Graphs: S S Eagerly " Even ST Suc
Graphs: S S Eagerly " Even ST Suc
l\frntion the real numbers to a mathematician and the image of a straight line will
probably form in her mind, quite involuntarily. And most likely she will neither
banish nor too eagerly embrace this mental picture of the real numbers. "Geomet-
ric intuition" will allow her to interpret statements about numbers in terms of this
picture, and may even suggest methods of proving them. Although the properties
of the real numbers which were studied in Part I are not greatly illuminated by a
geometric picture, such an interpretation will be a great aid in Part II.
You arc probably already familiar with the conventional method of considering
___l___ the straight line as a picture of the real numbers, i.e., of associating to each real
I
-1 0 2 2 3 number a point on a line. To do this (Figure 1) we pick, arbitrarily, a point which
FIGURE I we label 0, and a point to the right, which we label 1. The point twice as far to
the right is labeled 2, the point the same distance from O to I, but to the left of 0,
is labeled -1, etc. With this arrangement, if a < b, then the point corresponding
to a lies to the left of the point corresponding to b. \Ve can also draw rational
numbers, such as !, in the obvious way. It is usually taken for granted that the
irrational numbers also somehow fit into this scheme, so that every real number
can be drawn as a point on the line. \ Ve will not make too much fuss about
justifying this assumption, since this method of ''drawing" numbers is intended
solely as a method of picturing certain abstract ideas, and our proofs will ne\Tr
rely on these pictures (although we will frequently use a picture to suggest or help
explain a proof). Because this geometric picture plays such a prominent, albeit
inessential role, geometric terminology is frequently employed when speaking of
numbers thus a number is sometimes called a point, and R is often called the
real line.
The number la-bl has a simple interpretation in terms of this geometric picture:
it is the distance between a and b, the length of the line segment which has a as one
end point and b as the other. This means, to choose an example whose frequent
occurrence justifies special consideration, that the set of numbers x which satisfy
Ix - a I < £ may be pictured as the collection of points whose distance from a is
less than c. This set of points is the " interval'' from a - c to a+£, \\'hich may also
be clescribccl as the points corresponding to numbers x with a - £ < x < a + E
- -j- ---+ ----+--- (Figure 2).
a - E a
llCl ' RE 2 Sets of' numbers wl1ich correspond to intervals arise so frequently that it is desir-
able to have special names for them. The sct {x : a < .r < b} is denoted by (a, b)
and called the open interval from a to b. This notation naturally creates some
am biguity, sinn· (a. b) is also used to denote a pair of rn1111hcrs, but in context it is
always clear (or can easily be made clear) \\'hethcr one is talking about a pair or
a n i11tcr\'al. Note that if a ~ h, thcn (a. b) = 0 , the set \\'ith no clements; in prac-
56
.J. Graphs 5 7
a b a b tice, however, it is almost always assumed (explicitly if one has been careful, and
implicitly otherwise), that whenever an interval (a, b) is m entioned , the number a
the open interval (a, b) the closed interval [a, b]
is less than b.
a
the interval (-oo. a) The se t {x : a :::: x :::: b} is denoted by [a , b] and is called the closed interval
a
from a to b. This symbol is usually reserved for the case a < b , but it is som etim es
the interval (a. oo) used for a = b, also. The usual pictures for the intervals (a, b) and [a, b] are shown
a in Figure 3; since no reasonably accurate picture could ever indicate the difference
the interval (-oo, a] between the two intervals, various conventions have been adopted. Figure 3 also
a shows certain ''infinite" intervals. The set {x : x > a} is denoted by (a, oo),
the interval [a. oo) while the set {x: x ~ a} is denoted by [a, oo); th e sets (-oo, a) and (-oo, a] are
FIGURE3 defined similarly. At this point a standard warning must be issued: the symbols oo
an d -oo, though usually read "infinity" and "minus infinity," are purely sugges tive;
there is no number "oo" which satisfies oo ~ a for all numbers a . \Vhile the
symbols oo and -oo will appear in many contexts, it is always necessary to define
these uses in ways that refer only to numbers. The set R of all real numbers is
also considered to be an "interval," and is som etimes denoted by (-oo, oo).
(0 , b) (a, b)
-----. Of even greater interest to us than the method of drawing numbers is a method
I
I of drawing pairs of numbers. This procedure, probably also familiar to you, re-
(- 1, 1). .o. I) I
quires a ''coordinate system ,'' two straight lines intersecting at right angles. To
I
distinguish these straight lines, we call one the lzon"zontal axis, and one the vertical
(0, 0) (a, 0)
axis. (l\lore prosaic terminology, such as the "first" and "second" axes, is probably
(- ] , - ])• •(1, - ]) preferable from a logical point of view, but most people hold their books, or at
least their blackboards, in the same way, so that "horizontal" and "vertical" are
FIG U R E 4 more descripti\·e.) Each of the two axes could be labeled with real numbers, but
we can also label p oints on the horizontal axis with pairs (a, 0) and points on the
f (x) = l vertical axis with pairs (0, b), so that the intersection of the two axes, the "origin"
of the coordinate system , is labeled (0, 0). Any pair (a, b) can now be drawn as
in Figure 4, lying at th e vertex of the rectangle ~hose other three vertices are la-
beled (0, 0) , (a, 0), and (0, b). The numbers a and b are called the first and second
coordinates, respectively, of the point determined in this way.
f(x) = -! Our real concern, let us recall, is a m ethod of drawing functions. Since a func-
FIGURES tion is just a collection of pairs of numbers, we can draw a function by drawing
each of the pairs in the function. The drawing obtained in this way is called the
graph of the fun ction. In other words, the graph of f contains all the points cor-
responding to pairs (x, f (x) ). Since most functions contain infinitely many pairs,
drawing the graph promises to be a laborious undertaking, but, in fact, many
functions have graphs which are quite easy to draw.
Not surprisingly; the simplest functions of all, the constant functions f (x) = c,
have the simplest graphs. It is easy to see that the grap h of th e function f (x) = c
is a straight line para llel to the horizontal cLxis, at distance c from it (Figure 5).
The functions f (x) = ex also have particularly simple graphs straight lines
Fl(;URE 6 through (0, 0) , as in Figure 6. A proof of this fact is indicated in Figure 7:
58 Foundations
Let x be some number not equal to 0, and let L be the straight line which passes
through the origin O, corresponding to (0, 0), and through the point A, corre-
sponding to (x, ex). A point A', with first coordinate y, will lie on L when the
L triangle A' B' 0 is similar to the triangle ABO, thus when
A= (x,cx) A'B' AB
- - - - - -e·
OB' - OB - '
this is precisely the condition that A' corresponds to the pair (y, ey), i.e., that A'
lies on the graph of f. The argument has implicitly assumed that e > 0, but the
other cases arc treated easily enough. The number e, which measures the ratio of
the sides of the triangles appearing in the proof, is called the slope of the straight
FIG URE 7 line, and a line parallel to this line is also said to have slope e.
This demonstration has neither been labeled nor treated as a formal proo(
Indeed, a rigorous demonstration would necessitate a digression which we are
not at all prepared to follow. The rigorous proof of mry statement connecting
geometric and algebraic concepts would first require a real proof (or a precisely
stated assumption) that the points on a straight line correspond in an exact way
to the real numbers. Aside from this, it would be necessary to develop plane
geometry as precisely as we intend to develop the properties of real numbers.
Now the detailed development of plane geometry is a beautiful subject, but it is by
length c - a no means a prerequisite for the study of calculus. \Ve shall use geometric pictures
only as an aid to intuition; for our purposes (and for most of mathematics) it is
perfectly satisfactory to defi,ne the plane to be the set of all pairs of real numbers,
and to defi,ne straight lines as certain collections of pairs, including, among others,
F I GU RES the collections {(x, ex) : x a real number}. To provide this artificially constructed
geometry with all the structure of geometry studied in high school, one more
definition is required. If (a, b) and (c, d) are two points in the plane, i.e., pairs of
real numbers, we defi,ne the distance between (a, b) and (c, d) to be
/(a - e) 2 + (b - d)2.
If the motivation for this definition is not clear, Figure 8 should serve as adequate
J(x) = ex +d
explanation- with this definition the Pythagorean theorem has been built into our
/
/
geometry*
/1(x)=cx Reverting once more to our informal geometric picture, it is not hard to see
/
/
/
(Figure 9) that the graph of the function f (x) = ex + d is a straight line
with slope e, passing through the point (0. d). For this reason, the functions
f (x) = ex + d are called linear functions. Simple as they are, linear func-
tions occur frequently, and you should feel comfortable \\'Orking \\'ith them. The
following is a typical problem whose solution should not cause any trouble. Given
two distinct points (a, h) and (e, d), find the linear function f \\'hose graph goes
II ClRE9 through (a, b) and (e, d). This amounts to saying that .f(a) =band .f(c) = d. If
* The fa stidious reader might object to this d<"finition 011 the grounds that 11on11cgatin' m1111hers
are not yet k11ow11 to han· square roots. This ol~jert ion is really unanswerable at the 1110111e11t the
definition will just ha, ·c to be accepted with rcsen·ations, until this little point is settled.
4. Graphs 59
• (2.4)
(- 1, I) (I, I)
• •
(0, 0)
Fl(;URE 11
60 Foundations
It is not hard to convince yourself that all the pairs (x, x 2 ) lie along a curve like
the one shown in Figure 12; this curve is known as a parabola.
Since a graph is just a drawing on paper, made (in this case) with printer's ink,
the question ''Is this what the graph really looks like?" is hard to phrase in any
sensible manner. No drawing is ever realbJ correct since the line has thickness.
Nevertheless, there arc some questions which one can ask: for example, how can
you be sure that the graph does not look like one of the drawings in Figure 13?
It is easy to sec, and even to prove, that the graph cannot look like (a); for if
0 < x < y, then x 2 < y 2 , so the graph should be higher at y than at x, which is
not the case in (a) . It is also easy to see, simply by drawing a very accurate graph,
first plotting many pairs (x, x 2 ), that the graph cannot have a large 'jump" as in (b)
or a ''corner" as in (c). In order to prove these assertions, however, we first need
to say, in a mathematical way, what it means for a function not to have a 'jump"
or "corner"; these ideas already involve some of the fundamental concepts of
FIGURE 12 calculus. Eventually we will be able to define them rigorously, but meanwhile you
may amuse yourself by attempting to define these concepts, and then examining
your definitions critically. Later these definitions may be compared with the ones
mathematicians have agreed upon. If they compare favorably, you arc certainly
to be congratulated!
The functions f (x) = x 11 , for various natural numbers 11, arc sometimes called
power functions. Their graphs arc most easily compared as in Figure 14, by
drawing several at once.
The power functions are only special cases of polynomial functions, introduced
x y in the previous chapter. Two particular polynomial functions are graphed in
(a)
I f(x) = x 3 / f(x) = x2
I
J I
I
I
I
\
f( x) = x4 · I
I
I
I
I
f (x) = x
\ I
(-1. 1)\ (I, 1)
(b)
(c)
FI CU RE 14
I· I c; l ' R L I 3
4. GrajJhs 61
Figure I 5, while Figure 16 is meant to give a general idea of the graph of the
polynomial function
fG) Hr', =
f (~I ) I = (- )"+ t,
-2
(b)
f (x) = 1, Jx I ::: I,
FIGURE 15 and is a linear function on each interval [I/ (n + 1), 1/ n] and [ -1 / n, - I/ (11 + I)].
(The number O is not in the domain of f .) Of course, one can write out an explicit
formula for f (x) , when x is in [ 1/ (n + I), 1/ n]; this is a good exercise in the use
of linear functions, and will also convince you that a picture is worth a thousand
words.
x y
n even n odd
(a) (b)
FIGURE 16
62 Foundations
(a) (b)
x
f (x) = 1 + x2
(c) (d)
FIGURE!?
l
3
Fl G RE I
4. Graphs 63
FIGURE 19
1 1 1
f(x) =0 for x = -, , , ... ,
7T 2 7T 3 7T
1 1 1
f (x) = 1 for x = -1 - ' 1 . ' 1 ' ... '
2n 2n + 2n 2n + 4n
1 1 1
f(x) =-1 fo r x = -3- ' 3 ' 3 ' . . ..
2n 2n + 2n 2n + 4n
Notice that when x is large, so that 1/x is small, f (x) is al o small· when x is
. 1
m-
x
FI G RE 20
64 Foundations
/
/
/
/
/
/ . 1
/
J(x) = x sm -
/
x
/
/
FIG RE21
"large negative," that is, when Ix I is large for negative x , again f (x) is close to 0,
although f (x) < 0.
An interesting modification of this function is f (x) = x sin 1/ x. The graph of
this fun ction is sketched in Figure 21. Since sin 1/ x oscillates infinitely often near O
between 1 and -1 , the function f (x) = x sin 1/ x oscillates infinitely often between
x and - x. The behavior of the graph for x large or large negative is harder to
\
\ h
\ /,
\ I
\ I . l
\ f (x) = x 2 sm -
I x
\ I
\ I
\ I
I
/
\
\
\
\
\
\
\
\
\
\
Fl , R E 22
4. Graphs 65
f
analyze. Since sin l / x is getting close to 0, while x is getting larger and larger, there
seems to be no telling what the product will do. It is possible to decide, but this is
another question that is best deferred to Part III. The graph off (x) = x 2 sin 1/x
(a) has also been illustrated (Figure 22).
For these infinitely oscillating functions, it is clear that the graph cannot hope to
g be really "accurate." The best we can do is to show part of it, and leave out the
part near O (which is the interesting part). Actually, it is easy to find much simpler
functions whose graphs cannot be "accurately" drawn. The graphs of
f(x) = l x2,
2,
x < 1
x ~ 1
and g(x) =
l
x2
2,
'
x :::: 1
x > 1
(b) can only be distinguished by some convention similar to that used for open and
FIGURE23 closed intervals (Figure 23).
Out last example is a function whose graph is spectacularly nondrawable:
f(x) = l 0,
1,
x irrational
x rational.
x rational
f (x) = { l,
0, x irrational
FIG U RE 24
The graph of f must contain infinitely many points on the horizontal axis and
also infinitely many points on a line parallel to the horizontal axis, but it must not
(x,~ contain either of these lines entirely. Figure 24 shows the usual textbook picture
of the graph. To distinguish the two parts of the graph, the dots are placed closer
!ex - a) 2 + (y - 2
b) = r
or
(x - a) 2 + (y - b)2 = r 2.
66 Foundations
T he circle with center (0,0) and radius 1, often regarded as a sort of standard copy,
is called the unit circle.
A close relative of the circle is the ellipse. This is d fined as the set of points,
the sum of whose distances from two "focus" points is a con tant. (When the two
fo ci are the same, w obtain a circle.) If, for convenience, the focus points are
taken to be (- c , 0) and (c, 0) , and the sum of the distances is taken to be 2a (the
factor 2 simplifies som e algebra), then (x, y ) is on the ellipse if and only if
J (x - ( - c) ) 2 + y2 + J (x - c) 2 + y 2 = 2a
or
J (x + c)2 + y2 = 2a - J (x - c )2 + y2
or
x2 + 2cx + c2 + y 2 = 4a 2 - 4a J (x - c) 2 + y 2 + x 2 - 2cx + c 2 + y 2
or
4 (cx - a 2 ) = -4a Jex - c) 2 + y2
or
or
or
x2 y2 }
(0, b) { (x , y) : a2 + b2 = l
(- a , 0)
(0 , - b)
Fl ' R I·: 2 6
4. Graphs 67
FIUURE27 In this case, however, we must clearly choose c > a, so that a 2 - c 2 < 0. If
b= I c2 - a 2 , then ex, y) is on the hyperbola if and only if
x2 v2
a2 - !J2 = 1.
The picture is shown in Figure 27. It contains two pieces, because the difference
between the distances of ex. y) from e-c, 0) and ec. 0) may be taken in two dif-
ferent orders. The hyperbola intersects the horizontal axis when y = 0, so that
x = ±a, but it never intersects the ,·ertical axis.
It is interesting to compare (Figure 28) the hyperbola with a = b = J2 and
the graph of the function f ex) = 1/ x. The drawings look quite similar, and
the two sets are actually identical, except for a rotation through an angle of JT /4
(Problem 23 ).
Clearly no rotation of the plane will change circles or ellipses into the graphs of
functions. Nevertheless, the study of these important geometric figures can often
be reduced to the study of functions. Ellipses, for example, are made up of the
/
/
/
/
/
/
/
/
/
'' /
'//
(a) (b)
Fl c; L' RE 28
68 Foundations
-a~ x ~ a
and
-a~ x ~ a.
Of course, there are many other pairs of functions with this same property. For
example, we can take
-a~ x ~ 0
and
-bJI-(x 2 /a 2 ),
g(x) =
{
bJI-(x 2 /a 2 , -a~ x ~ 0.
We could also choose
f(x) =
b J1 - (x 2 /a 2 ), x rational, - a .:::: x ~ a
{
-b JI - (x 2 /a 2 ), x irrational, - a ~ x ~ a
and
x rational, - a ~ x ~a
x irrational, - a ~ x ~ a.
But all these other pairs necessarily involve unreasonable functions which jump
around. A proof, or even a precise statement of this fact, is too difficult at present.
Although you have probably already begun to make a distinction between those
functions with reasonable graphs, and those with unreasonable graphs, you may
find it very difficult to state a reasonable definition of reasonable functions. A
mathematical definition of this concept is by no means eas1~ and a great deal of this
book may be viewed as successive attempts to impose more and more conditions
that a ''reasonable" function must satisfy. As we define some of these conditions,
we will take time out to ask if we have really succeeded in isolating the function
which deserve to be called reasonable. The answer, unfortunately, will always be
"no," or at best, a qualified "yes."
PROBLEl\JS
1. Indicate on a straight line the set of all x satisfying the following conditions.
Also name each set, using the notation for intervals (in some cases you will
also need the U sign).
(i) Ix - 31 < I.
(ii) Ix - 31 ~ I.
(iii) Ix - al < c .
(iv) Ix 2 - 11 < {
4. Graphs 69
1 I
(v) --->-
1 + x2 - s.
1
(vi) ::S a (give an answer in terms of a, distinguishing various cases).
l+x 2
(vii) x 2 + 1 ~ 2.
(viii) (x + 1)(x - 1)(x - 2) > 0.
2. There is a very useful way of describing the points of the dosed interval [a, b]
(\vhcre we assume, as usual, that a < b).
(a) First consider the interval [O, b], for b > 0. Prove that if x is in [O, b],
then x = t b for some t with O ::S t ::S 1. What is the significance of the
number t? What is the mid-point of the interval [O, b]?
(b) Now prove that if x is in [a, b], then x = (1 - t)a + tb for some t with
0 ::S t ::S 1. Hint: This expression can also be written as a+ t(b - a).
What is the midpoint of the interval [a, b]? What is the point I /3 of the
way from a to b?
(c) Prove, conversely, that if O ::St ::S 1, then (1 - t)a + tb is in [a, b].
(d) The points of the open interval (a, b) arc those of the form (1 - t)a + tb
for O < t < 1.
3. Draw the set of all points (x, y) satisfying the following conditions. (In most
cases your picture will be a sizable portion of a plane, not just a line or curve.)
(i) x > y.
(ii) x +a> y +b.
(iii) y < x2.
(iv) y ::S x2.
(v) Ix - YI < 1.
(vi) Ix+ YI < 1.
(vii) x +y is an integer.
1 . .
(viii) - - 1s an mtegcr.
x+y
(ix) (x - 1) 2 + (v - 2) 2 < 1.
(x) x 2 < y < x 4 .
4. Draw the set of all points (x, y) satisfying the following conditions:
5. Draw the set of all points (x, y) satisfying the following conditions:
...,
(i) x = y-.
y2 x2
(ii) ---= 1
a2 b2 .
(iii) x = lyl.
(iv) x = smy.
Hint: You already know the answers when x and y are interchanged.
6. (a) Show that the straight line through (a, b) with slope 111 is the graph of the
function f (x) = m(.r - a)+ b. This formula, known as the "point-slope
form" is far more convenient than the equivalent expression f (x) =
mx + (b - ma); it is immediately dear from the point-slope form that the
slope is m, and that the value of f at a is b.
(b) For a -=f. c, show that the straight line through (a, b) and (c, d) is the
graph of the function
cl - b
f(x) = - - ( x - a)+ b.
c-a
(c) When are the graphs of f (x) = mx + b and g(x) = m'x + b' parallel
straight lines?
7. (a) For any numbers A, B, and C, with A and B not both 0, show that the
set of all (x. y) satisfying Ax+ By + C = 0 is a straight line (possibly a
vertical one). Hint: First decide when a vertical straight line is described.
(1, m)
(b) Show conversely that every straight line, including vertical ones, can be
described as the set of all (x, y) satisfying Ax+ By+ C = 0.
J (x) = mx + b,
g(x) = llX + C,
(1, n) are perpendicular if 11111 = -1, by computing the squares of the leng1hs
of the sides of the triangle in Figure 29. (\ \'hy is this special case, where
Ff(; U RE 2 9 the lines intersect at the origin, as good as the general case?)
(b) Prove that the two straight lines consisting of all (x, y) satisfying the con-
clitions
Ax+ By+ C = 0,
A'x + B' y + C' = 0.
2 2
+ J(x3 - x2) + (y3 - Y2) .
10. Sketch the graphs of the following functions, plotting enough points to get
a good idea of the general appearance. (Part of the problem is to make
a reasonable decision how many is "enough"; the queries posed below are
meant to show that a little thought will often be more valuable than hundreds
of individual points.)
1
(i) f (x) = x + -.
(What happens for x near 0, and for large x? \\There
x
does the graph lie in relation to the graph of the identify function? \\Thy
does it suffice to consider only positive x at first?)
1
(ii) f(x) =x - -.
x
') 1
(iii) f(x) = x-+ 7·
x-
l
(iv) f(x)=x 2 - 1 .
x-
12. Graph the functions f (x) = v'x form = 1, 2, 3, 4. (There is an easy way to
do this, using Figure 14. Be sure to remember, however, that "jx means the
positive mth root of x when m is even; you should also note that there will be
an important difference between the graphs when m is even and when m is
odd.)
15. Draw the graph of f (x) = ax 2 +bx+ c. Hint: Use the methods of Prob-
lem l-18.
16. Suppose that A and C are not both 0. Show that the set of all (x, y) satisfying
2
Ax + Bx + cy2 + Dy + E = 0
17. The symbol [x J denotes the largest integer which is :S x. Thus, [2.1 J = [2] =
2 and [ -0.9] = [ -1 J = -1. Draw the graph of the following functions
(they are all quite interesting, and several will reappear frequently in other
problems).
(v) f(x) = [~ J
(vi)
f(x) = [H 1
(i) f (x) = {x}, where {x} is defined to be the distance from x to the nearest
integer.
(ii) .f(x) = {2x}.
(iii) f(x) = {x} + !{2x}.
(iv) .f(x) = {4x}.
(v) f(x) = {x} + !{2x} + ±{4x}.
4. Graphs 73
FIGURE 30 *22. (a) Show that the square of the distance from (c, d) to (x. mx) is
x 2 (m 2 + I) +x(-2md- 2c) + d 2 + c2 .
Using Problem 1-18 to find the minimum of these numbers, show that
the distance from (c, d) to the graph of f (x) = m x is
lcm - dl/Jm 2 + I.
(b) Find the distance from (c, d) to the graph of f (x) = mx + b. (Reduce
this case to part (a).)
74 Foundations
*23. (a) Using Problem 22, show that the numbers x ' and y' indicated in Fig-
ure 31 are given by
I 1 1
x = h x + h y,
I 1 1
y = - h x +h y.
(b) Show that the et of all (x , y) with (x' / .J2 )2 - (y' / .J2 )2 = 1 is th same
F I GU RE 3 1 as the set of all (x, y) with xy = 1.
4) Appendix 1. rectors 75
APPENDIX 1. VECTORS
For convenience, we will use this convention that subscripts indicate the first and
second pairs of a point that has been described by a single letter. Thus, if we
mention the points w and z, it will be understood that w is the pair (w1. w2),
while.: is the pair (z1, z2).
Instead of the actual pair of numbers (v1, v2), we often picture v as an arrow
from the origin O to this point (Figure 1), and we refer to these arrows as vectors
in the plane. Of course, we've haven't really said anything new yet, we've simply
introduced an alternate term for a point of the plane, and another mental picture.
The real point of the new terminology is to emphasize that we are going to do
some new things with points in the plane.
For example, suppose that we have two vectors (i.e., points) in the plane,
0
FIGURE I Then we can define a new vector (a new point of the plane) v + w by the equation
(1)
Notice that all the letters on the right side of this equation are numbers, and the
+ sign is just our usual addition of numbers. On the other hand, the + sign on
the left side is new: previously, the sum of two points in the plane wasn't defined,
and we've simply used equation (1) as a defi,nitioiz.
A very fussy mathematician might want to use some new symbol for this newly
defined operation, like
v + w, or perhaps v EB w.
but there's really no need to insist on this; since v + w hasn't been defined before,
there's no possibility of confusion, so we might as well keep the notation simple.
Of course, any one can make new notation; for example, since it's our definition,
we could just as well have defined v + w as (v1 + wi · w2. v2 + w1 2), or by some
other equally weird formula. The real question is, does our new construction have
l any particular significance?
Figure 2 shows two vectors v and w, as well as the point
) (v2 + w,) - v2 which, for the moment, we have simply indicated in the usual way, without drawing
an arrow. Note that it is easy to compute the slope of the line L between v and
our new point: as indicated in Figure 2, this slope is just
( v2 + w2) - v2 w2
(vi + wi) - vi wi
and this, of course, is the slope of our vector w, from the origin O to (w1, w 2). In
FIGLiRE 2 other words, the line L is parallel to w.
76 Foundations
Similarly, the slope of the line M between (w1, w2) and our new point is
( v2+ w2) - w2 v2
(VI + W i) - v2 VI
which is the slope of the vector v; so M is parallel to v. In short, the new point
v + w lies on the parallelogram having v and w as sides. \ Vhen we draw v + w as
an arrow (Figure 3), it points along the diagonal of this parallelogram. In physics,
vectors are used to symbolize forces, and the sum of two vectors represents the
resultant force when two different forces arc applied simultaneously to the same
object.
FIGURE3 Figure 4 shows another way of visualizing the sum v+w. Ifwe use "w" to denote
an arrow parallel to w, and having the same length, but starting at v instead of at
the origin, then v + w is the vector from O to the final endpoint; thus we get to
v + w by first following v, and then following w.
Many of the properties of + for ordinary numbers also hold for this new + for
vectors. For example, the "commutative law''
V +W = W + V,
is obvious from the geometric picture, since the parallelogram spanned by v and
w is the same as the parallelogram spanned by w and v . It is also easily checked
analytically, since it states that
VJ+ W] = WJ + VJ,
v2 + w2 = w2 + v2.
FI G U R E 4 Similarly, unraveling definitions, we find the "associative law''
[ v + w] +.:: = v + [w + .::].
Figure 5 indicates a method of finding v + w + z.
The origin O = (0, 0) is an "additive identity,"
0 + v = v+ 0 = v,
and if we define
w - v = w+( - v),
I· I< ; l ' R L 5
4) Appendix 1. T1ectors 77
v + (w - v) = w.
" w-v ,, Figure 6(a) shows v and an arrow "w - v" that is parallel to w - v but that starts
at the endpoint of v. As we established with Figure 4, the vector from the origin
to the endpoint of this arrow is just v + (w - v) = w (Figure 6(b)) . In other words,
we can picture w - v geometrically as the arrow that goes from v to w (except that
(a) it must then be moved back to the origin).
There is also a way of multiplying a number by a vector: For a number a and
a vector v = (v1, v2), we define
O· v = 0,
-l·v=-v.
2v Notice that we have only defined a product of a number and a vector, we haYe
not defined a way of 'multiplying' two vectors. to get another vector.* However,
there are various ways of 'multiplying' vectors to get numbers, which are explored
in the following problems.
PROBLEl\IS
I
--v 1. Given a point v of the plane, let Re(v) be the result of rotating v around the
2
origin through an angle of e (Figure 8). The aim of this problem is to obtain
FIGURE? a formula for Re, with minimal calculation.
(a) Show that
Re (L 0) = (cose, sine), [we should really write Re((l, 0)), etc.]
Re(O, l) = (-sine,cose).
(b) Explain why we have
Re(v + w) = Re(v) + Re(w),
Re(a · w) =a· Re(w).
, Re(v)
I
(c) Now show that for any point (.c y) we have
/.....,_() .. v
I
I
..,
) _., Re (x, y) = (x cos e - y sine. x sin e + y cos e ).
..,
* If you jump to Chapter 25 , you 'll find that there is an import a nt way of defining a product, but
this is something \'ery special for the pla ne- it doesn't work for \'Cc tors in 3-space, for example, c\·cn
FH; URE 8 though the other constructions do.
78 Foundations
this is often called the 'clot product' or 'scalar product' of v and w ('scalar'
being a rather old-fashioned word for a number, as opposed to a vector).
(a) Given v, find a vector w such that v • w = 0. Now describe the set of all
such vectors w.
(b) Show that
V•W=W•V
v • (w + z) = v•w + v •z
and that
a· (v · w) =(a· v) · w = v ·(a· w).
Notice that the last of these equations involves three products: the clot
product · of two vectors; the product · of a number and a vector; and
the ordinary product · of two numbers.
(c) Show that v • v ~ 0, and that v • v = 0 only when v = 0. Hence we can
define the norm II v II as
llvll = Jij":v,
which will be O only for v = 0. \Vhat is the geometric interpretation of
the norm?
(cl) Prove that
llv + wll .::: llvll + llwll,
and that equality holds if and only if v = 0 or w = 0 or w = a · v for
some number a > 0.
(c) Show that
llv + wll 2 - llv - wll 2
V • UI =
4
3. (a) Let Re be rotation by an angle of e (Problem l ). Show that
Re(v) • Re(w) = v • w.
(b) Let e = ( l, 0) be the \'Cctor of length 1 pointing along the first axis, and
let w = (cos e, sin B); this is a \·ector of length l that makes an angle of e
with the first axis (compare Problem l ). Calculate that
e. w = cose.
Conclude that in general
4. Given two vectors v and w, we'd expect to have a simple formula, involving
the coordinates v 1, v2, w l, w2, for the area of the parallelogram they span.
Figure 9 indicates a strategy for finding such a formula: since the triangle
with vertices w, A, v + w is congruent to the triangle O Bv, we can reduce
L
the problem to an easier one where one side of the parallelogram lies along
the horizontal axis:
(a) The line L passes through v and is parallel to w, so has slope w2/w1.
Conclude that the point B has coordinate
vi w2 - WJ v2
F I GURE9 This formula, which defines the determinant det, certainly seems to be simple
enough, but it can't really be true that det(v, w) always gives the area. After
all, we clearly have
det(w, v) = - det(u, w),
so sometimes det will be negative! Indeed, it is easy to see that our "deriva-
tion'' made all sorts of assumptions (that w2 was positive, that B had a positive
coordinate, etc.) Nevertheless, it seems likely that det(u, w) is± the area; the
next problem gives an independent proo(
5. (a) If v points along the positive horizontal axis, show that det( v, w) is the
area of the parallelogram spanned by v and w for w above the horizontal
axis (w2 > 0), and the negative of the area for w below this axis.
(b) If Re is rotation by an angle of e (Problem J), show that
det(Rev, Rew)= det(v, w).
Conclude that det(u, w) is the area of the parallelogram spanned by
v and w \vhen the rotation from v to w is counterclockwise, and the
negative of the area when it is clockwise.
6. Show that
det(u, w + z) = det(u, w) + det(u, z)
det(u + w, z) = det(u, z) + det(w, z)
and that
a det(v, w) = det(a · v, w) = det(u, a· w).
: llwll · sine 7. Using the method of Problem 3, show that
FIGCRE IO which is also obvious from the geometric interpretation (Figure 10).
80 }oundations
FIGURE2
For any given first two coordinates x and y, the point (.x, y , 0) in the horizontal
plane has distance Jx2 + y2 from the origin, and thus
(I) (x, y , z) is on the cone if and only if z = ±CJ .x 2 + y 2 .
\tVe can descend from these three-dimensional vistas to the more familiar two-
dimensional one by asking what happens wh en we intersect this cone with som e
plane P (Figure 3).
FI GU RE 3
Ir the plane is parallel to the horizontal pla11c, there's ce rta inly no mystery- the
intersection is just a circle. Otherwise, the plane P intersects the horizo ntal plane
4) Appendix 2. The Conic Sections 81
in a straight line. We can make things a lot simpler for ourselves if we rotate
everything around the vertical axis so that this intersection line points straight out
from the plane of the paper, while the first axis is in the usual position that we
are familiar with. The plane P is thus viewed "straight on," so that all we see
slope C (Figure 4) is its intersection L with the plane of the first and third axes; from this
view-point the cone itself simply appears as two straight lines.
L If this line L happens to be vertical, consisting of all points (a, z) for som e a,
then equation (1) says that the intersection of the cone and the plane consists of
all points (a,y,z) with
which is an hyperbola.
Otherwise, in the plane of the first and third axes, the line L can be described
FIGURE4 as the collection of all points of the form
(x, Mx + B),
where M is the slope of L. For an arbitrary point (x, y, z) it follows that
Mx + B = ±cff+ y2.
Now we have to choose coordinate axes in the plane P. We can choose L as the
first axis, measuring distances from the intersection Q with the horizontal plane
(Figure 5); for the second axis we just choose the line through Q parallel to our
original second axis. If the first coordinate of a point in P with respect to these
axes is x, then the first coordinate of this point with respect to the original axes
can be written in the form
ax+ f3
ax+ f3
for some a and f3. On the other hand, if the second coordinate of the point with
I respect to these axes is y, then y is also the second coordinate with respect to the
FIGURES original axes.
Consequently, (*) says that the point lies on the intersection of the plane and the
cone if and only if
a 2 (M 2 + Ex + F
2
C _v2 - - C 2 )x 2 = 0
( 1) If M = ±C we obtain a parabola;
(2) If C 2 > M 2 we obtain an ellipse;
(3) If c 2 < M 2 WC obtain an hyperbola.
(1 ) If our plane is parallel to one of the generating lines of the cone we obtain
a parabola;
(2) If our plane slopes less than the generating line of the cone (so that our
intersection omits one half of the cone) we obtain an ellipse;
(3) If our plane slopes more than the generating line of the cone we obtain an
hyperbola.
FIGURE6
In fact, the very names of these "conic sections" are related to this description.
The word Jwrabola comes from a Greek root meaning 'alongside,' the same root
that appears in parable, not to mention paradig1n, paradox, paragon, paragraph,
paralegal, parallax, parallel, and even parachute. Ellipse comes from a Greek root
meaning 'defect,' or omission, as in ellipsis (an omission, ... or the dots that in-
dicate it). And hyjJerbola comes from a Greek root meaning 'throwing beyond,' or
excess. \\Tith the currency of words like hyperactive, hypersensitive, and h)verven-
tilate, not to mention hype, one can probably say, without risk of hyperbole, that
this root is familiar to almost everyone.*
PROBLEl\IS
* ,\!though the corrcspm1uc1HT between these root s and the gcomctric picture correspond so beau-
tif11lly, !cir the sake of dull accuracy it ha s to be reported that the Gree-ks originally applied the words
tu describe fi:atures of cntain equations im·oh-ing th e conic sections.
4, ApjJendix 2. The Conic Sections 83
(a) (b )
F J< ; L' RE9
84 Foundations
In this chapter we've been acting all along as if there's only one way to label
y p points in the plane with pairs of numbers. Actually, there are many different
ways, each giving rise to a different "coordinate system. " The usual coordinates
of a point are called its cartesian coordinates, after the French mathematician
and philosopher Rene Descartes (1596- 1650), who first introduced the idea of
x coordinate systems. In many situations it is more convenient to introduce polar
coordinates, which are illustrated in Figure I. To the point P we assign the polar
FIGURE I coordinates (r, 8), where r is the distance from the origin O to P , and e is the
measure, in radians, of the angle between the horizontal axis and the line from
O to P. This e is not determined unambiguously. For example, points on the
right side of the horizo ntal axis could have either e = 0 ore = 2n; moreover, e
is completely ambiguous at the origin O. So it is necessary to exclude some ray
through the origin if we want to assign a unique pair (r, 8) to each point under
consideration.
On the other hand, there is no problem associating a unique point to any pair
(r. 8) . In fact, it is possible (though not approved of by all) to associate a point
to (r, 8) when r < 0, according to the schem e indicated in Figure 2. Thus, it
always makes sense to talk about "the point with polar coordinates (r, 8)," (with
or without the possibility of r < 0), even though there is some ambiguity when we
talk about "the polar coordinates" of a given point.
It is clear from Figure I (and Figure 2) that the point with polar coordinates
(r , 0) has cartesian coordinates (x, y) given by
x = r ros e, y = r sin e .
4, Appendix 3. Polar Coordinates 85
Conversely, if a point has cartesian coordinates (x, y ), then (any of) its polar co-
ordinates (r, ()) satisfy
r = ±Jx2 + y2
y
tan()= - if x =J. 0.
x
a Now suppose that f is a function . Then by the graph off in polar co-
ordinates we mean the collection of all points P with polar coordinates (r, ())
satisfying r = f (()). In other words, the graph of f in polar coordinates is the
collection of all points with polar coordinates (!(()), ()). No special significance
should be attached to the fact that we are considering pairs (!(()),()),with f (())
FIGURE3 first, as opposed to pairs (x, f (x)) in the usual graph of f; it is purely a matter of
convention that r is considered the first polar coordinate and () is considered the
second.
The graph of f in polar coordinates is often described as "the graph of the
equation r = f (())." For example, suppose that f is a constant function, f (()) = a
for all (). The graph of the equation r = a is simply a circle with center O and
radius a (Figure 3). This example illustrates, in a rather blatant way, that polar
coordinates are likely to make things simpler in situations that involve symmetry
with respect to the origin O.
The graph of the equation r =()is shown in Figure 4. The solid line corresponds
to all values of() :::::. 0, while the dashed line corresponds to values of () ::::: 0.
I
I
I
I
I
I
I
I
I
I .....
I
I
' '\
I \
I
(a)
\
\
\
\
''
' ' ..... ..........
As another example involving both positive and negative r, consider the graph of
the equation r = cos(). Figure 5(a) shows the part that corresponds to O ::::: () ::::: JT /2
Figure 5(b) shows the part corresponding to n/2 ::::: () ::::: n; here r < 0. You can
check that no new points are added for () > JT or () < 0. It is easy to describe
(b) this same graph in terms of the cartesian coordinates of its points. Since the polar
Fl c; UR E 5 coordinates of any point on the graph satisfy
,. =cos(),
86 Foundations
and hence
2
r = r cose ,
which describes a circle (Problem 4-16). [Conversely, it is clear that if the cartesian
coordinates of a point satisfy x 2 + y 2 = x then it lies on the graph of the equation
r = cos e.J
Although we've now gotten a circle in two different ways, we might well be
hesitant about trying to find the equation of an ellipse in polar coordinates. But
it turns out that we can get a very nice equation if we choose one of the foci as
the origin. Figur 6 shows an ellipse with one focus at O, with the sum of the
distances of all points from O and the other fo cus f being 2a . We've chosen f to
the left of O , with coordinates written as
(-2w, 0).
FIGURE 6
(1)
By as umption, the di tan e from (x, y) to f i 2a - r , h nee
or
2
(2) 4a 2 - 4ar + r2 = x
2
+ 4sax + 4s 2 a 2 +
ublra Ling (1) fr m (2) and dividing by 4a , w g t
a - r = sx + s 2 a,
4) Appendix 3. Polar Coordinates 87
or
2
r = a - 8X - 8 a
= (1 - t:2)a - 8X,
r=A-8rcos8,
r(l + 8 cos&)= A,
and thus
A
(4) r=----
l + 8 cos e
In Chapter 4 we found that
x2 y2
(5) -+-=1
a2 b2
is the equation in cartesian coordinates for an ellipse with 2a as the um of the
distances to the foci, but with the foci at (-c, 0) and (c, 0) , where
b = .Ja 2 - c 2 .
Since the distance between the foci is 2c, when this ellipse i moved left by c
units, so that the focus (c, 0) is now at the origin, we get the ellipse (4) when we
take c = w or 8 = c/a (with equation (3) determining A). Conver ely, gi en
the ellipse described by (4), for the corresponding equation (5) the value of a is
determined by (3),
A
a=--2,
1- 8
and again using c = 8a, we get
Thu , we can obtain a and b, the lernrths of the m ajor and minor ax , imm diat 1
from 8 and A.
The numb r
PROBLEMS
1. If two points have polar coordinates (r1, B1 ) and (r2, B2), show that the dis-
tance d between them is given by
d
2 = r1 2 + r} - 2r1r2 cos(B1 - B2).
(i) r = a sin e.
(ii) r = a sec e. Hint: It is a very simple graph!
(iii) r = cos 2e. Good luck on this one!
(iv) r = cos 3e.
(v) r = I cos 2e1.
(vi) r = lcos3e1.
4. Find equations for the cartesian coordinates of points on the graphs (i), (ii)
and (iii) in Problem 3.
5. Consider a hyperbola, where the difference of the distance between the two
foci is the constant 2a, and choose one focu s at O and the other at (-2w, 0).
(In this case, we must have t: > 1). Show that we obtain the exact same
equation in polar coordinates
A
r= - - - -
1 + c: cos e
as we obtained for an ellipse.
6. Consider the set of points (x, y) such that the distance (x, y) to O is equal to
the distance from (x, y) to the line y = a (Figure 7). Show that the distance
to the line is a - r cos e, and conclude that the equation can be written
(x , y )
•
a = ,. (1 + cos e) .
y=a
Notice that this equation for a parabola is again of the same form as (4).
7. Now, for any A and t: , co nsider the graph in polar coordinates of the equa-
0
tion (4), which implies (3). Show that the point s satisfying this equation satisfy
(I - c: 2 )x 2 + y2 = A 2 - 2/\c:x.
11 (; i ·RI : 7 Using Problem 4-16, show that this is an ellipse for F < 1, a parabola for
c = 1, and a h)11crhola for c > 1.
4) Appendix 3. Polar Coordinates 89
x2 + y2 = J x2 + y2 _ y,
and conclude that it can be described by the equation
(x2 + y2 + y )2 = x2 + y2
9. Sketch the graphs of the following equations.
(i) r = 1- i sin e.
(ii) r = 1 - 2 sin e.
(iii) r = 2 + cos e.
10. (a) Sketch the graph of the Lemniscate
(-a, 0) (a, 0) r
2
= 2a 2 cos 2e.