math.wiki
Calculus › Vectors and space curves › Vector-valued functions
Sign in

Vector-valued functions

A waterslide winds twice around a tower as it drops to the pool, and its shape is three formulas in one parameter. What do the limit, the derivative and the integral of such a vector formula mean, and which rules of one-variable calculus survive the move to space?

Where you are 5 in Vectors and space curves

Why this exists

Vector valued functions and motion in the plane differentiated and integrated \mathbf{r}(t) = \langle x(t), y(t) \rangle one component at a time. Space adds a third component and new questions. A curve in space is often given as the place where two surfaces meet, not by formulas, and has to be parametrized first. Its tangent is a line in space, which no slope describes. And space has a product the plane lacks, the cross product of The cross product and the triple product, so we need the derivative of \mathbf{u}(t) \times \mathbf{v}(t).

Here limits are defined by distance, the derivative rules are proved, and we find which one-variable theorems survive. The mean value theorem does not.

Definition

Let I be an interval. A vector-valued function on I is a rule \mathbf{r} : I \to \mathbb{R}^3 that assigns to each t \in I a vector

\mathbf{r}(t) = \langle x(t), y(t), z(t) \rangle = x(t)\,\mathbf{i} + y(t)\,\mathbf{j} + z(t)\,\mathbf{k}.

The real functions x, y, z are its components. When \mathbf{r} is given by a formula and no interval is named, its domain is the set of t at which all three components are defined, the intersection of their domains. Drawn from the origin, the tips of the vectors \mathbf{r}(t) form the curve traced by \mathbf{r}, and \mathbf{r} is a parametrization of that curve. A curve is a set of points; a parametrization also says in what order, and how fast, the points are visited (Parametric equations).

Limit. Let \mathbf{r} be defined on an interval containing a, except possibly at a. Then \lim_{t \to a} \mathbf{r}(t) = \mathbf{L} means: for every \varepsilon > 0 there is a \delta > 0 such that \lVert \mathbf{r}(t) - \mathbf{L} \rVert < \varepsilon whenever 0 < \lvert t - a \rvert < \delta, as in The epsilon delta definition of a limit with distance in place of absolute value. One-sided limits at an endpoint of I take t on one side only.

Continuity. \mathbf{r} is continuous at a when \lim_{t \to a} \mathbf{r}(t) = \mathbf{r}(a), and continuous on I when it is continuous at every point of I, one-sided at an endpoint.

Derivative. The derivative of \mathbf{r} at a is

\mathbf{r}'(a) = \lim_{h \to 0} \frac{\mathbf{r}(a + h) - \mathbf{r}(a)}{h},

when the limit exists; then \mathbf{r} is differentiable at a. If \mathbf{r}'(a) \ne \mathbf{0}, it is a tangent vector to the curve at \mathbf{r}(a), and the tangent line there is the line \boldsymbol{\ell}(s) = \mathbf{r}(a) + s\,\mathbf{r}'(a) of Lines and planes in space.

Smooth. A parametrization \mathbf{r} is smooth on I when \mathbf{r}' exists and is continuous on I and \mathbf{r}'(t) \ne \mathbf{0} for every t \in I. The unit tangent and the curvature of Arc length and curvature are built on smooth parametrizations.

Integral. If x, y, z are continuous on [a, b],

\begin{aligned} \int_a^b \mathbf{r}(t)\,dt = \Big\langle &\int_a^b x(t)\,dt, \int_a^b y(t)\,dt, \\ &\int_a^b z(t)\,dt \Big\rangle. \end{aligned}

The components of a vector Riemann sum \sum \mathbf{r}(t_i^*)\,\Delta t are Riemann sums of x, y, z, so this is its limit (The definite integral as a limit of Riemann sums). An antiderivative of \mathbf{r} on I is any \mathbf{R} with \mathbf{R}' = \mathbf{r} on I.

Intuition

z P Q
\mathbf{r}'(a)
Figure 1Jin’s waterslide of Example 2, \mathbf{r}(t) = \langle 4\cos t, 4\sin t, 14 - t \rangle, far side dashed. The chord from P = \mathbf{r}(a) to Q = \mathbf{r}(a + h) turns toward the tangent vector \mathbf{r}'(a) as h \to 0.

The vector \mathbf{r}(a + h) - \mathbf{r}(a) runs from P = \mathbf{r}(a) to a nearby point Q = \mathbf{r}(a + h) of the curve: a chord. Dividing by h rescales it (and for h < 0 reverses a backward chord), so the quotient points the way t increases. As h \to 0 the chord turns toward the direction in which the curve leaves P, and the quotient settles on \mathbf{r}'(a), whose length says how fast the parametrization covers the curve there.

That length belongs to the parametrization; the tangent line belongs to the curve. Covering the curve twice as fast, \mathbf{r}(2s), doubles the tangent vector (Theorem 2(e)) and leaves the line where it was. Where \mathbf{r}'(a) = \mathbf{0} the quotient has no direction to settle on, and the curve may turn a corner (Boundaries).

Results

Throughout, \mathbf{r} = \langle x, y, z \rangle, \mathbf{u} and \mathbf{v} are vector-valued functions on an interval I, f is a real function on I, c is a number and \mathbf{c} a constant vector.

Theorem 1 (components). \lim_{t \to a} \mathbf{r}(t) = \langle L_1, L_2, L_3 \rangle if and only if x(t) \to L_1, y(t) \to L_2 and z(t) \to L_3 as t \to a. Consequently \mathbf{r} is continuous at a if and only if x, y, z are, and differentiable at a if and only if x, y, z are, with

\mathbf{r}'(a) = \langle x'(a), y'(a), z'(a) \rangle.

Theorem 2 (derivative rules). If \mathbf{r} is differentiable at t, it is continuous at t. If \mathbf{u}, \mathbf{v} and f are differentiable at t, then so are the following, with

  • (a) (\mathbf{u} + \mathbf{v})' = \mathbf{u}' + \mathbf{v}' and (c\,\mathbf{u})' = c\,\mathbf{u}';
  • (b) (f\mathbf{u})' = f'\mathbf{u} + f\mathbf{u}';
  • (c) (\mathbf{u} \cdot \mathbf{v})' = \mathbf{u}' \cdot \mathbf{v} + \mathbf{u} \cdot \mathbf{v}';
  • (d) (\mathbf{u} \times \mathbf{v})' = \mathbf{u}' \times \mathbf{v} + \mathbf{u} \times \mathbf{v}', the order of each product kept;
  • (e) if \varphi is a real function differentiable at s with \varphi(s) = t, then (\mathbf{u} \circ \varphi)'(s) = \varphi'(s)\,\mathbf{u}'(\varphi(s)).

Theorem 3 (constant length). Let \mathbf{r} be differentiable on I. If \lVert \mathbf{r}(t) \rVert is constant on I, then \mathbf{r}(t) \cdot \mathbf{r}'(t) = 0 for every t \in I. Conversely, if \mathbf{r} \cdot \mathbf{r}' = 0 on I, then \lVert \mathbf{r} \rVert is constant on I.

Theorem 4 (integrals). Let \mathbf{r} be continuous on [a, b].

  • (a) If \mathbf{R}' = \mathbf{r} on [a, b], then \int_a^b \mathbf{r}(t)\,dt = \mathbf{R}(b) - \mathbf{R}(a). The function t \mapsto \int_a^t \mathbf{r}(\tau)\,d\tau is an antiderivative of \mathbf{r}, and any two antiderivatives on an interval differ by a constant vector.
  • (b) \mathbf{c} \cdot \int_a^b \mathbf{r}(t)\,dt = \int_a^b \mathbf{c} \cdot \mathbf{r}(t)\,dt.
  • (c) \left\lVert \int_a^b \mathbf{r}(t)\,dt \right\rVert \le \int_a^b \lVert \mathbf{r}(t) \rVert\,dt.
Why it is true

Theorem 1. For any vector \mathbf{w} = \langle w_1, w_2, w_3 \rangle and each i,

\lvert w_i \rvert \le \lVert \mathbf{w} \rVert \le \lvert w_1 \rvert + \lvert w_2 \rvert + \lvert w_3 \rvert.

The first holds because \lVert \mathbf{w} \rVert^2 = w_1^2 + w_2^2 + w_3^2 \ge w_i^2; the second because the square of the right side is w_1^2 + w_2^2 + w_3^2 plus nonnegative cross terms. Take \mathbf{w} = \mathbf{r}(t) - \mathbf{L}. If \mathbf{r}(t) \to \mathbf{L}, the \delta that makes \lVert \mathbf{r}(t) - \mathbf{L} \rVert < \varepsilon also makes \lvert x(t) - L_1 \rvert < \varepsilon, and the same for y and z. Conversely, if each component converges, choose \delta_1, \delta_2, \delta_3 making the three component differences less than \varepsilon/3, and let \delta be the smallest; then \lVert \mathbf{r}(t) - \mathbf{L} \rVert < \varepsilon. Continuity is the case \mathbf{L} = \mathbf{r}(a). The difference quotient of \mathbf{r} has as components the difference quotients of x, y, z, so applying the first part to it, as a function of h, gives the derivative statement.

Theorem 2. Continuity first: for h \ne 0, \displaystyle \mathbf{r}(t + h) - \mathbf{r}(t) = h \cdot \frac{\mathbf{r}(t + h) - \mathbf{r}(t)}{h}, and by Theorem 1, component by component, this tends to 0 \cdot \mathbf{r}'(t) = \mathbf{0}.

For (d), add and subtract \mathbf{u}(t) \times \mathbf{v}(t + h) and use the distributive law (Theorem 2 of The cross product and the triple product):

\begin{aligned} &\mathbf{u}(t + h) \times \mathbf{v}(t + h) - \mathbf{u}(t) \times \mathbf{v}(t) \\ &= [\mathbf{u}(t + h) - \mathbf{u}(t)] \times \mathbf{v}(t + h) \\ &\quad + \mathbf{u}(t) \times [\mathbf{v}(t + h) - \mathbf{v}(t)]. \end{aligned}

Divide by h, which a cross product lets us move into either factor. As h \to 0 the first bracket over h tends to \mathbf{u}'(t) and \mathbf{v}(t + h) \to \mathbf{v}(t) by continuity. The components of \mathbf{p} \times \mathbf{q} are polynomials in the six components of \mathbf{p} and \mathbf{q}, so the cross product is continuous, and the first term tends to \mathbf{u}'(t) \times \mathbf{v}(t); the second tends to \mathbf{u}(t) \times \mathbf{v}'(t). The same computation with the dot product proves (c), and with a number times a vector proves (b). Part (a) is Theorem 1 with the sum and constant multiple rules for each component, and (e) is Theorem 1 with the one-variable chain rule (The chain rule) applied to x(\varphi(s)), y(\varphi(s)), z(\varphi(s)).

Theorem 3. By Theorem 2(c) with \mathbf{u} = \mathbf{v} = \mathbf{r}, (\mathbf{r} \cdot \mathbf{r})' = 2\,\mathbf{r} \cdot \mathbf{r}'. If \lVert \mathbf{r} \rVert = k on I, then \mathbf{r} \cdot \mathbf{r} = k^2 is constant and its derivative is 0. Conversely, if \mathbf{r} \cdot \mathbf{r}' = 0 on I, then \mathbf{r} \cdot \mathbf{r} has derivative 0 on an interval, so it is constant by the mean value theorem (The mean value theorem), and so is its square root.

Theorem 4. (a) is the fundamental theorem of calculus (The second fundamental theorem of calculus) in each component, together with the fact that a real function with derivative 0 on an interval is constant. (b): both sides equal c_1 \int_a^b x\,dt + c_2 \int_a^b y\,dt + c_3 \int_a^b z\,dt, by linearity of the integral. (c): let \mathbf{D} = \int_a^b \mathbf{r}\,dt. If \mathbf{D} = \mathbf{0} there is nothing to prove. Otherwise, by (b) with \mathbf{c} = \mathbf{D},

\lVert \mathbf{D} \rVert^2 = \mathbf{D} \cdot \mathbf{D} = \int_a^b \mathbf{D} \cdot \mathbf{r}(t)\,dt.

Since \mathbf{D} \cdot \mathbf{r}(t) \le \lVert \mathbf{D} \rVert\,\lVert \mathbf{r}(t) \rVert (by \cos\theta \le 1, The dot product and the angle between vectors, or trivially when \mathbf{r}(t) = \mathbf{0}) and integrals of continuous functions preserve \le, we get \lVert \mathbf{D} \rVert^2 \le \lVert \mathbf{D} \rVert \int_a^b \lVert \mathbf{r} \rVert\,dt. Divide by \lVert \mathbf{D} \rVert > 0.

Worked examples

Example 1

A limit, and the value that makes it continuous

Let \displaystyle \mathbf{r}(t) = \left\langle \frac{\sin 3t}{t}, \frac{e^{2t} - 1}{t}, \sqrt{4 - t} \right\rangle. Find the domain of \mathbf{r} and \lim_{t \to 0} \mathbf{r}(t), and say how to define \mathbf{r}(0) so that \mathbf{r} becomes continuous on (-\infty, 4].

Solution

The first two components are defined for t \ne 0, the third for t \le 4. The domain is the intersection, (-\infty, 0) \cup (0, 4].

By Theorem 1 the limit is taken one component at a time. Writing \displaystyle \frac{\sin 3t}{t} = 3 \cdot \frac{\sin 3t}{3t} and using \displaystyle \frac{\sin u}{u} \to 1 as u \to 0 gives 3. Writing \displaystyle \frac{e^{2t} - 1}{t} = 2 \cdot \frac{e^{2t} - 1}{2t} and using \displaystyle \frac{e^u - 1}{u} \to 1, the derivative of e^u at 0 (The derivative at a point), gives 2. The root is continuous at 0, so it tends to \displaystyle \sqrt{4} = 2. Hence

\lim_{t \to 0} \mathbf{r}(t) = \langle 3, 2, 2 \rangle.

Define \mathbf{r}(0) = \langle 3, 2, 2 \rangle. Then each component is continuous on (-\infty, 4] (from the left at 4), and so is \mathbf{r}, by Theorem 1.

Example 2

Jin’s waterslide

Jin designs a waterslide that winds twice around a round tower as it drops to the pool. With the z-axis up the center of the tower and lengths in meters, the centerline of the slide is

\mathbf{r}(t) = \langle 4\cos t, 4\sin t, 14 - t \rangle, \quad 0 \le t \le 4\pi,

where t is the angle turned. Show that this parametrization is smooth and that the slide descends at the same angle everywhere, find that angle, and find the tangent line at \displaystyle t = \frac{\pi}{3}.

Solution

By Theorem 1, differentiate each component:

\mathbf{r}'(t) = \langle -4\sin t, 4\cos t, -1 \rangle.

It is continuous, and never \mathbf{0} because its third component is -1, so \mathbf{r} is smooth. Its length is \displaystyle \sqrt{16\sin^2 t + 16\cos^2 t + 1} = \sqrt{17} for every t.

The angle \beta between \mathbf{r}'(t) and the downward vertical -\mathbf{k} has \displaystyle \cos\beta = \frac{\mathbf{r}'(t) \cdot (-\mathbf{k})}{\sqrt{17}} = \frac{1}{\sqrt{17}}. The slide dips below the horizontal by \displaystyle \alpha = \frac{\pi}{2} - \beta, so \displaystyle \sin\alpha = \frac{1}{\sqrt{17}}, and since the horizontal part \langle -4\sin t, 4\cos t \rangle has length 4, \displaystyle \tan\alpha = \frac{1}{4}. Thus \displaystyle \alpha = \arctan\frac{1}{4} \approx 0.2450 radians, about 14.04°, at every point.

At \displaystyle t = \frac{\pi}{3} the point is \displaystyle \mathbf{r}\left(\frac{\pi}{3}\right) = \left\langle 2, 2\sqrt{3}, 14 - \frac{\pi}{3} \right\rangle and the tangent vector is \displaystyle \mathbf{r}'\left(\frac{\pi}{3}\right) = \langle -2\sqrt{3}, 2, -1 \rangle. The tangent line is

\boldsymbol{\ell}(s) = \left\langle 2 - 2\sqrt{3}\,s, 2\sqrt{3} + 2s, 14 - \tfrac{\pi}{3} - s \right\rangle.

Check: its horizontal part \displaystyle \langle -2\sqrt{3}, 2 \rangle is perpendicular to the radius \displaystyle \langle 2, 2\sqrt{3} \rangle, so from above the line touches the circle x^2 + y^2 = 16.

Example 3

A curve where two surfaces meet

The cylinder x^2 + y^2 = 4 meets the saddle z = x^2 - y^2 in a closed curve C. Parametrize C, find its tangent line at \displaystyle (\sqrt{2}, \sqrt{2}, 0), and find the points where the tangent is horizontal.

Solution

The points of the cylinder are (2\cos t, 2\sin t, z), and 0 \le t < 2\pi reaches each (x, y) on the circle once. On the saddle, z = 4\cos^2 t - 4\sin^2 t = 4\cos 2t. So

\mathbf{r}(t) = \langle 2\cos t, 2\sin t, 4\cos 2t \rangle, \quad 0 \le t \le 2\pi,

traces all of C: a point on both surfaces has (x, y) on the circle, so (x, y) = (2\cos t, 2\sin t) for some t, and then z is forced.

Differentiating, \mathbf{r}'(t) = \langle -2\sin t, 2\cos t, -8\sin 2t \rangle, never \mathbf{0} because \sin t and \cos t never vanish together. The point \displaystyle (\sqrt{2}, \sqrt{2}, 0) is \displaystyle t = \frac{\pi}{4}, where \displaystyle \mathbf{r}'\left(\frac{\pi}{4}\right) = \langle -\sqrt{2}, \sqrt{2}, -8 \rangle, so the tangent line is

\boldsymbol{\ell}(s) = \langle \sqrt{2} - \sqrt{2}\,s, \sqrt{2} + \sqrt{2}\,s, -8s \rangle.

The tangent is horizontal where z'(t) = -8\sin 2t = 0, which is \displaystyle t = 0, \frac{\pi}{2}, \pi, \frac{3\pi}{2}: the points (2, 0, 4) and (-2, 0, 4), highest since 4\cos 2t \le 4, and (0, 2, -4) and (0, -2, -4), lowest.

z y x P
\boldsymbol{\ell}
(2, 0, 4)
(0, 2, -4)
Figure 2The curve of Example 3 where the cylinder x^2 + y^2 = 4 meets the saddle z = x^2 - y^2, with its tangent line \boldsymbol{\ell} at \displaystyle P(\sqrt{2}, \sqrt{2}, 0). The tangent is horizontal at the two high and two low points.
Example 4

The cross product rule, checked

Let \mathbf{u}(t) = \langle 1, t, t^2 \rangle and \mathbf{v}(t) = \langle t, 2, -t \rangle. Find (\mathbf{u} \times \mathbf{v})' by Theorem 2(d), and check it by expanding \mathbf{u} \times \mathbf{v} first.

Solution

Here \mathbf{u}' = \langle 0, 1, 2t \rangle and \mathbf{v}' = \langle 1, 0, -1 \rangle. Expanding each determinant along its first row,

\begin{aligned} \mathbf{u}' \times \mathbf{v} &= \langle -5t, 2t^2, -t \rangle, \\ \mathbf{u} \times \mathbf{v}' &= \langle -t, 1 + t^2, -t \rangle, \end{aligned}

so (\mathbf{u} \times \mathbf{v})' = \langle -6t, 1 + 3t^2, -2t \rangle.

Check: expanding, \mathbf{u} \times \mathbf{v} = \langle -3t^2, t + t^3, 2 - t^2 \rangle, whose derivative is \langle -6t, 1 + 3t^2, -2t \rangle. The order matters: \mathbf{u}' \times \mathbf{v} + \mathbf{v}' \times \mathbf{u} is \langle -4t, t^2 - 1, 0 \rangle, which is not the derivative.

Example 5

A proof: acceleration along the position keeps a curve in a plane

Let \mathbf{r} be twice differentiable on an interval, with \mathbf{r}''(t) = \lambda(t)\,\mathbf{r}(t) for some real number \lambda(t) at each t. Prove that \mathbf{r}(t) \times \mathbf{r}'(t) is a constant vector \mathbf{L}, and that if \mathbf{L} \ne \mathbf{0} the curve lies in the plane through the origin with normal \mathbf{L}. Illustrate with \mathbf{r}(t) = \langle \cos t, \cos t + \sin t, 2\sin t \rangle.

Solution

By Theorem 2(d),

(\mathbf{r} \times \mathbf{r}')' = \mathbf{r}' \times \mathbf{r}' + \mathbf{r} \times \mathbf{r}'' = \mathbf{0} + \lambda\,(\mathbf{r} \times \mathbf{r}) = \mathbf{0},

because a vector crossed with itself is \mathbf{0} (Theorem 2 of The cross product and the triple product). A vector function with zero derivative on an interval is constant (Theorem 4(a)), so \mathbf{r} \times \mathbf{r}' = \mathbf{L}. Then \mathbf{L} \cdot \mathbf{r}(t) = \mathbf{r}(t) \cdot (\mathbf{r}(t) \times \mathbf{r}'(t)) = 0, a triple product with a repeated vector (Theorem 1 of The cross product and the triple product). Every point of the curve satisfies \mathbf{L} \cdot \mathbf{r} = 0, the equation of the plane through the origin with normal \mathbf{L}.

For the illustration, \mathbf{r}''(t) = \langle -\cos t, -\cos t - \sin t, -2\sin t \rangle = -\mathbf{r}(t), so \lambda = -1. Since \mathbf{L} is constant, compute it at t = 0: \mathbf{r}(0) = \langle 1, 1, 0 \rangle, \mathbf{r}'(0) = \langle 0, 1, 2 \rangle, and \mathbf{L} = \langle 1, 1, 0 \rangle \times \langle 0, 1, 2 \rangle = \langle 2, -2, 1 \rangle. The curve lies in the plane 2x - 2y + z = 0. Check: 2\cos t - 2(\cos t + \sin t) + 2\sin t = 0 for every t.

Example 6

Integrals over one turn of the slide

For Jin’s slide of Example 2 over its first turn, 0 \le t \le 2\pi: (a) compute \int_0^{2\pi} \mathbf{r}'(t)\,dt and compare it with \mathbf{r}(2\pi) - \mathbf{r}(0); (b) show that no c in (0, 2\pi) has \mathbf{r}(2\pi) - \mathbf{r}(0) = 2\pi\,\mathbf{r}'(c); (c) compare \lVert \mathbf{r}(2\pi) - \mathbf{r}(0) \rVert with \int_0^{2\pi} \lVert \mathbf{r}'(t) \rVert\,dt.

Solution

(a) Component by component,

\begin{aligned} \int_0^{2\pi} \mathbf{r}'(t)\,dt &= \Big\langle \big[4\cos t\big]_0^{2\pi}, \big[4\sin t\big]_0^{2\pi}, -2\pi \Big\rangle \\ &= \langle 0, 0, -2\pi \rangle. \end{aligned}

This is \mathbf{r}(2\pi) - \mathbf{r}(0), as Theorem 4(a) says: one turn takes the rider 2\pi \approx 6.28 meters straight down.

(b) For \langle 0, 0, -2\pi \rangle to equal 2\pi\,\mathbf{r}'(c) we need -4\sin c = 0 and 4\cos c = 0 at once, impossible since \sin^2 c + \cos^2 c = 1. Each component has its own mean value point (c = \pi for x, \displaystyle c = \frac{\pi}{2} for y), but no single c serves all three. The mean value theorem fails for vector functions.

(c) \lVert \mathbf{r}(2\pi) - \mathbf{r}(0) \rVert = 2\pi, while \displaystyle \int_0^{2\pi} \sqrt{17}\,dt = 2\pi\sqrt{17} \approx 25.91. Theorem 4(c), applied to \mathbf{r}', holds with room to spare. The right side is the length of one turn, the subject of Arc length and curvature.

Boundaries

  • There is no mean value theorem (Example 6). What survives is \lVert \mathbf{r}(b) - \mathbf{r}(a) \rVert \le (b - a)\max \lVert \mathbf{r}' \rVert (Exercise 17), and that \mathbf{r}' = \mathbf{0} on an interval forces \mathbf{r} constant (Exercise 16).
  • Differentiable components do not give a smooth curve. \langle t^2, t^3, t^2 \rangle has components with derivatives of every order, yet \mathbf{r}'(0) = \mathbf{0} and the curve has a cusp at the origin. Smoothness belongs to the parametrization: \langle t^3, t^3, t^3 \rangle traces a straight line but is not smooth at t = 0.
  • Theorem 2(d) holds only in \mathbb{R}^3, and only with the order of each factor kept, since \mathbf{v} \times \mathbf{u} = -(\mathbf{u} \times \mathbf{v}).

Where this goes next

Arc length and curvature integrates \lVert \mathbf{r}' \rVert to measure length and differentiates the unit tangent to measure bending. Velocity and acceleration in space reads \mathbf{r}' and \mathbf{r}'' as velocity and acceleration and splits the acceleration along and across the motion, using Theorems 2 and 3. Line integrals integrate a function or a force along a parametrized curve, and Theorem 2(e) is why the answer does not depend on the speed at which the curve is traced.

Exercises

Domains, limits and tangent lines

  1. Find the domain of \displaystyle \mathbf{r}(t) = \left\langle \sqrt{t + 3}, \frac{1}{t^2 - 4}, \ln(5 - t) \right\rangle.
  2. Find \displaystyle \lim_{t \to 1} \left\langle \frac{t^3 - 1}{t - 1}, \frac{\ln t}{t - 1}, \cos \pi t \right\rangle.
  3. Show that \mathbf{r}(t) = \langle t\cos t, t\sin t, t \rangle lies on the cone z^2 = x^2 + y^2, and find its tangent line at \displaystyle t = \frac{\pi}{2}.
  4. The cylinder x^2 + y^2 = 1 meets the plane x + y + z = 2 in a closed curve. Parametrize it and find its highest and lowest points.
  5. Jin’s second design is \mathbf{r}(t) = \langle 3\cos t, 3\sin t, 10 - t \rangle, 0 \le t \le 3\pi. Find the angle at which it descends, compare it with Example 2, and find the tangent line at the bottom end, t = 3\pi.

Derivative rules

  1. For \mathbf{u}(t) = \langle e^t, t, 1 \rangle and \mathbf{v}(t) = \langle \cos t, \sin t, t \rangle, find (\mathbf{u} \times \mathbf{v})'(0) by Theorem 2(d).
  2. For \mathbf{r}(t) = \langle t, t^2, t^3 \rangle let f(t) = \mathbf{r} \cdot (\mathbf{r}' \times \mathbf{r}''). Find f(t), then find f'(t) twice: by differentiating your formula, and as \mathbf{r} \cdot (\mathbf{r}' \times \mathbf{r}''') (Exercise 15 shows why this works).
  3. Elena rides the slide of Exercise 5. After \tau seconds she has turned through the angle \displaystyle t = \varphi(\tau) = \frac{\pi\tau^2}{50}. Find her velocity \displaystyle \frac{d}{d\tau}\mathbf{r}(\varphi(\tau)) and her speed at \tau = 5, and the time at which she reaches the bottom.
  4. For \mathbf{r}(t) = \langle e^t\cos t, e^t\sin t, e^t \rangle, find \displaystyle \frac{d}{dt}\lVert \mathbf{r}(t) \rVert using \mathbf{r} \cdot \mathbf{r}', and show that the angle between \mathbf{r}(t) and \mathbf{r}'(t) is the same for every t. Find it.

Integrals

  1. Evaluate \displaystyle \int_0^1 \left\langle 2te^{t^2}, \frac{1}{1 + t}, \pi\sin \pi t \right\rangle dt.
  2. Find \mathbf{r}(t) if \mathbf{r}'(t) = \langle 3t^2, \sin t, e^{-t} \rangle and \mathbf{r}(0) = \langle 1, 0, 2 \rangle.
  3. For \mathbf{r}(t) = \langle \cos t, \sin t, t \rangle, evaluate \int_0^{\pi} \mathbf{r}(t) \times \mathbf{r}'(t)\,dt.

Smoothness and the mean value theorem

  1. Which of these are smooth on (-\infty, \infty)? (a) \langle t^3 - 1, 2t^3, t^3 + 1 \rangle; (b) \langle t^2, t^3 - 3t, t \rangle; (c) \langle t^2 - 2t, t^3 - 3t, (t - 1)^3 \rangle.
  2. For \mathbf{r}(t) = \langle t^2, t^3, 2t \rangle on [0, 1], show that no c in (0, 1) satisfies \mathbf{r}(1) - \mathbf{r}(0) = \mathbf{r}'(c).

Proofs

  1. For differentiable \mathbf{u}, \mathbf{v}, \mathbf{w}, prove that the derivative of \mathbf{u} \cdot (\mathbf{v} \times \mathbf{w}) is the sum of \mathbf{u}' \cdot (\mathbf{v} \times \mathbf{w}), \mathbf{u} \cdot (\mathbf{v}' \times \mathbf{w}) and \mathbf{u} \cdot (\mathbf{v} \times \mathbf{w}'). Then explain why f' = \mathbf{r} \cdot (\mathbf{r}' \times \mathbf{r}''') in Exercise 7.
  2. Prove that if \mathbf{r}'(t) = \mathbf{0} for every t in an interval I, then \mathbf{r} is constant on I.
  3. Let \mathbf{r}' be continuous on [a, b]. Prove that \lVert \mathbf{r}'(t) \rVert has a largest value M on [a, b] and that \lVert \mathbf{r}(b) - \mathbf{r}(a) \rVert \le (b - a)M.
Answers
  1. [-3, -2) \cup (-2, 2) \cup (2, 5). The root needs t \ge -3, the fraction t \ne \pm 2, the logarithm t < 5; intersect.
  2. \langle 3, 1, -1 \rangle. For t \ne 1, \displaystyle \frac{t^3 - 1}{t - 1} = t^2 + t + 1 \to 3; \displaystyle \frac{\ln t}{t - 1} = \frac{\ln t - \ln 1}{t - 1} \to 1, the derivative of \ln at 1; \cos \pi = -1.
  3. x^2 + y^2 = t^2(\cos^2 t + \sin^2 t) = t^2 = z^2. Here \mathbf{r}'(t) = \langle \cos t - t\sin t, \sin t + t\cos t, 1 \rangle, so \displaystyle \mathbf{r}\left(\frac{\pi}{2}\right) = \left\langle 0, \frac{\pi}{2}, \frac{\pi}{2} \right\rangle, \displaystyle \mathbf{r}'\left(\frac{\pi}{2}\right) = \left\langle -\frac{\pi}{2}, 1, 1 \right\rangle, and the tangent line is \displaystyle \boldsymbol{\ell}(s) = \left\langle -\frac{\pi}{2}s, \frac{\pi}{2} + s, \frac{\pi}{2} + s \right\rangle.
  4. \mathbf{r}(t) = \langle \cos t, \sin t, 2 - \cos t - \sin t \rangle, 0 \le t \le 2\pi. Since \displaystyle \cos t + \sin t = \sqrt{2}\sin\left(t + \frac{\pi}{4}\right), z is largest, \displaystyle 2 + \sqrt{2}, at \displaystyle t = \frac{5\pi}{4}, the point \displaystyle \left(-\frac{\sqrt{2}}{2}, -\frac{\sqrt{2}}{2}, 2 + \sqrt{2}\right), and smallest, \displaystyle 2 - \sqrt{2}, at \displaystyle t = \frac{\pi}{4}, the point \displaystyle \left(\frac{\sqrt{2}}{2}, \frac{\sqrt{2}}{2}, 2 - \sqrt{2}\right).
  5. \displaystyle \arctan\frac{1}{3} \approx 0.3218 radians, about 18.43°, steeper than 14.04°. \mathbf{r}'(t) = \langle -3\sin t, 3\cos t, -1 \rangle has horizontal part of length 3 and vertical part 1. At t = 3\pi, \mathbf{r} = \langle -3, 0, 10 - 3\pi \rangle and \mathbf{r}' = \langle 0, -3, -1 \rangle, so \boldsymbol{\ell}(s) = \langle -3, -3s, 10 - 3\pi - s \rangle.
  6. \langle -1, -1, 0 \rangle. At 0: \mathbf{u} = \langle 1, 0, 1 \rangle, \mathbf{u}' = \langle 1, 1, 0 \rangle, \mathbf{v} = \langle 1, 0, 0 \rangle, \mathbf{v}' = \langle 0, 1, 1 \rangle. Then \mathbf{u}' \times \mathbf{v} = \langle 0, 0, -1 \rangle and \mathbf{u} \times \mathbf{v}' = \langle -1, -1, 1 \rangle.
  7. f(t) = 2t^3 and f'(t) = 6t^2 both ways. With \mathbf{r}' = \langle 1, 2t, 3t^2 \rangle and \mathbf{r}'' = \langle 0, 2, 6t \rangle, the determinant with rows \mathbf{r}, \mathbf{r}', \mathbf{r}'' is t(12t^2 - 6t^2) - t^2(6t) + t^3(2) = 2t^3. With \mathbf{r}''' = \langle 0, 0, 6 \rangle, expanding along the last row gives 6(2t^2 - t^2) = 6t^2.
  8. Velocity \displaystyle \left\langle -\frac{3\pi}{5}, 0, -\frac{\pi}{5} \right\rangle, speed \displaystyle \frac{\pi\sqrt{10}}{5} \approx 1.99 meters per second, bottom at \displaystyle \tau = 5\sqrt{6} \approx 12.25 seconds. By Theorem 2(e) the velocity is \varphi'(\tau)\,\mathbf{r}'(\varphi(\tau)); at \tau = 5, \displaystyle \varphi = \frac{\pi}{2}, \displaystyle \varphi' = \frac{2\pi\tau}{50} = \frac{\pi}{5} and \displaystyle \mathbf{r}'\left(\frac{\pi}{2}\right) = \langle -3, 0, -1 \rangle. The bottom is \displaystyle \frac{\pi\tau^2}{50} = 3\pi, so \tau^2 = 150.
  9. \displaystyle \frac{d}{dt}\lVert \mathbf{r} \rVert = \sqrt{2}\,e^t; the angle is \displaystyle \arccos\frac{\sqrt{6}}{3} \approx 0.6155 radians, about 35.26°. \mathbf{r}' = e^t\langle \cos t - \sin t, \sin t + \cos t, 1 \rangle, so \mathbf{r} \cdot \mathbf{r}' = 2e^{2t}, \displaystyle \lVert \mathbf{r} \rVert = \sqrt{2}\,e^t and \displaystyle \lVert \mathbf{r}' \rVert = \sqrt{3}\,e^t. Then \displaystyle \frac{2e^{2t}}{\sqrt{2}\,e^t} = \sqrt{2}\,e^t and \displaystyle \cos\theta = \frac{2e^{2t}}{\sqrt{6}\,e^{2t}} = \frac{\sqrt{6}}{3}, free of t.
  10. \langle e - 1, \ln 2, 2 \rangle. Antiderivatives e^{t^2}, \ln(1 + t) and -\cos \pi t.
  11. \mathbf{r}(t) = \langle t^3 + 1, 1 - \cos t, 3 - e^{-t} \rangle. An antiderivative is \langle t^3, -\cos t, -e^{-t} \rangle, equal to \langle 0, -1, -1 \rangle at 0, so add \langle 1, 1, 3 \rangle.
  12. \langle 4, -\pi, \pi \rangle. \mathbf{r} \times \mathbf{r}' = \langle \sin t - t\cos t, -\cos t - t\sin t, 1 \rangle, and over [0, \pi]: \int \sin t = 2, \int t\cos t = \big[t\sin t + \cos t\big]_0^{\pi} = -2, \int \cos t = 0, \int t\sin t = \big[\sin t - t\cos t\big]_0^{\pi} = \pi.
  13. (a) No: \mathbf{r}'(t) = 3t^2\langle 1, 2, 1 \rangle is \mathbf{0} at t = 0, though the curve is a line through (-1, 0, 1). (b) Yes: the third component of \mathbf{r}' = \langle 2t, 3t^2 - 3, 1 \rangle is never 0. (c) No: \mathbf{r}' = \langle 2t - 2, 3t^2 - 3, 3(t - 1)^2 \rangle is \mathbf{0} at t = 1.
  14. \mathbf{r}(1) - \mathbf{r}(0) = \langle 1, 1, 2 \rangle and \mathbf{r}'(c) = \langle 2c, 3c^2, 2 \rangle. The first components need \displaystyle c = \frac{1}{2}, the second \displaystyle c = \frac{1}{\sqrt{3}}; these differ, so no c works.
  15. By Theorem 2(c) the derivative is \mathbf{u}' \cdot (\mathbf{v} \times \mathbf{w}) plus \mathbf{u} \cdot (\mathbf{v} \times \mathbf{w})', and by 2(d) (\mathbf{v} \times \mathbf{w})' = \mathbf{v}' \times \mathbf{w} + \mathbf{v} \times \mathbf{w}'; distribute the dot product. With \mathbf{u} = \mathbf{r}, \mathbf{v} = \mathbf{r}', \mathbf{w} = \mathbf{r}'', the first term \mathbf{r}' \cdot (\mathbf{r}' \times \mathbf{r}'') is a triple product with a repeated vector, and the second has \mathbf{r}'' \times \mathbf{r}'' = \mathbf{0}, so only \mathbf{r} \cdot (\mathbf{r}' \times \mathbf{r}''') remains.
  16. By Theorem 1, x' = y' = z' = 0 on I. For s < t in I the mean value theorem (The mean value theorem) gives x(t) - x(s) = x'(c)(t - s) = 0 for some c between them, and likewise for y and z. So \mathbf{r}(t) = \mathbf{r}(s).
  17. \displaystyle \lVert \mathbf{r}' \rVert = \sqrt{x'^2 + y'^2 + z'^2} is a composition of continuous functions, so it is continuous on [a, b] and has a largest value M by the The extreme value theorem. By Theorem 4(a), \mathbf{r}(b) - \mathbf{r}(a) = \int_a^b \mathbf{r}'\,dt. By Theorem 4(c) its length is at most \int_a^b \lVert \mathbf{r}' \rVert\,dt, which is at most \int_a^b M\,dt = (b - a)M.
math.wiki · updated 2026-11-17 · CC-BY-SA-4.0