Swarthmore College

Works

Mathematics & Statistics Faculty Works Mathematics & Statistics

3-1-1983

Some Unusual Locus Problems

Stephen B. Maurer , '67
Swarthmore College, [email protected]

Follow this and additional works at: https://works.swarthmore.edu/fac-math-stat

Part of the Discrete Mathematics and Combinatorics Commons

Let us know how access to these works benefits you

Recommended Citation
Stephen B. Maurer , '67. (1983). "Some Unusual Locus Problems". Two-Year College Mathematics
Journal. Volume 14, Issue 2. 146-153. DOI: 10.2307/3026591
https://works.swarthmore.edu/fac-math-stat/72

This work is brought to you for free by Swarthmore College Libraries' Works. It has been accepted for inclusion in
Mathematics & Statistics Faculty Works by an authorized administrator of Works. For more information, please
contact [email protected].
Some Unusual Locus Problems
Author(s): Stephen B. Maurer
Source: The Two-Year College Mathematics Journal, Vol. 14, No. 2 (Mar., 1983), pp. 146-153
Published by: Mathematical Association of America
Stable URL: http://www.jstor.org/stable/3026591
Accessed: 19-10-2016 15:23 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Mathematical Association of America is collaborating with JSTOR to digitize, preserve and extend access
to The Two-Year College Mathematics Journal

This content downloaded from 130.58.65.20 on Wed, 19 Oct 2016 15:23:04 UTC
All use subject to http://about.jstor.org/terms
 Some Unusual Locus Problems

Stephen B. Maurer

Stephen B. Maurer is Associate Professor of Mathematics at

Swarthmore College, where he has taught since 1979. He grew
up in Silver Spring, Maryland. He received a B.A. from
Swarthmore in 1967 and a Ph.D. at Princeton in 1972. He has
taught previously at Princeton, the University of Waterloo,
Hampshire College, and the Phillips Exeter Academy. His major
scholarly interest has been research and curricular develop-

tE 1: lwl l O ment in discrete mathematics, but he has also made

sometimes continuous-into mathematical biology, economics,
and anthropology. He is a MAA Visiting Lecturer, Chairman of
the MAA Committee on High School Contests, and a member
of CUPM. In 1982-83 he is on leave at the Alfred P. Sloan
Foundation, working primarily on a new program to encourage
liberal arts colleges to bring mathematical, technological, and
computer literacy more into the core of the general curriculum.

Most locus problems one sees are distance problems, and in geometry texts distance
almost always means either distance between points or distance between a point
and a line. These two distance concepts are special cases of a more general concept,
the distance between two point sets. Once one understands this more general
concept, many new distance locus problems arise. The solutions are often surprising
(at least they were to me) and they provide a good review of the locus properties of
the conic sections.
In this article, we explain this more general distance concept and give several
examples of its use, leaving more and more of the justification to the reader as we
go along.
To give an idea where we are headed, here is the last problem we will analyze:
Find the "locus" of all circles equidistant from two given circles.
We will assume that all point sets are on the plane and that the distance between
points a = (x, y) and b = (x', y') has been defined as usual:

d(a, b) = Y(x- + (y _yT)2

If A and B are point sets, define their distance d(A, B) to be

glb{d(a,b)Ia E A,b E B

where as usual glb means greatest lower bound. Also, we define the distance d(a, B)
from point a to set B to be d({a},B).
The glb need not be in the set of point distances; that is, there need not exist a
closest pair a E A and b E B with d(a, b) = d(A, B). For instance, let A be the line

146

This content downloaded from 130.58.65.20 on Wed, 19 Oct 2016 15:23:04 UTC
All use subject to http://about.jstor.org/terms
y = <r12 and B be the graph of y = Arctanx. Then d(A,B) = 0 although A n B
= 0. Another example: Let A be the open left half-plane {(x, y) I x < 0} and B be
the open right half-plane {(x, y) Ix > 1}. Then d(A, B) = 1 yet d(a, b) > 1 for all
a E A and b E B. If these examples seem surprising, it is because when we think of
point sets we generally think of closed bounded figures, and in that case one can
show (in a course in analysis or general topology) that there is always a closest pair
of points. (In fact, it suffices that only one set be bounded so long as both are
closed. Also, if both sets are convex, one of them strictly, then the pair of closest
points is unique.)
If A = {a} and B= {b}, then clearly d(A,B) = d(a,b). If A = {a} and B is a
straight line, then there is a closest point in B to a, namely the foot of the
perpendicular to B from a; so by our definition, d(A,B) is the length of this
perpendicular. But this length is also precisely the special definition usually given in
texts for the distance from a point to a line. Thus our general definition does
include the two special cases mentioned earlier.

(3, -4)

(A) (B)
Figure 1. Distance from a point to a circle.

To develop a feel for the general definition, we begin by letting C be (the point
set of) the circle x2 + y2 = 9. What is the distance from a = (1,2) to C and from
b = (3, -4) to C? Draw radii of C through these points; see Figure 1(A). This
suggests that d(a, C) = 3 - d5 and d(b, C) = 5 - 3 = 2. The smaller circles drawn
around a and b in Figure 1(B) make clear that we do have the right answers,
because no other points of C are as close to a and b respectively as the points of
tangency of these two new circles to C.
It is equally easy to find the distance between a point and, say, a square or a
triangle. However, as soon as the point set C becomes slightly more complicated,
the problem gets much harder. For instance, what is the distance between (0, 1) and
the parabola y = x2? This is hard to answer without calculus. One can do it without
calculus if one is given that the slope of y = x2 at (x, x2) is 2x, and that the segment
from a point a to its closest point b on a smooth curve without endpoints (e.g., C)
is perpendicular to the curve at b. The answer in this case is V3 /2 and there are two
closest points on the parabola, (? /2 , 1/2).

147

This content downloaded from 130.58.65.20 on Wed, 19 Oct 2016 15:23:04 UTC
All use subject to http://about.jstor.org/terms
 What is the distance between two circles if (1) they are external to each other, or
(2) one is inside the other, or (3) they intersect? Hint: draw some pictures and
connect the centers by line segments.

(2, 3)

(0,1)

(2,0) (3,0)

Figure 2. Distance between two line segments.

Now let us take lines and line segments for our sets. Clearly, the distance
between two lines is 0 if they intersect and equals the length of any perpendicular
between them if they are parallel. What is the distance between two line segments?
That depends! In Figure 2, the distance is 3 2 and the closest points are given as
large dots. Can you see that, so long as the two segments are not parallel and don't
intersect, at least one of the unique pair of closest points is an endpoint of its
segment?
Finally, what is the distance between the parabolas y = x2 + 1 and x =y2 + 1?
One can answer this using algebra and symmetry if one again knows that the slope
at (x, y) on the first parabola is 2x. Note: the parabolas are symmetric to each
other through the line y = x.
Now let's do some locus problems. The simplest sort is: describe the locus of all
points distance d from some set A. If A is a point a, we have a standard easy
problem; the answer is the circle of radius d with center a. But what if A is itself a
circle of radius r and center a? The answer depends on whether or not d < r. If it is,
then the locus has two components, a circle of radius r + d and a circle of radius
r - d, both with center a. If r < d (or if A is the disk of radius r and center a, i.e., A
is all points on or inside the circle), then the locus of points distance d from A is just
the circle around a of radius r + d.
What is the locus of points distance d from a square or a triangle? The latter
problem is illustrated in Figure 3. The original triangle, drawn thicker, is equilateral
of side s, with d < sVY /6. By picking various polygonal sets for A, one can get
some very pretty pictures as loci-and some good area problems. For instance, what
is the area of the region between the two components of the locus in Figure 3?

148

This content downloaded from 130.58.65.20 on Wed, 19 Oct 2016 15:23:04 UTC
All use subject to http://about.jstor.org/terms
 Figure 3. Points a fixed distance from a triangle.

Again, once one allows A to be curved, these questions become much more
difficult. For instance, what is the locus of all points distance 1 from the parabola
y - x2? Here too the locus has two components, below the parabola and above it.

Figure 4. Construction of all points a fixed distance below a parabola.

The component below is easily described in words: for every point on the parabola
draw an outward facing perpendicular of length 1; the points on the far ends of
these segments form the component. See Figure 4. However, the same construction
with inward facing perpendiculars gives too many points. For instance, it gives (0, 1)
at the end of the perpendicular from (0,0 ), and we have already remarked that (0, 1)
is not distance 1 from this parabola. The asymmetry between the inside and outside
components of the locus is due to the fact that the parabola is concave up, not
down. Question: Is the component of the locus below the parabola another
parabola? A much harder question (but still doable by elementary means given the
information about slopes on this parabola stated earlier): What points above the
parabola are on the locus?

149

This content downloaded from 130.58.65.20 on Wed, 19 Oct 2016 15:23:04 UTC
All use subject to http://about.jstor.org/terms
 Now let us consider the locus of points equidistant from two sets A and B. The
locus is well known if A and B are both points (it's the perpendicular bisector of the
connecting segment) and if A is a point and B a line (a parabola). But there are
many more possibilities to consider. Here are some interesting ones.
1. Two intersecting lines
2. Two rays with the same origin (be careful!)
3. Two line segments
4. Two mutually external circles
a. of equal radius
b. of different radii
5. A circle and a line outside it
6. A circle and an intersecting line
7. Two intersecting circles
8. A square and a straight line outside it
To give an indication of how to do these, let us sketch solutions to 6 and to the
special case of 8 when the line is parallel to a side of the square.

The answers are shown in Figures 5 and 6. In Figure 5, the circle C has radius r
and center a, and L is the original intersecting line. M and N are parallel lines
distance r on either side of L. The locus is the union of two parabolas, one with
focus a and directrix M, the other with focus a and directrix N. One way to show
this is by considering four cases for where a locus point might be: either left or right
of L and either inside or outside C. The union of the points from the four cases,
along with the two points of intersection of L and C, gives the complete locus. We
illustrate two of the cases. If p is outside C and right of L and equidistant from
both, then p is equidistant from the point a and the line N, hence on the second
parabola mentioned above. If p is inside C and right of L, then p is distance d from
both sets iff it is distance r - d from a and r - d from M. Hence this case gives
those points on the first parabola inside C. And so on.

For problem 8, the curve PQRSTU in Figure 6 is the locus. Segment RS is a

straight line segment, halfway between AD and L. Segment QR is part of a
parabola with focus A and directrix L. Segment PQ is linear; it is halfway between
side BA of the square and L. The infinite section of the locus left of P is part of a
parabola with focus B and directrix L. The lower segments of the locus are
described similarly. Why is it that no point on the locus has its closest point of the
square inside side BC?
Finally, although we have generalized the sets from which we find distances, we
have still been looking for loci of points with distance properties relative to these
sets. We could just as well ask for loci of figures. That is, we could pose problems
like: Find all circles distance d from a given circle C of radius r. Or, find all circles
equidistant from two given circles. The solution to the first problem depends on the
relative sizes of d and r. We state the solution when r > d; there are more circles in
the locus in this case. Let K and K' be circles concentric with C of radius r - d and
r + d respectively. The circles distance d from C are: K, all circles internally tangent
to K; K', all circles with K' internally tangent to them; and all circles externally
tangent to K'. See Figure 7, where C is drawn heavy, K and K' are dotted, and the
other circles are drawn normally. Why is this the (entire) solution?

150

This content downloaded from 130.58.65.20 on Wed, 19 Oct 2016 15:23:04 UTC
All use subject to http://about.jstor.org/terms
 P L
N L M

B A R

C C D

Points equidistant from line L and circle C. Points equidistant from a line and a square.
Figure 5. Figure 6.

Figure 7. Circles a fixed distance from circle C.

151

This content downloaded from 130.58.65.20 on Wed, 19 Oct 2016 15:23:04 UTC
All use subject to http://about.jstor.org/terms
 Figure 8. A circle externally equidistant from Cl and C2.

As for finding all circles equidistant from two given circles, let us take the case
where the circles Cl, C2 are mutually external, with radii r1 > r2, and with d(C,, C2)
> rp. Let a,, a2 be the two centers. The circles we seek, equidistant from these two,
are of several types.
(1) Distance 0 from both Cl and C2, i.e., any circle intersecting both.
(2) Circles D inside C1. There are none such equidistant from Cl and C2, since
for any such D, d(C1, D) < r, and d(C2, D) > d(C2, C1) > r,.
(3) Circles inside C2. Again there are none.
(4) Circles D external to both C1 and C2. Let b be the center of such a circle
and let R- be its radius. As Figure 8 shows,

d(b,a,) - R - = d(C ,D)= d(C2,D)= d(b,a2)- R - r2.

Thus

d(b,a,) - d(b,a2)= r- r2.

That is, the locus of centers b of such circles D is one branch of a hyperbola with
foci al and a2! R is any value which keeps D external to both C1 and C2.
(5) Circles D containing C1 and external to C2. With a1, a2,r1,r2,b, R as before,
we see from Figure 9 that

R - d(b, a,) - = d(D, C1) = d(D, C2) = d(b, a2)- R - r2.

So

d(h-a1) + d(b, a2) = 2R + r2 - r > R + r2.

152

This content downloaded from 130.58.65.20 on Wed, 19 Oct 2016 15:23:04 UTC
All use subject to http://about.jstor.org/terms
 Figure 9. Another circle equidistant from Cl and C2.

That is, the centers of all such circles D of radius R are on an elipse with foci a,
a2 and sum of lengths 2R + r2- r,.
(6) Circles external to C1 and containing C2. Similar to (5).
(7) Circles D containing both C1 and C2. Let the reader show that the centers
of such circles are on the other branch of the same hyperbola as in (4). The reader
should also verify that there are no further cases.
One can easily make up many more challenging locus problems of these sorts. Or
add additional twists. For instance, a referee has suggested asking all the same
questions again with different definitions of distance.
Literature: the author is unaware of any previous literature on the whole class of
locus problems discussed here-those which arise from the definition of distance
between general sets. For more traditional locus problems, consult any "classical"
text in high school or college geometry, e.g., [1, Section 1.2] or [2, Ch. 18].

REFERENCES

1. N. Altshiller-Court, College Geometry, Johnson Publ., Richmond, Va., 1925.

2. A. W. Weeks and J. B. Adkins, A Course in Geometry, Ginn & Co., Boston, 1961.

153

This content downloaded from 130.58.65.20 on Wed, 19 Oct 2016 15:23:04 UTC
All use subject to http://about.jstor.org/terms