    5. Length of a polar curve. If r′r' is continuous on [α,β]\icc{\alpha}{\beta}, the polar curve of radius rr is rectifiable and its length is ∫αβr2+r′2 dθ\int_{\alpha}^{\beta} \sqrt{r^2 + r'^2} \, \mathrm{d}\theta.

      Uses: 18. Length of a curve with a continuous derivative. Rests on: F (Algebra and Geometry), E (Calculus I). Proved in: Arc length and curvature.

    6. Curvature of a parametric curve. If x′x' and y′y' are continuous, γ′(t0)≠(0,0)\gamma'(t_0) \neq (0, 0), and x′x' and y′y' are differentiable on t0t_0, there is an angle function of the tangent differentiable on t0t_0, and the curvature, the absolute value of the derivative of the angle with respect to arc length, is ∣x′y′′−y′x′′∣(x′2+y′2)32\frac{\abs{x' y'' - y' x''}}{\left(x'^2 + y'^2\right)^{\frac{3}{2}}}.

      Uses: 18. Length of a curve with a continuous derivative. Rests on: E (Calculus I), F (Algebra and Geometry), C (Definite integral). Proved in: Arc length and curvature.

    7. The curvature of a graph. For the curve (t,f(t))(t, f(t)), with f′f' continuous and ff twice differentiable on x0x_0, the curvature at x0x_0 is ∣f′′(x0)∣(1+f′(x0)2)32\frac{\abs{f''(x_0)}}{\left(1 + f'(x_0)^2\right)^{\frac{3}{2}}}, that of the Calculus I lesson on curvature.

      Uses: 21. Curvature of a parametric curve. Rests on: E (Calculus I). Proved in: Arc length and curvature.

  4. Segment 4 The error of the numerical rules

    • Recognize in the midpoint, trapezoidal and Simpson’s rules exact integrals of interpolating polynomials, and relate them to one another.
    • Prove the bounds for their errors with the iterated Cauchy theorem and the derivative of an integral with variable limits, without the Taylor polynomial.
    • Decide how many subintervals guarantee a given error and how the error decreases when the number of subintervals is doubled.
    1. Each rule integrates an interpolating polynomial. On each subinterval, the midpoint rule integrates the constant f(ck)f(c_k), the trapezoidal rule a polynomial of degree at most one that passes through the points at the endpoints, and Simpson’s rule, on each pair of subintervals, one of degree at most two that passes through three points of the curve.

      Rests on: E (Calculus I), C (Definite integral). Proved in: Numerical integration.

    2. Relations among the three rules. For every natural number nn, T2n=Tn+Rn2T_{2n} = \frac{T_n + R_n}{2} and S2n=Tn+2Rn3S_{2n} = \frac{T_n + 2 R_n}{3}; if ff is integrable, RnR_n, TnT_n and S2nS_{2n} tend to ∫abf\int_a^b f.

      Rests on: D (Sequences), C (Definite integral). Proved in: Numerical integration.

    3. Iterated Cauchy theorem. If FF is continuous on [0,ℓ]\icc{0}{\ell}, differentiable mm times on [0,ℓ)\ico{0}{\ell} with its derivatives up to order m−1m - 1 equal to zero at 00, and G(u)=upG(u) = u^p with m≤pm \leq p, there exists ξ\xi in (0,ℓ)\ioo{0}{\ell} with F(ℓ) G(m)(ξ)=G(ℓ) F(m)(ξ)F(\ell) \, G^{(m)}(\xi) = G(\ell) \, F^{(m)}(\xi).

      Rests on: E (Calculus I). Proved in: Numerical integration.

    4. Error of the midpoint and trapezoidal rules. If ff is continuous on [a,b]\icc{a}{b} and twice differentiable on (a,b)\ioo{a}{b}, with ∣f′′∣≤B2\abs{f''} \leq B_2, then ∣ER∣≤(b−a)h224B2\abs{E_R} \leq \frac{(b - a) h^2}{24} B_2 and ∣ET∣≤(b−a)h212B2\abs{E_T} \leq \frac{(b - a) h^2}{12} B_2.

      Rests on: H (Iterated Cauchy theorem), E (Calculus I), C (Definite integral). Proved in: Numerical integration.

    5. Error of Simpson’s rule. If ff is continuous on [a,b]\icc{a}{b} and four times differentiable on (a,b)\ioo{a}{b}, with ∣f(4)∣≤B4\abs{f^{(4)}} \leq B_4, and nn is even, then ∣ES∣≤(b−a)h4180B4\abs{E_S} \leq \frac{(b - a) h^4}{180} B_4.

      Rests on: H (Iterated Cauchy theorem), E (Calculus I), C (Definite integral). Proved in: Numerical integration.

    6. Degree of exactness and sign of the error. The midpoint and trapezoidal rules are exact for polynomials of degree at most one, and Simpson’s rule for those of degree at most three; if f′′≥0f'' \geq 0, then Rn≤∫abf≤TnR_n \leq \int_a^b f \leq T_n, and if f(4)≥0f^{(4)} \geq 0, then ∫abf≤Sn\int_a^b f \leq S_n.

      Uses: 26. Error of the midpoint and trapezoidal rules, 27. Error of Simpson’s rule. Rests on: E (Calculus I). Proved in: Numerical integration.

    7. The error when the number of subintervals is doubled. If f′′f'' is continuous on [a,b]\icc{a}{b}, then n2ET(n)→−(b−a)212(f′(b)−f′(a))n^2 E_T(n) \to -\frac{(b - a)^2}{12} \left(f'(b) - f'(a)\right), and half of that with the opposite sign for the midpoint rule; if f′(b)≠f′(a)f'(b) \neq f'(a), each error is divided by four, in the limit, when nn is doubled.

      Uses: 26. Error of the midpoint and trapezoidal rules. Rests on: D (Sequences), E (Calculus I), C (Definite integral). Proved in: Numerical integration.

Toward improper integrals. The chain stops where regions and solids cease to be bounded. All the quantities of this unit were defined with integrals of bounded functions on closed and bounded intervals, because only for them are there lower and upper sums. The unit on improper integrals measures a region under a graph on an unbounded interval, or under an unbounded function, as the limit of the measures of its bounded portions: the solid generated by the graph of 1x\frac{1}{x}, for x≥1x \geq 1, truncated at x=cx = c, has by disks the volume π(1−1c)\pi \left(1 - \frac{1}{c}\right), and its volume will be the limit as c→∞c \to \infty. There the full length of a spiral that makes infinitely many turns will also be computed, and the area under a decreasing graph, enclosed between sums of rectangles as in this unit, will give, in the unit on series, the integral test.