Appendix 4

Curves and Arc Length

From the discussion of Section 7 of Chapter 5, it doesnt seem at all unreasonable to conjecture that arcsin(t) represents the radian measure of the angle (signed length of the arc of the unit circle) going from (1,0) to the point (COS (arc sin(t)), SIN (arc sin (t))). In order to verify this conjecture, we have to better pin down the concept of angle. To do this we will introduce the concepts of curve and length of arc (curve length). Intuitively, a curve in two or three dimensions is just a set of points, (x1 (t) x 2 ( t )) or ( x 1 ( t ), x 2 (t ), x 3 ( t )) indexed by some variable, t , which could (but need not) represent time. If t does stand for time, then (x1 (t), x2(t)) might represent the position in plane of some object at time t. Similarly, (x1 (t), x2(t), x3(t)) would represent the position of an object in three-dimensional space at time t .We illustrate this concept in Figure A4-1.

No understanding should be lost by giving the following general denition of curve.

Denition A4.1: Curve

A curve in n dimensions is a sequence x = (x1, x2, ..., xn) of real valued functions of a real variable (Denition 2.1 on page 21) having a common interval domain (Denition 1.1 on page 2).

As already mentioned, a curve may be viewed as an ordered set of points being traced out in n-dimensional space over time. Although our development for our particular application only requires two dimensions (n = 2) it should be evident that this restriction is not necessary for the algebraic manipulations we will soon perform.

1999 by CRC Press LLC

It is convenient to introduce the concept of tangent line to a curve at this point.

Denition A4.2: Tangent to a curve

Suppose x = (x1, x2, ..., xn) is a curve (Denition A4.1) whose derivative (Denition 2.5 on page 30)
, x ,, x ) x = ( x1 2 n

exists at t = t0. The tangent to x at t = t0 is the vector valued function whose value, T t (t) at t is given by
0

T t (t) = x(t0)+(t t0 ) x(t0)

0

(t0), x2(t0)+(t t0) x (t0),..., x1(t0)+(t t0) x (t0)) = (x1(t0)+(t t0) x 1 2 1

We note that under our denition, the tangent to a two-dimensional curve is not restricted to being nonvertical, and if the range (see Denition 1.1 on page 2) of a curve intersects itself, there are likely to be two different tangent lines at the point of intersection, as can be seen in Figure A4-1. We can approximate the length of the curve illustrated in Figure A4-1 from t = a to t = b by the sum of the lengths of the line segments connecting the indicated points in this gure. Recall that the Pythagorean denition of length (Denition 8.6 on page 270) as motivated in this books preface where we establish the Pythagorean theorem, is given by the following.

Denition A4.3: Pythagorean length, || ||

The length, a b , of the line segment connecting the points a = (a1, a2, ..., an) and b = (b1, b2,...,bn) in n dimensions is given by

ab =

( ai bi )2 .
i=0

The quantity a b is also referred to as the distance between the vectors or points a and b.

1999 by CRC Press LLC

Applying Denition A4.3 to the hypotenuse line segment in Figure A4-2, we see that its length is [ x 2 ( t i +1 ) x 2 ( t i ) ]2 [ x 1 ( t i + 1 ) x 1 ( t i ) ]2 .

Hence the approximate length of the planar curve x from t = a to t = b is given by

n 1

i =0

[ x 2 ( t i + 1 ) x 2 ( t i ) ]2 [ x 1 ( t i +1 ) x 1 ( t i ) ]2 .

By the mean value theorem (Theorem 3.1 on page 62) this expression may be written in the form
n 1

i =0

2 2 * ( )(t [ x2 i i + 1 t i ) ] [ x 1 ( i ) ( t i +1 t i ) ] ,

where i and i* are (unspecied) points between ti and ti +1. This last sum may be rewritten as
n 1

i =0

( ) ]2 [ x ( * ) ]2 ( t [ x2 i 1 i i +1 t i ) .

This last form of the approximation to the length of the planar curve x from t = a to t = b is also seen to be essentially (if these derivatives are continuous, Denition 2.6 on page 34) a Riemann sum approximation to the integral

(see Denition 4.2 on page 102). This then leads to the denitions for curve length in two and in n dimensions.

b ( ) ]2 + [ x ( ) ]2 d . [ x1 2

Denition A4.4: Length of a curve (arc length)

If x = (x1, x2) is a two-dimensional curve whose domain is the interval [a;b], then the length of x is dened to be

1999 by CRC Press LLC

b ( ) ]2 [ x ( ) ]2 d [ x1 2

provided this integral exists. Similarly, if x = (x1, x2,..., xn) is an n-dimensional curve, whose domain is the interval [a;b], the following integral (provided it exists) is the length of x:

a [ x1 ( ) ]2 d .
i =1

The extension of these denitions of a portion of a curve (i.e., to the part of x in which t is restricted to a subinterval of its domain) is simply to carry out the above integration over this subinterval.

Example A4.1: Arc length for x = (t 2/2, (2t + 4)3/2/3)

We want to nd the length of this curve from t = 2 to t = 7. Using Denition ( t ) = t, x ( t ) = A4.4, we need x 1 2 t + 4 , so that 2

b ( ) ]2 + [ x ( ) ]2 d = [ x1 2

2 2

2 + 2 + 4 d ( + 2 )2 d

= =

2 ( + 2 ) d
=7 2

( + 2 )2 = -----------------2

is the desired arc length.

I

At this point I should point out that almost any real-life arc length problems that you will run into are not likely to have simple answers like the one above; most will require numerical integration. But if you feel compelled to go through some exercises which yield such simple answers, here is the way this example was constructed. Starting off with some integrand which is nonnegative and whose integral allows a formula evaluation, such as + 2 for going from 2 to 7 above, we write this integrand in the form

1999 by CRC Press LLC

( + 2 )2 which we expand out to 2 + 2 + 4 . Then choose some of the terms under the square root sign to represent ( ) ] 2 and the remaining terms to represent [ x ( ) ] 2 , these choices [ x1 2 being made so that it is possible to determine formulas for x 1 ( ) and x 2 ( ) ( ) ] 2 = 2, [ x ( ) ] 2 = 2 + 4 so that we easily. In this case we chose [ x 1 2 were led to the original formulas for x 1 ( ) and x 2 ( ).

Exercises A4.2
For those with access to the Mathematica program, make up your own arc length exercises following the suggestions of Example A4.1. You can check the various stages of your computation as well as your answers using the Mathematica tools we introduced earlier (see the discussion just preceding Exercises 2.15 on page 54, as well as Exercise 4.20.5 on page 125). Returning from the arc length denition diversion, if we look specically at the curve x given for 1 < t < 1 by x1 ( t ) = we see that This range of this curve is the part of the unit circle centered at the origin, which is at (1,0) for t = 0, and goes 1/4 revolution counterclockwise as t goes from 0 to 1, and 1/4 revolution clockwise as t goes from 0 to 1, as illustrated in Figure A4-3. 1 t , x2 ( t ) = t ,
2

It is not difcult to verify that

()] + [ x ()] = [ x1 2 2 2

1 2 -. ------------- + 1 = ----------------2 1 1 2

1999 by CRC Press LLC

This shows that the value, arcsin ( t ), of the arcsine function given in Denition 5.13 on page 160, for 1 < t < 1, represents the arc length along the unit circle centered at the origin starting from the point (1,0) ending at the point ( 1 t 2, t ) . So = arc sin ( t ) is the signed radian measure of the angle (wedge) determined by the origin, (0,0), and the ordered pair of points (1,0) , ( 1 t 2, t ) . The quantity t = SIN ( ) = SIN ( arc sin ( t )) , being the signed length of the vertical leg of the right triangle in Figure A4-3 is the familiar ratio opposite / hypotenuse since the hypotenuse is of unit length. The quantity COS() = SIN() = adjacent/hypotenuse. 1 SI N ( ) =
2

1 t is the familiar

So we see that for any given angle (wedge) as illustrated in Figure A4-4.,

whose radian measure is dened by the equation signed length of indicated circular arc = -----------------------------------------------------------------------------------------c we nd that a opposite -, SIN ( ) = -------------------------- = -c hypotenuse b adjacent -, COS ( ) = -------------------------- = -c hypotenuse

where a is the signed length1 of the vertical leg of the right triangle in Figure A4-4. The introduction of the usual geometric interpretation of SIN and COS furnishes a reasonable way to extend the denitions of these functions beyond their original domains. First of all we notice that from examination of Figure A4-3, we expect arc sin ( t ) = - d 0 ----------------2 1
t

1. In referring to the signed length, a is taken to be positive if the vertical leg is above the horizontal one. The circular arc is assigned a positive length if it arises from a counterclockwise rotation.

1999 by CRC Press LLC

to have a nite limit as t approaches 1 (from below, of course), since this limit corresponds to the length of 1/4-th of the unit circle. We leave the proof of this assertion of Problem A4.3 on page 779. The length of the entire unit circle is denoted by 2. ( itself is the arc length of any circle measured in diameters of this circle.) From a practical viewpoint, if we want to compute with the knowledge we currently have, its better to compute /4 = arc sin(1/ 2 ) (see the discussion preceding equation 5.59 on page 162). Referring to our copy of Figure A4-3 given below as Figure A4-3, if we continue moving counterclockwise on the unit circle, having started at (1,0),

the vertical line of the right triangle continues to exist, remaining positive until a half circle has been traversed (covering an arc length of ), and then becoming negative until an arc length of 2 has been traversed. The symmetry of this process, together with the analogous reasoning if we had chosen to move clockwise starting from (1,0) provides the geometric motivation for the denition of the sine function that we are about to introduce. We will do this formally for reference purposes.

Denition A4.5:
The number represents the arc length of a half-cricle of radius 1 unit.1One formula for which is amenable to digital computer evaluation is = 4

1/ 2

1 d . ----------------2 1

Problem A4.3
Show that lim
t 1 0 t <1

1 d exists. ----------------2 1

1. Arc length measured in the same units being used to measure the radius length.

1999 by CRC Press LLC

1 1 1 Hints: ----------------- = ------------------------------------- --------------2 1 1 (1 )(1 + )

for

0 < 1. But using

standard techniques the right side is easy to integrate. Youll also need equation 4.18 on page 107 of Theorem 4.5, and Theorem 3.8 on page 86, which you may need to modify a little to apply to the given situation.

1999 by CRC Press LLC