Skip to content
Topos Uranos

Summary

This lesson defines the length of a plane curve and computes it with the integral. A parametric curve is a point that moves according to two continuous functions; its length is defined as the supremum of the lengths of the polygonal paths inscribed in it, which do not get shorter when vertices are added, and the curve is rectifiable if that supremum exists. The central theorem proves that, if the two functions have continuous derivatives, the curve is rectifiable and its length is the integral of the speed: Lagrange’s mean value theorem, applied to each coordinate separately, writes each side of the polygonal path almost as a term of a Riemann sum, and the Heine-Cantor theorem bounds the difference. From it follow the length of a graph and that of a polar curve, the arc length function and the curvature, defined as the rate at which the tangent turns per unit of length traveled, which for a graph coincides with that of Calculus I. The problems measure the circle, the catenary, the cycloid and a spiral, and show a continuous curve that has no length.

Learning objectives

  1. Define the length of a plane parametric curve as the supremum of the lengths of its inscribed polygonal paths, and recognize rectifiable and non-rectifiable curves.
  2. Prove that a curve with a continuous derivative is rectifiable and that its length is the integral of its speed, and derive the formulas for graphs and polar curves.
  3. Define the arc length function and the curvature as the rate at which the tangent turns, prove its formula for parametric curves and check that it coincides with that of Calculus I for a graph.
  4. Compute lengths and curvatures of classical curves, choosing the identity or the splitting of the interval that removes the square root.

Curves and inscribed polygonal paths

The lesson on angles and their measure assumed that every arc of a circle has a length, defined the number π\pi as half the length of the circle of radius 11 and announced that the length of an arc is defined with inscribed polygonal paths and computed with the integral. This lesson fulfills that announcement for a broad family of curves. As the lesson on the area problem did with area, it is best to settle first what is being measured: the length of a line segment is the distance between its endpoints, and a curve is approximated by polygonal paths whose vertices lie on it. The bare minimum about parametric curves suffices here; their general study, in space and with their geometric properties, belongs to Calculus III.

DefinitionPlane parametric curve

Let a<ba < b. A plane parametric curve on [a,b]\icc{a}{b} is a pair of functions xx and yy continuous on [a,b]\icc{a}{b}; one writes γ(t)=(x(t),y(t))\gamma(t) = (x(t), y(t)), the point of the plane where the curve is at the instant tt. For a point or vector A=(p,q)A = (p, q) one writes ∣A∣=p2+q2\abs{A} = \sqrt{p^2 + q^2}, its norm, from the lesson on vectors in the plane, so that ∣A−B∣\abs{A - B} is the distance between the points AA and BB. If xx and yy are differentiable at tt, the vector γ′(t)=(x′(t),y′(t))\gamma'(t) = (x'(t), y'(t)) is the velocity of the curve at tt, and its norm, the speed:

∣γ′(t)∣=x′(t)2+y′(t)2\abs{\gamma'(t)} = \sqrt{x'(t)^2 + y'(t)^2}

The graph of a continuous function ff on [a,b]\icc{a}{b} is the curve γ(t)=(t,f(t))\gamma(t) = (t, f(t)); the circle of radius RR centered at the origin is the curve γ(t)=(Rcos⁡t,Rsin⁡t)\gamma(t) = (R \cos t, R \sin t) on [0,2π]\icc{0}{2\pi}. A curve is more than the set of its points: the curve (cos⁡2t,sin⁡2t)(\cos 2t, \sin 2t) on [0,2π]\icc{0}{2\pi} traverses the same circle twice, and its length will be twice as large.

DefinitionLength of a curve

Let γ\gamma be a plane parametric curve on [a,b]\icc{a}{b} and P={t0,t1,…,tn}P = \{t_0, t_1, \ldots, t_n\} a partition of [a,b]\icc{a}{b}, in the sense of the lesson on the Riemann integral, with Δtk=tk−tk−1\Delta t_k = t_k - t_{k - 1}. The inscribed polygonal path of PP is the one that joins, in order, the points γ(t0),γ(t1),…,γ(tn)\gamma(t_0), \gamma(t_1), \ldots, \gamma(t_n), and its length is the sum of the lengths of its sides:

L(γ,P)=∑k=1n∣γ(tk)−γ(tk−1)∣L(\gamma, P) = \sum_{k=1}^{n} \abs{\gamma(t_k) - \gamma(t_{k - 1})}

The curve is rectifiable if the set of the numbers L(γ,P)L(\gamma, P), for all partitions PP of [a,b]\icc{a}{b}, is bounded above. In that case, its supremum, which exists by the completeness axiom, from the lesson on number systems, is the length of the curve, L(γ)L(\gamma).

The letter is that of the lower sums of this unit, but its first argument is a curve, not a function, and no lower sums appear in this lesson. The supremum is taken, and not a limit, because an inscribed polygonal path is never longer than the curve: the lemma below makes this plausible, and the rectification theorem, further on, will show that, for curves with a continuous derivative, the supremum is also the limit as the norm tends to 00, which is how Algebra and Geometry described it.

LemmaThe polygonal path does not get shorter under refinement

Let γ\gamma be a plane parametric curve on [a,b]\icc{a}{b}, and PP and QQ partitions of [a,b]\icc{a}{b} with P⊆QP \subseteq Q. Then L(γ,P)≤L(γ,Q)L(\gamma, P) \leq L(\gamma, Q). In particular, ∣γ(b)−γ(a)∣≤L(γ,Q)\abs{\gamma(b) - \gamma(a)} \leq L(\gamma, Q) for every partition QQ. When a vertex is added, one side is replaced by two that form a triangle with it.

Proof

  1. ∣γ(tk)−γ(tk−1)∣≤∣γ(c)−γ(tk−1)∣+∣γ(tk)−γ(c)∣→L(γ,P)≤L(γ,P∪{c})\abs{\gamma(t_k) - \gamma(t_{k - 1})} \leq \abs{\gamma(c) - \gamma(t_{k - 1})} + \abs{\gamma(t_k) - \gamma(c)} \rightarrow \resaltar{L(\gamma, P) \leq L(\gamma, P \cup \{c\})}

    First let Q=P∪{c}Q = P \cup \{c\}, with tk−1<c<tkt_{k - 1} < c < t_k. The sides of the two polygonal paths coincide except the one joining γ(tk−1)\gamma(t_{k - 1}) to γ(tk)\gamma(t_k), which in the path of QQ is replaced by two; the triangle inequality theorem, from the lesson on vectors in the plane, applied to the three points, gives the inequality.

  2. L(γ,P)≤L(γ,Q),∣γ(b)−γ(a)∣=L(γ,{a,b})≤L(γ,Q)L(\gamma, P) \leq \resaltar{L(\gamma, Q)}, \qquad \abs{\gamma(b) - \gamma(a)} = L(\gamma, \{a, b\}) \leq \resaltar{L(\gamma, Q)}

    If QQ has mm more points than PP, they are added one at a time; by induction on mm, each addition applies the preceding case to the partition obtained so far, and the inequalities are combined. The partition {a,b}\{a, b\} is contained in every partition QQ, and its polygonal path is the line segment from γ(a)\gamma(a) to γ(b)\gamma(b).

Length is a property of the path traced, not of the way its instants are numbered: if the curve is traversed in the same order, or in the reverse order, with another parameter, the inscribed polygonal paths are the same.

PropositionLength does not depend on the parametrization

Let γ\gamma be a plane parametric curve on [a,b]\icc{a}{b}, c<dc < d, and φ\varphi a strictly monotonic function on [c,d]\icc{c}{d} whose image is [a,b]\icc{a}{b}. Then the curve γ∘φ\gamma \circ \varphi, on [c,d]\icc{c}{d}, is rectifiable if and only if γ\gamma is, and in that case the two have the same length.

Proof

  1. Q↦φ(Q)={φ(u0),…,φ(un)}Q \mapsto \varphi(Q) = \{\varphi(u_0), \ldots, \varphi(u_n)\}

    The function φ\varphi is injective, being strictly monotonic, and its image is [a,b]\icc{a}{b}: it is a bijection from [c,d]\icc{c}{d} onto [a,b]\icc{a}{b}, increasing or decreasing. By the theorem on monotonic functions whose image is an interval, from the lesson on the continuity of the elementary functions, φ\varphi is continuous, and by the theorem on the composition of continuous functions, from the lesson on limit and continuity, the coordinates of γ∘φ\gamma \circ \varphi are continuous on [c,d]\icc{c}{d}: it is a plane parametric curve. If φ\varphi is increasing, φ(c)\varphi(c) and φ(d)\varphi(d) are the smallest and the largest of its values, which are aa and bb; if it is decreasing, they are bb and aa. Hence, if Q={u0,…,un}Q = \{u_0, \ldots, u_n\} is a partition of [c,d]\icc{c}{d}, the points φ(uk)\varphi(u_k) form a partition of [a,b]\icc{a}{b}, in the same order if φ\varphi is increasing and in the reverse order if it is decreasing; and every partition of [a,b]\icc{a}{b} is obtained in this way, from the partition formed by its preimages.

  2. L(γ∘φ,Q)=L(γ,φ(Q))→L(γ∘φ)=L(γ)L(\gamma \circ \varphi, Q) = L(\gamma, \varphi(Q)) \rightarrow \resaltar{L(\gamma \circ \varphi) = L(\gamma)}

    The sides of the polygonal path of QQ for γ∘φ\gamma \circ \varphi are the line segments joining γ(φ(uk−1))\gamma(\varphi(u_{k - 1})) to γ(φ(uk))\gamma(\varphi(u_k)), which are the sides of the polygonal path of φ(Q)\varphi(Q) for γ\gamma, traversed in the same order or in the reverse order; the sums coincide. The two sets of lengths of polygonal paths are equal: they are bounded at the same time, and they have the same supremum.