math.wiki
Calculus › Vectors and space curves › Velocity and acceleration in space
Sign in

Velocity and acceleration in space

A luge sled is pushed along the ice and pushed sideways by the wall of a banked curve at the same time. How does one acceleration vector carry both pushes, the one that changes the speed and the one that bends the path?

Where you are 5 in Vectors and space curves

Why this exists

Vector valued functions and motion in the plane read the derivatives of a position in the plane as velocity and acceleration. By Newton’s second law the force on a body of mass m is m\mathbf{a}, so acceleration is what a rider feels. A luge slider feels two pushes at once: one along the track that changes the speed, and one from the wall of a curve that turns the sled. The components \langle a_x, a_y, a_z \rangle mix them. The frame \mathbf{T}, \mathbf{N} of Arc length and curvature separates them: acceleration has no third part, its part along the motion is the rate of change of speed, and its part across is \kappa times the square of the speed.

Definition

A motion is a vector function \mathbf{r}(t) = \langle x(t), y(t), z(t) \rangle, twice differentiable on an interval I, read as a position at time t. Its velocity is \mathbf{v}(t) = \mathbf{r}'(t), its speed is \lVert \mathbf{v}(t) \rVert, and its acceleration is \mathbf{a}(t) = \mathbf{v}'(t) = \mathbf{r}''(t).

When \mathbf{r} is also smooth (\mathbf{v} continuous and never \mathbf{0}), let s be arc length, so \displaystyle \frac{ds}{dt} = \lVert \mathbf{v} \rVert, and take \mathbf{T}, \kappa and, where \kappa > 0, \mathbf{N} and \mathbf{B} = \mathbf{T} \times \mathbf{N} from Arc length and curvature. The tangential component of acceleration is

a_{\mathbf{T}} = \mathbf{a} \cdot \mathbf{T},

and the normal component is a_{\mathbf{N}} = \mathbf{a} \cdot \mathbf{N} where \kappa > 0, and 0 where \kappa = 0. These are the scalar components \operatorname{comp}_{\mathbf{T}} \mathbf{a} and \operatorname{comp}_{\mathbf{N}} \mathbf{a} of Vector and scalar projection.

Intuition

Think of the speed \displaystyle \frac{ds}{dt} and the direction \mathbf{T} as two dials, with \displaystyle \mathbf{v} = \frac{ds}{dt}\,\mathbf{T}. Turning the speed dial moves \mathbf{v} along \mathbf{T}. Turning the direction dial moves \mathbf{v} along \mathbf{T}', which is perpendicular to \mathbf{T} because \mathbf{T} has constant length, and points along \mathbf{N}. Nothing moves \mathbf{v} along \mathbf{B}, so the acceleration lies in the osculating plane, the plane of the bend.

P
\mathbf{a}
a_{\mathbf{T}}\mathbf{T}
a_{\mathbf{N}}\mathbf{N}
Figure 1The acceleration at P drawn in the osculating plane. Its part along \mathbf{T} changes the speed; its part along \mathbf{N}, toward the inside of the bend, turns the velocity.

In a short time \Delta t the body covers about \displaystyle \frac{ds}{dt}\,\Delta t of path, so its direction turns through about \displaystyle \kappa \frac{ds}{dt}\,\Delta t radians, and a velocity of length \displaystyle \frac{ds}{dt} swings sideways by about \displaystyle \kappa \left(\frac{ds}{dt}\right)^2 \Delta t. Speed enters twice, so on the same curve twice the speed needs four times the sideways push. The curvature belongs to the track; the speed, and with it a_{\mathbf{N}}, belongs to the slider.

Results

Theorem 1 (splitting the acceleration). Let \mathbf{r} be a smooth, twice differentiable motion on I. Then \displaystyle \frac{ds}{dt} is differentiable, \displaystyle \mathbf{v} = \frac{ds}{dt}\,\mathbf{T}, and

\mathbf{a} = \frac{d^2 s}{dt^2}\,\mathbf{T} + \kappa \left(\frac{ds}{dt}\right)^2 \mathbf{N},

the last term read as \mathbf{0} where \kappa = 0. Hence \displaystyle a_{\mathbf{T}} = \frac{d^2 s}{dt^2}, \displaystyle a_{\mathbf{N}} = \kappa \left(\frac{ds}{dt}\right)^2 \ge 0, and \mathbf{a} \cdot \mathbf{B} = 0 wherever \kappa > 0.

Theorem 2 (the components from \mathbf{v} and \mathbf{a}). Under the same hypotheses,

\begin{aligned} a_{\mathbf{T}} &= \frac{\mathbf{v} \cdot \mathbf{a}}{\lVert \mathbf{v} \rVert}, \\ a_{\mathbf{N}} &= \frac{\lVert \mathbf{v} \times \mathbf{a} \rVert}{\lVert \mathbf{v} \rVert} \\ &= \sqrt{\lVert \mathbf{a} \rVert^2 - a_{\mathbf{T}}^2}, \end{aligned}

and where \kappa > 0, \mathbf{N} = \dfrac{\mathbf{a} - a_{\mathbf{T}}\mathbf{T}}{a_{\mathbf{N}}} and \kappa = \dfrac{a_{\mathbf{N}}}{\lVert \mathbf{v} \rVert^2}.

Theorem 3 (speeding up and slowing down). Let \mathbf{r} be twice differentiable on I.

  • (a) Where \mathbf{v} \ne \mathbf{0}, the speed has derivative \displaystyle \frac{\mathbf{v} \cdot \mathbf{a}}{\lVert \mathbf{v} \rVert}. If \mathbf{v} \cdot \mathbf{a} > 0 on an interval, the speed is strictly increasing there; if \mathbf{v} \cdot \mathbf{a} < 0, strictly decreasing.
  • (b) The speed is constant on I if and only if \mathbf{v} \cdot \mathbf{a} = 0 at every t \in I.

Theorem 4 (motion from acceleration). Let \mathbf{a} be continuous on I, t_0 \in I, and \mathbf{r}_0, \mathbf{v}_0 any vectors. Exactly one twice differentiable \mathbf{r} on I has \mathbf{r}'' = \mathbf{a}, \mathbf{r}'(t_0) = \mathbf{v}_0 and \mathbf{r}(t_0) = \mathbf{r}_0, namely

\begin{aligned} \mathbf{v}(t) &= \mathbf{v}_0 + \int_{t_0}^t \mathbf{a}(u)\,du, \\ \mathbf{r}(t) &= \mathbf{r}_0 + \int_{t_0}^t \mathbf{v}(u)\,du. \end{aligned}

For a constant acceleration \mathbf{a}_0 this is \displaystyle \mathbf{r} = \mathbf{r}_0 + \tau\mathbf{v}_0 + \frac{1}{2}\tau^2\mathbf{a}_0 with \tau = t - t_0. If \mathbf{v}_0 \times \mathbf{a}_0 \ne \mathbf{0}, the path lies in the plane through \mathbf{r}_0 with normal \mathbf{v}_0 \times \mathbf{a}_0 and is an arc of a parabola there with axis parallel to \mathbf{a}_0; otherwise it lies on one line.

Why it is true

Theorem 1. By Theorem 2 of Arc length and curvature, \displaystyle \frac{ds}{dt} = \lVert \mathbf{v} \rVert = \sqrt{\mathbf{v} \cdot \mathbf{v}}. The dot product rule (Theorem 2(c) of Vector-valued functions) makes \mathbf{v} \cdot \mathbf{v} differentiable, and it is positive, so \displaystyle \frac{ds}{dt} and \mathbf{T} = \mathbf{v} / \lVert \mathbf{v} \rVert are differentiable. From \displaystyle \mathbf{v} = \frac{ds}{dt}\,\mathbf{T} the product rule (Theorem 2(b) there) gives

\mathbf{a} = \frac{d^2 s}{dt^2}\,\mathbf{T} + \frac{ds}{dt}\,\mathbf{T}'.

By Theorem 3(a) of Arc length and curvature, \displaystyle \lVert \mathbf{T}' \rVert = \kappa \frac{ds}{dt}. Where \kappa > 0, \displaystyle \mathbf{T}' = \lVert \mathbf{T}' \rVert\,\mathbf{N} = \kappa \frac{ds}{dt}\,\mathbf{N}; where \kappa = 0, \mathbf{T}' = \mathbf{0}. Dotting the result with the mutually perpendicular unit vectors \mathbf{T}, \mathbf{N}, \mathbf{B} gives the rest.

Theorem 2. First, \displaystyle a_{\mathbf{T}} = \mathbf{a} \cdot \frac{\mathbf{v}}{\lVert \mathbf{v} \rVert}. Since \mathbf{T} \times \mathbf{T} = \mathbf{0}, Theorem 1 gives \displaystyle \mathbf{v} \times \mathbf{a} = \frac{ds}{dt}\,a_{\mathbf{N}}\,\mathbf{B} where \kappa > 0 and \mathbf{0} where \kappa = 0; as \lVert \mathbf{B} \rVert = 1 and a_{\mathbf{N}} \ge 0, \lVert \mathbf{v} \times \mathbf{a} \rVert = \lVert \mathbf{v} \rVert\,a_{\mathbf{N}}. As \mathbf{T} and \mathbf{N} are perpendicular unit vectors, \lVert \mathbf{a} \rVert^2 = a_{\mathbf{T}}^2 + a_{\mathbf{N}}^2, and a_{\mathbf{N}} \ge 0 picks the positive root. Where \kappa > 0, a_{\mathbf{N}} > 0, and solving Theorem 1 for \mathbf{N} and \kappa gives the last two formulas.

Theorem 3. (a) By the dot product rule (\lVert \mathbf{v} \rVert^2)' = 2\,\mathbf{v} \cdot \mathbf{a}; where \mathbf{v} \ne \mathbf{0} the chain rule for the square root gives the derivative. The sign of a derivative on an interval decides strict monotonicity, by the mean value theorem (The mean value theorem). (b) is Theorem 3 of Vector-valued functions applied to \mathbf{v}.

Theorem 4. By the fundamental theorem of calculus in each component (The second fundamental theorem of calculus, Theorem 4(a) of Vector-valued functions), the displayed \mathbf{v} has \mathbf{v}' = \mathbf{a} and is continuous, so the displayed \mathbf{r} has \mathbf{r}' = \mathbf{v}, \mathbf{r}'' = \mathbf{a} and the given values at t_0. If \tilde{\mathbf{r}} is another, \mathbf{w} = \tilde{\mathbf{r}} - \mathbf{r} has \mathbf{w}'' = \mathbf{0}. A vector function with zero derivative on an interval is constant, so \mathbf{w}' = \mathbf{w}'(t_0) = \mathbf{0}, and then \mathbf{w} = \mathbf{w}(t_0) = \mathbf{0}.

For constant \mathbf{a}_0 the integrals are \tau\mathbf{a}_0 and \displaystyle \tau\mathbf{v}_0 + \frac{1}{2}\tau^2\mathbf{a}_0. Let \mathbf{n} = \mathbf{v}_0 \times \mathbf{a}_0 \ne \mathbf{0}; it is perpendicular to \mathbf{v}_0 and \mathbf{a}_0 (The cross product and the triple product), so \mathbf{n} \cdot (\mathbf{r} - \mathbf{r}_0) = 0. Let \displaystyle \mathbf{e}_1 = \frac{\mathbf{a}_0}{\lVert \mathbf{a}_0 \rVert} and \mathbf{p} = \mathbf{v}_0 - (\mathbf{v}_0 \cdot \mathbf{e}_1)\mathbf{e}_1, which is not \mathbf{0} because \mathbf{v}_0 is not a multiple of \mathbf{a}_0, and \displaystyle \mathbf{e}_2 = \frac{\mathbf{p}}{\lVert \mathbf{p} \rVert}. In the perpendicular unit directions \mathbf{e}_2, \mathbf{e}_1 of the plane, \mathbf{r} - \mathbf{r}_0 has coordinates X = \tau\lVert \mathbf{p} \rVert and \displaystyle Y = \tau\,b + \frac{1}{2}\tau^2\lVert \mathbf{a}_0 \rVert, where b = \mathbf{v}_0 \cdot \mathbf{e}_1. Substituting \displaystyle \tau = \frac{X}{\lVert \mathbf{p} \rVert},

Y = \frac{b}{\lVert \mathbf{p} \rVert}X + \frac{\lVert \mathbf{a}_0 \rVert}{2\lVert \mathbf{p} \rVert^2}X^2,

a parabola with axis along \mathbf{e}_1. If \mathbf{n} = \mathbf{0}, then \mathbf{a}_0 = \mathbf{0} or \mathbf{v}_0 = c\,\mathbf{a}_0, and \mathbf{r} - \mathbf{r}_0 is a multiple of \mathbf{v}_0 or of \mathbf{a}_0 for every t.

Worked examples

Example 1

The splitting at one instant

For \displaystyle \mathbf{r}(t) = \left\langle 2t, \frac{t^2}{2}, \frac{2}{3}t^3 \right\rangle, find a_{\mathbf{T}}, a_{\mathbf{N}}, \mathbf{N}, \mathbf{B} and \kappa at t = 1, and check that \mathbf{a} = a_{\mathbf{T}}\mathbf{T} + a_{\mathbf{N}}\mathbf{N}.

Solution

Differentiating each component, \mathbf{v}(t) = \langle 2, t, 2t^2 \rangle and \mathbf{a}(t) = \langle 0, 1, 4t \rangle, so \mathbf{v}(1) = \langle 2, 1, 2 \rangle, of length 3, and \mathbf{a}(1) = \langle 0, 1, 4 \rangle, with \lVert \mathbf{a}(1) \rVert^2 = 17. By Theorem 2,

\begin{gathered} a_{\mathbf{T}} = \frac{0 + 1 + 8}{3} = 3, \\ a_{\mathbf{N}} = \sqrt{17 - 9} = 2\sqrt{2}. \end{gathered}

The cross product agrees: \mathbf{v} \times \mathbf{a} = \langle 2, -8, 2 \rangle has length \displaystyle 6\sqrt{2}, and \displaystyle \frac{6\sqrt{2}}{3} = 2\sqrt{2}. Since \displaystyle \mathbf{T} = \frac{1}{3}\langle 2, 1, 2 \rangle, a_{\mathbf{T}}\mathbf{T} = \langle 2, 1, 2 \rangle and

\begin{aligned} \mathbf{N} &= \frac{\langle 0, 1, 4 \rangle - \langle 2, 1, 2 \rangle}{2\sqrt{2}} \\ &= \frac{1}{\sqrt{2}}\langle -1, 0, 1 \rangle, \end{aligned}

perpendicular to \mathbf{T}, as it must be: -2 + 0 + 2 = 0. Then \displaystyle \mathbf{B} = \mathbf{T} \times \mathbf{N} = \frac{1}{3\sqrt{2}}\langle 1, -4, 1 \rangle, a positive multiple of \mathbf{v} \times \mathbf{a}, and \displaystyle \kappa = \frac{a_{\mathbf{N}}}{\lVert \mathbf{v} \rVert^2} = \frac{2\sqrt{2}}{9}. Check: \displaystyle 3\mathbf{T} + 2\sqrt{2}\,\mathbf{N} = \langle 2, 1, 2 \rangle + \langle -2, 0, 2 \rangle = \langle 0, 1, 4 \rangle.

Example 2

A spiral on a cone

The motion \mathbf{r}(t) = \langle e^t\cos t, e^t\sin t, 2e^t \rangle runs on the cone z^2 = 4(x^2 + y^2). Find the speed, a_{\mathbf{T}}, a_{\mathbf{N}} and \kappa for every t, and show that the angle between \mathbf{v} and \mathbf{a} never changes.

Solution

By the product rule in each component,

\begin{aligned} \mathbf{v} &= e^t\langle \cos t - \sin t, \sin t + \cos t, 2 \rangle, \\ \mathbf{a} &= e^t\langle -2\sin t, 2\cos t, 2 \rangle. \end{aligned}

Since (\cos t - \sin t)^2 + (\sin t + \cos t)^2 = 2, the speed is \displaystyle \sqrt{6}\,e^t, and \displaystyle \lVert \mathbf{a} \rVert = 2\sqrt{2}\,e^t. In \mathbf{v} \cdot \mathbf{a} the terms \mp 2\sin t\cos t cancel, leaving e^{2t}(2\sin^2 t + 2\cos^2 t + 4) = 6e^{2t}. Theorem 2 gives \displaystyle a_{\mathbf{T}} = \frac{6e^{2t}}{\sqrt{6}\,e^t} = \sqrt{6}\,e^t, and \displaystyle a_{\mathbf{N}} = \sqrt{8e^{2t} - 6e^{2t}} = \sqrt{2}\,e^t. So

\kappa = \frac{\sqrt{2}\,e^t}{6e^{2t}} = \frac{\sqrt{2}}{6}e^{-t}.

The cosine of the angle between \mathbf{v} and \mathbf{a} is \displaystyle \frac{a_{\mathbf{T}}}{\lVert \mathbf{a} \rVert} = \frac{\sqrt{6}}{2\sqrt{2}} = \frac{\sqrt{3}}{2} for every t, an angle of exactly 30^\circ.

Example 3

The slider or the track

Lucas’s sled carries a sensor that reads velocity and acceleration in meters and seconds. As he passes a point P of a banked curve on the luge track it reads \mathbf{v} = \langle 24, 0, -7 \rangle and \mathbf{a} = \langle 0.96, 16, -0.28 \rangle. Find his tangential and normal components and the curvature and radius of curvature of the track at P. Naomi passes P on the same path at 30 meters per second, gaining speed at 0.5 meters per second squared. Find her acceleration vector.

Solution

Lucas’s speed is \displaystyle \sqrt{576 + 49} = 25. By Theorem 2, \displaystyle a_{\mathbf{T}} = \frac{23.04 + 1.96}{25} = 1, and since \lVert \mathbf{a} \rVert^2 = 257, \displaystyle a_{\mathbf{N}} = \sqrt{257 - 1} = 16: he gains 1 meter per second each second, and the wall pushes him sideways at 16 meters per second squared. Then \displaystyle \kappa = \frac{16}{625} = 0.0256 per meter, a radius of curvature of 39.0625 \approx 39.1 meters.

The curvature, \mathbf{T} and \mathbf{N} belong to the path, so Naomi has the same \displaystyle \mathbf{T} = \frac{1}{25}\langle 24, 0, -7 \rangle and the same \displaystyle \mathbf{N} = \frac{\mathbf{a} - \mathbf{T}}{16} = \langle 0, 1, 0 \rangle. Her normal component is \kappa \cdot 30^2 = 23.04 and her tangential one is 0.5, so by Theorem 1

\begin{gathered} \mathbf{a} = 0.5\,\mathbf{T} + 23.04\,\mathbf{N} \\ = \langle 0.48, 23.04, -0.14 \rangle, \end{gathered}

of length about 23.05 meters per second squared. Her speed is 1.2 times his, so her sideways push is 1.2^2 = 1.44 times his.

Example 4

A snowball in a crosswind

From the start platform, 64 feet above the ground, Naomi tosses a snowball horizontally with velocity \langle 6, 8, 0 \rangle feet per second. Gravity gives 32 feet per second squared downward, and a steady crosswind is modeled as a further constant 2 feet per second squared in the direction of -\mathbf{j}; ignore drag. With the origin on the ground below her hand, find the path, where it lands, and the plane it stays in. Is the snowball speeding up as it leaves her hand?

Solution

The acceleration is the constant \mathbf{a}_0 = \langle 0, -2, -32 \rangle, with \mathbf{r}_0 = \langle 0, 0, 64 \rangle and \mathbf{v}_0 = \langle 6, 8, 0 \rangle at t = 0. By Theorem 4,

\mathbf{r}(t) = \langle 6t, 8t - t^2, 64 - 16t^2 \rangle.

It lands when 64 - 16t^2 = 0, at t = 2, at the point (12, 12, 0). As \mathbf{v}_0 and \mathbf{a}_0 are not parallel, the path is an arc of a parabola in the plane through (0, 0, 64) with normal

\begin{gathered} \mathbf{v}_0 \times \mathbf{a}_0 = \langle -256, 192, -12 \rangle \\ = -4\langle 64, -48, 3 \rangle, \end{gathered}

that is, 64x - 48y + 3z = 192; the landing point checks, 768 - 576 = 192. At release \mathbf{v}_0 \cdot \mathbf{a}_0 = -16, so \displaystyle a_{\mathbf{T}} = -\frac{16}{10} = -1.6: the snowball is slowing, since gravity is perpendicular to \mathbf{v}_0 while the wind brakes the motion along \mathbf{j}. In general \mathbf{v}(t) = \langle 6, 8 - 2t, -32t \rangle and \mathbf{v} \cdot \mathbf{a}_0 = 1028t - 16, negative for \displaystyle t < \frac{4}{257} \approx 0.016 and positive after. By Theorem 3 the speed falls for 0.016 seconds, then rises to \displaystyle \lVert \langle 6, 4, -64 \rangle \rVert = 2\sqrt{1037} \approx 64.4 feet per second at the ground.

Example 5

A proof: equal areas in equal times

A motion \mathbf{r} never passes through the origin, and its acceleration is central: \mathbf{a}(t) is a multiple of \mathbf{r}(t) for every t. Prove that \mathbf{L} = \mathbf{r} \times \mathbf{v} is constant and that, if \mathbf{L} \ne \mathbf{0}, the segment from the origin to \mathbf{r}(t) sweeps area at the constant rate \displaystyle \frac{1}{2}\lVert \mathbf{L} \rVert. Check it on \mathbf{r}(t) = \langle 4\cos t, 2\sin t, 0 \rangle.

Solution

By the cross product rule, \mathbf{L}' = \mathbf{v} \times \mathbf{v} + \mathbf{r} \times \mathbf{a} = \mathbf{0}, since a vector crossed with a multiple of itself is \mathbf{0}. So \mathbf{L} is constant and, as in the proof example of Vector-valued functions, the motion stays in the plane through the origin with normal \mathbf{L}. Choose right-handed axes with \mathbf{L} = \lVert \mathbf{L} \rVert\,\mathbf{k} and write \mathbf{r} = \rho\,\mathbf{u} with \rho = \lVert \mathbf{r} \rVert > 0 and \mathbf{u} = \langle \cos\theta, \sin\theta, 0 \rangle, taking the polar angle \theta(t) differentiable (possible because \mathbf{r} is differentiable and never \mathbf{0}). With \mathbf{w} = \langle -\sin\theta, \cos\theta, 0 \rangle, the chain rule gives \mathbf{u}' = \theta'\mathbf{w}, so \mathbf{v} = \rho'\mathbf{u} + \rho\theta'\mathbf{w} and, as \mathbf{u} \times \mathbf{w} = \mathbf{k},

\mathbf{L} = \rho\,\mathbf{u} \times (\rho'\mathbf{u} + \rho\theta'\mathbf{w}) = \rho^2\theta'\,\mathbf{k}.

So \rho^2\theta' = \lVert \mathbf{L} \rVert > 0 and \theta is increasing. By the polar area formula of Area in polar coordinates and the substitution \theta = \theta(t), the area swept from t_0 to t_1, while \theta turns at most once around, is

\begin{aligned} \frac{1}{2}\int_{\theta(t_0)}^{\theta(t_1)} \rho^2\,d\theta &= \frac{1}{2}\int_{t_0}^{t_1} \rho^2\theta'\,dt \\ &= \frac{\lVert \mathbf{L} \rVert}{2}(t_1 - t_0). \end{aligned}

For the ellipse, \mathbf{a} = -\mathbf{r} is central, the pull of a spring anchored at the origin, and \mathbf{L} = \langle 0, 0, 8 \rangle, since 4 \cdot 2\cos^2 t + 4 \cdot 2\sin^2 t = 8. Area is swept at rate 4, and one trip, 0 \le t \le 2\pi, sweeps 8\pi, the area \pi \cdot 4 \cdot 2 of the ellipse. For a planet pulled by the sun this is Kepler’s second law.

x y O
Figure 2Example 5. The motion \langle 4\cos t, 2\sin t, 0 \rangle sweeps the sector over 0.2 \le t \le 1.2 and the sector over 2.7 \le t \le 3.7 in equal times, and both have area 4.
Example 6

A proof: a magnetic field bends the path into a helix

A charged particle in a uniform magnetic field along \mathbf{k} moves with \mathbf{a} = \omega\,(\mathbf{v} \times \mathbf{k}) for a constant \omega \ne 0. Prove that its speed and \mathbf{v} \cdot \mathbf{k} are constant. Then show that the only such motion with \mathbf{r}(0) = \mathbf{0} and \mathbf{v}(0) = \langle c, 0, h \rangle, c > 0, is

\begin{aligned} \mathbf{r}(t) = \Big\langle &\tfrac{c}{\omega}\sin\omega t, \\ &\tfrac{c}{\omega}(\cos\omega t - 1), ht \Big\rangle, \end{aligned}

and find its curvature.

Solution

Here \mathbf{v} \cdot \mathbf{a} = \omega\,\mathbf{v} \cdot (\mathbf{v} \times \mathbf{k}) = 0, a triple product with a repeated vector (The cross product and the triple product), so the speed is constant by Theorem 3(b). Likewise (\mathbf{v} \cdot \mathbf{k})' = \omega\,\mathbf{k} \cdot (\mathbf{v} \times \mathbf{k}) = 0.

For any \mathbf{v} = \langle v_1, v_2, v_3 \rangle, \mathbf{v} \times \mathbf{k} = \langle v_2, -v_1, 0 \rangle. The given \mathbf{r} has \mathbf{r}(0) = \mathbf{0} and

\begin{aligned} \mathbf{v} &= \langle c\cos\omega t, -c\sin\omega t, h \rangle, \\ \mathbf{a} &= -c\omega\langle \sin\omega t, \cos\omega t, 0 \rangle, \end{aligned}

so \mathbf{v}(0) = \langle c, 0, h \rangle and \omega(\mathbf{v} \times \mathbf{k}) = \mathbf{a}. Theorem 4 does not give uniqueness, because this \mathbf{a} depends on \mathbf{v}. Instead, let \tilde{\mathbf{r}} be another solution with the same start and \mathbf{d} = \tilde{\mathbf{v}} - \mathbf{v}. By distributivity \mathbf{d}' = \omega(\mathbf{d} \times \mathbf{k}), so \mathbf{d} \cdot \mathbf{d}' = 0 and \lVert \mathbf{d} \rVert is constant (Theorem 3 of Vector-valued functions), equal to \lVert \mathbf{d}(0) \rVert = 0. So \tilde{\mathbf{r}} - \mathbf{r} has derivative \mathbf{0} and starts at \mathbf{0}: \tilde{\mathbf{r}} = \mathbf{r}.

The path satisfies \displaystyle x^2 + \left(y + \frac{c}{\omega}\right)^2 = \frac{c^2}{\omega^2}: it winds around the vertical line through \displaystyle \left(0, -\frac{c}{\omega}, 0\right) at radius \displaystyle \frac{c}{\lvert \omega \rvert} while climbing at rate h, a helix (a circle when h = 0). Its speed is \displaystyle \sqrt{c^2 + h^2} and a_{\mathbf{T}} = 0, so a_{\mathbf{N}} = \lVert \mathbf{a} \rVert = c\lvert \omega \rvert and, by Theorem 2,

\kappa = \frac{c\lvert \omega \rvert}{c^2 + h^2}.

Boundaries

  • The splitting needs \mathbf{v} \ne \mathbf{0}. A ball thrown straight up, \mathbf{r}(t) = \langle 0, 0, 48t - 16t^2 \rangle in feet, has \mathbf{v}(1.5) = \mathbf{0}: \mathbf{T} jumps from \mathbf{k} to -\mathbf{k}, and \mathbf{a} = -32\mathbf{k} has no tangential and normal parts there.
  • Where \kappa = 0 there is no \mathbf{N}. At an inflection of a plane path \mathbf{N} jumps to the other side while a_{\mathbf{N}} \ge 0 keeps its sign; a signed normal component exists only in the plane, measured against a fixed turn of \mathbf{T} such as the J\mathbf{T} of Exercise 16 in Arc length and curvature.
  • Theorem 4 needs the acceleration as a known function of time. Drag depends on the velocity, as the magnetic force of Example 6 does; then the motion solves a differential equation, whose general existence and uniqueness belong to a course in differential equations.

Where this goes next

Only the tangential part does work. If a force \mathbf{F} = m\mathbf{a} moves a body, \mathbf{F} \cdot \mathbf{v} = m\,a_{\mathbf{T}}\lVert \mathbf{v} \rVert, the derivative of \displaystyle \frac{1}{2}m\lVert \mathbf{v} \rVert^2; the normal part turns the body and never changes its energy. Line integrals integrate \mathbf{F} \cdot \mathbf{v} along a path, extending Work as an integral, and Conservative fields and potential functions turn this into conservation of energy. How fast \mathbf{B} turns, the torsion, belongs to Differential Geometry.

Exercises

Velocity, speed and the splitting

  1. For \mathbf{r}(t) = \langle t, e^t, e^{-t} \rangle, find \mathbf{v}(0), \mathbf{a}(0), the speed, a_{\mathbf{T}}, a_{\mathbf{N}}, \kappa and \mathbf{N} at t = 0.
  2. For \mathbf{r}(t) = \langle \cos 2t, \sin 2t, t^2 \rangle, find the speed, a_{\mathbf{T}} and a_{\mathbf{N}} as functions of t, and check that a_{\mathbf{T}} is the derivative of the speed.
  3. At one instant a particle has \mathbf{v} = \langle 2, -1, 2 \rangle and \mathbf{a} = \langle 1, 4, 7 \rangle. Find a_{\mathbf{T}}, a_{\mathbf{N}}, \mathbf{N} and the curvature of its path. Is it speeding up or slowing down?
  4. Lucas rides a flat circular turn of radius 40 meters. At one instant his speed is 12 meters per second and is increasing at 1.5 meters per second squared. Find the length of his acceleration and the angle between his acceleration and his velocity.
  5. For \displaystyle \mathbf{r}(t) = \left\langle t\cos t, t\sin t, \frac{t^2}{2} \right\rangle, find a_{\mathbf{T}} as a function of t, and a_{\mathbf{N}} and \kappa at t = 0.
  6. For \mathbf{r}(t) = \langle t, 1 - \cos t, \sin t - t \rangle, find \kappa and \mathbf{N} at \displaystyle t = \frac{\pi}{2}.

The slider and the track

  1. The tightest curve of a luge track has curvature 0.032 per meter, and the normal acceleration of a sled must stay at most 40 meters per second squared. What is the fastest a slider may take that curve?
  2. Lucas passes a point Q of the track at 20 meters per second with a_{\mathbf{N}} = 10 meters per second squared. Naomi passes Q on the same path at 26 meters per second. Find her normal component and the radius of curvature of the track at Q.

From acceleration to motion

  1. Find the motion with \mathbf{a}(t) = \langle 6t, 2, e^{-t} \rangle, \mathbf{v}(0) = \langle 1, 0, -1 \rangle and \mathbf{r}(0) = \langle 0, 3, 1 \rangle, and its position at t = 1.
  2. A snowball leaves the ground at the origin with velocity \langle 12, 5, 40 \rangle feet per second, with no wind, so \mathbf{a} = \langle 0, 0, -32 \rangle. When and where does it land, how high does it rise, what is its smallest speed, and in what plane does it move?
  3. Naomi throws a snowball from the origin on level ground with velocity \langle 10, 0, 16 \rangle feet per second into a crosswind, so \mathbf{a} = \langle 0, 3, -32 \rangle. Find where it lands and the plane of its path.

Proofs

  1. Let \mathbf{r} be a smooth, twice differentiable motion on an interval. Prove that a_{\mathbf{T}} = 0 throughout if and only if the speed is constant, and that a_{\mathbf{N}} = 0 throughout if and only if the motion runs along one line.
  2. Let \mathbf{r} be smooth and twice differentiable, let \varphi be twice differentiable with \varphi' > 0, and let \mathbf{q}(u) = \mathbf{r}(\varphi(u)), the same path at another pace. Prove that a_{\mathbf{N}} of \mathbf{q} at u is \varphi'(u)^2 times a_{\mathbf{N}} of \mathbf{r} at \varphi(u), and express a_{\mathbf{T}} of \mathbf{q} in terms of \mathbf{r}.
  3. A motion has constant acceleration \mathbf{a}_0 \ne \mathbf{0} and \mathbf{v}(0) = \mathbf{v}_0, for all real t. Prove that its smallest speed is \displaystyle \frac{\lVert \mathbf{v}_0 \times \mathbf{a}_0 \rVert}{\lVert \mathbf{a}_0 \rVert}, reached exactly when \mathbf{v} \perp \mathbf{a}_0.
  4. A motion never passes through the origin, has central acceleration, and has \mathbf{r}(t_0) \times \mathbf{v}(t_0) = \mathbf{0} at one time t_0. Prove that it stays on one line through the origin.
  5. A smooth motion has continuous acceleration, constant speed c > 0 and acceleration of constant length A > 0. Prove that its curvature is the constant \displaystyle \frac{A}{c^2}, and that if it also stays in a plane it runs on a circle of radius \displaystyle \frac{c^2}{A}.
Answers
  1. \mathbf{v}(0) = \langle 1, 1, -1 \rangle, \mathbf{a}(0) = \langle 0, 1, 1 \rangle, speed \displaystyle \sqrt{3}, a_{\mathbf{T}} = 0, \displaystyle a_{\mathbf{N}} = \sqrt{2}, \displaystyle \kappa = \frac{\sqrt{2}}{3}, \displaystyle \mathbf{N}(0) = \frac{1}{\sqrt{2}}\langle 0, 1, 1 \rangle. Here \mathbf{v} = \langle 1, e^t, -e^{-t} \rangle and \mathbf{a} = \langle 0, e^t, e^{-t} \rangle; \mathbf{v}(0) \cdot \mathbf{a}(0) = 0, so a_{\mathbf{N}} = \lVert \mathbf{a}(0) \rVert and \displaystyle \mathbf{N} = \frac{\mathbf{a}}{a_{\mathbf{N}}}.
  2. Speed \displaystyle 2\sqrt{1 + t^2}, \displaystyle a_{\mathbf{T}} = \frac{2t}{\sqrt{1 + t^2}}, \displaystyle a_{\mathbf{N}} = 2\sqrt{\frac{5 + 4t^2}{1 + t^2}}. Here \lVert \mathbf{a} \rVert^2 = 16 + 4 = 20 and \mathbf{v} \cdot \mathbf{a} = 4t, so \displaystyle a_{\mathbf{N}}^2 = 20 - \frac{4t^2}{1 + t^2}; and \displaystyle \frac{d}{dt}\,2\sqrt{1 + t^2} = \frac{2t}{\sqrt{1 + t^2}}.
  3. a_{\mathbf{T}} = 4, \displaystyle a_{\mathbf{N}} = 5\sqrt{2}, \displaystyle \mathbf{N} = \frac{1}{15\sqrt{2}}\langle -5, 16, 13 \rangle, \displaystyle \kappa = \frac{5\sqrt{2}}{9}; speeding up, as a_{\mathbf{T}} > 0. \lVert \mathbf{v} \rVert = 3, \mathbf{v} \cdot \mathbf{a} = 12, \lVert \mathbf{a} \rVert^2 = 66, and \displaystyle \mathbf{a} - 4\mathbf{T} = \frac{1}{3}\langle -5, 16, 13 \rangle.
  4. 3.9 meters per second squared, at about 67.4^\circ to the velocity. \displaystyle a_{\mathbf{N}} = \frac{144}{40} = 3.6 and a_{\mathbf{T}} = 1.5, so \displaystyle \lVert \mathbf{a} \rVert = \sqrt{15.21}, and the cosine of the angle is \displaystyle \frac{1.5}{3.9} = \frac{5}{13}.
  5. \displaystyle a_{\mathbf{T}} = \frac{2t}{\sqrt{1 + 2t^2}}; at t = 0, \displaystyle a_{\mathbf{N}} = \sqrt{5} and \displaystyle \kappa = \sqrt{5}. The speed is \displaystyle \sqrt{1 + 2t^2}; \mathbf{a}(0) = \langle 0, 2, 1 \rangle is perpendicular to \mathbf{v}(0) = \langle 1, 0, 0 \rangle, and \displaystyle \kappa(0) = \frac{\sqrt{5}}{1}.
  6. \displaystyle \kappa = \frac{\sqrt{6}}{9} and \displaystyle \mathbf{N} = \frac{1}{\sqrt{6}}\langle -1, -1, -2 \rangle. Here \mathbf{v} = \langle 1, 1, -1 \rangle and \mathbf{a} = \langle 0, 0, -1 \rangle, so \displaystyle a_{\mathbf{T}} = \frac{1}{\sqrt{3}}, \displaystyle a_{\mathbf{N}} = \sqrt{\frac{2}{3}}, and \displaystyle \mathbf{a} - a_{\mathbf{T}}\mathbf{T} = \frac{1}{3}\langle -1, -1, -2 \rangle.
  7. \displaystyle 25\sqrt{2} \approx 35.4 meters per second. 0.032\,v^2 \le 40 gives v^2 \le 1250.
  8. 16.9 meters per second squared; radius 40 meters. \displaystyle \kappa = \frac{10}{20^2} = \frac{1}{40}, and her \displaystyle a_{\mathbf{N}} = \frac{26^2}{40}.
  9. \mathbf{r}(t) = \langle t^3 + t, 3 + t^2, e^{-t} \rangle, at (2, 4, e^{-1}) when t = 1. First \mathbf{v}(t) = \langle 1 + 3t^2, 2t, -e^{-t} \rangle, since -1 + \int_0^t e^{-u}\,du = -e^{-t}; then integrate again from \mathbf{r}(0).
  10. It lands at \displaystyle t = \frac{5}{2} at \displaystyle \left(30, \frac{25}{2}, 0\right), rises to 25 feet at \displaystyle t = \frac{5}{4}, and has smallest speed 13 feet per second, at the top, in the plane 5x - 12y = 0. Here \mathbf{r} = \langle 12t, 5t, 40t - 16t^2 \rangle; at the top \mathbf{v} = \langle 12, 5, 0 \rangle is perpendicular to \mathbf{a}, and \mathbf{v}_0 \times \mathbf{a} = \langle -160, 384, 0 \rangle.
  11. It lands at t = 1 at \displaystyle \left(10, \frac{3}{2}, 0\right), in the plane 24x - 160y - 15z = 0. Here \displaystyle \mathbf{r} = \left\langle 10t, \frac{3}{2}t^2, 16t - 16t^2 \right\rangle and \mathbf{v}_0 \times \mathbf{a} = \langle -48, 320, 30 \rangle.
  12. By Theorem 1, a_{\mathbf{T}} is the derivative of the speed, which is 0 on an interval exactly when the speed is constant (mean value theorem). Also \displaystyle a_{\mathbf{N}} = \kappa\left(\frac{ds}{dt}\right)^2 with \displaystyle \frac{ds}{dt} > 0, so a_{\mathbf{N}} = 0 throughout exactly when \kappa = 0 throughout, which by Theorem 5 of Arc length and curvature happens exactly when every \mathbf{r}(t) lies on one line.
  13. By the chain rule, \mathbf{q}' = \varphi'\,\mathbf{r}'(\varphi) and \mathbf{q}'' = \varphi''\,\mathbf{r}'(\varphi) + \varphi'^2\,\mathbf{r}''(\varphi), so \mathbf{q} is smooth. As \mathbf{r}' \times \mathbf{r}' = \mathbf{0}, \mathbf{q}' \times \mathbf{q}'' = \varphi'^3\,\mathbf{r}' \times \mathbf{r}'', while \lVert \mathbf{q}' \rVert = \varphi'\,\lVert \mathbf{r}' \rVert. In Theorem 2 the numerator gains the factor \varphi'^3 and the denominator \varphi', so a_{\mathbf{N}} gains \varphi'^2. By Theorem 1, a_{\mathbf{T}} of \mathbf{q} is the derivative of its speed: \varphi''(u)\,\lVert \mathbf{r}'(\varphi(u)) \rVert + \varphi'(u)^2\,a_{\mathbf{T}}(\varphi(u)).
  14. As \mathbf{v} = \mathbf{v}_0 + t\mathbf{a}_0, \lVert \mathbf{v} \rVert^2 is a quadratic in t with leading coefficient \lVert \mathbf{a}_0 \rVert^2 > 0 and derivative 2\,\mathbf{v} \cdot \mathbf{a}_0, which is 0 only at \displaystyle t^* = -\frac{\mathbf{v}_0 \cdot \mathbf{a}_0}{\lVert \mathbf{a}_0 \rVert^2}. There the minimum is \displaystyle \lVert \mathbf{v}_0 \rVert^2 - \frac{(\mathbf{v}_0 \cdot \mathbf{a}_0)^2}{\lVert \mathbf{a}_0 \rVert^2}, which Lagrange’s identity (The cross product and the triple product) turns into \displaystyle \frac{\lVert \mathbf{v}_0 \times \mathbf{a}_0 \rVert^2}{\lVert \mathbf{a}_0 \rVert^2}. Take square roots.
  15. By Example 5, \mathbf{r} \times \mathbf{v} is constant, so it is \mathbf{0} for every t. Let \rho = \lVert \mathbf{r} \rVert > 0 and \displaystyle \mathbf{u} = \frac{\mathbf{r}}{\rho}, differentiable as in Theorem 1, so \mathbf{u} \cdot \mathbf{u}' = 0 (Theorem 3 of Vector-valued functions). From \mathbf{v} = \rho'\mathbf{u} + \rho\mathbf{u}', \mathbf{0} = \mathbf{r} \times \mathbf{v} = \rho^2\,\mathbf{u} \times \mathbf{u}'. As \mathbf{u} \perp \mathbf{u}' and \lVert \mathbf{u} \rVert = 1, \lVert \mathbf{u} \times \mathbf{u}' \rVert = \lVert \mathbf{u}' \rVert, so \mathbf{u}' = \mathbf{0}, \mathbf{u} is a constant \mathbf{u}_0, and \mathbf{r} = \rho\,\mathbf{u}_0 stays on the line along \mathbf{u}_0.
  16. Constant speed gives a_{\mathbf{T}} = 0 (Theorem 1), so by Theorem 2 a_{\mathbf{N}} = \lVert \mathbf{a} \rVert = A and \displaystyle \kappa = \frac{A}{c^2}. With s = c(t - t_0), the arc length parametrization \displaystyle \mathbf{q}(s) = \mathbf{r}\left(t_0 + \frac{s}{c}\right) has \displaystyle \mathbf{q}'' = \frac{1}{c^2}\mathbf{a} continuous. Taking the plane as the xy-plane (a rigid change of axes keeps lengths and curvature), Exercise 16 of Arc length and curvature puts the curve on a circle of radius \displaystyle \frac{1}{\kappa} = \frac{c^2}{A}.
math.wiki · updated 2026-11-17 · CC-BY-SA-4.0