Hojoo Lee Inequalities Through Problems
Hojoo Lee Inequalities Through Problems
VOLUME 6
Solutions to the
'Cover Problems'
in Volumes 1 to 5
BY
1906 - 1988
A C B
An arbelos Is formed by erecting semicircles on
I.
I
Cover problem dedicated to Sam Greltzer.
I
,
I Copyright © 1988
Committee on the American Mathematics Competitions
t Mathematical Association of America.
ii -- ---------------'--
A Tribute to Professor Samuel L Greitzer
A Second Diversion , 17
Another Diversion 29
Take a Break 42
Epilogue 47
ii
Professor Greitler, who was the first chairman of the U.S.A. Mathe
matical Olympiad Committee, died on February 22, 1988. These notes on
the cover problems of "his journal", the Arbelo., being published posthu
mously, are his last known mathematical writings.
Among those of his friends and others who, because of their great
respect for his work, transformed Sam's handwritten notes to this publica
tion were: Dr Stanley Rabinowitz, Alliant Computer Systems Corporation
[author of the cover problem for this special issue]; and Professors Robert
Bumerot, Hofstra University; Darrell Horwath, John Carroll University;
Murray Klamkin, University of Alberta; Walter Mientka, University of Ne
braska; and Leo Schneider, John Carroll University.
iii
The November 1982 Arbelo" Cover Problem 1
During the brief life-span of the student journal, Arbelo", the cover
problems produced the majority of correspondence. There were a few so
lutions and many requests for solutions. Since some of th~se involv~d fine
points in geometry, I have prepared these solutions in the hope that I may
be able to advance the cause of an important branch of mathematics that
has b~en shabbily treated.
Now to the task. The first issue was actually named Aftermath. The
cover had a formula for the volume ofa tetrahedron, which I will not discuss.
A o c B
IFigure 1.1
The November, 19S2 issue (the first Arbelo") introduced the Arbelos,
which is the region bounded by the semicircles in Figure 1. Archimedes
loved to play with the diagram and discovered many properties of it. Since
all the curves are semi-circles, LAXC = LATB = 90°, so C X and YT
are parallel. Similarly, for CY and XT. Thus, TXCY is a rectangle.
Therefore, CT and XY are equal and bisect each other at O.
Since 6.0XC is isosceles, LOXC = LOCX. Draw radius DX. Then
6.CDX is also isosceles, so LDXC = LDCX. Adding,
LOXD = LOXC + LCXD = LOCX + LXCD = 90°.
Hence XY is tangent to arc AXC, and, in the same way, to arc CY B.
Next, let AC = a, CB = b. The area bounded by the three semicircles
will be
1rab
4
The area of the circumcircle about TXCY equals 1r(TO)2 = ~~~ since
TC = =
v;ib so that TO ~v;ib. Thu", the area within the Arbelo" equal"
the area of the circle with TC a" diameter.
2 The January 1983 Arbelol Cover Problem
T - -__
A D c p
IFlaure 2.l
I\ Now consider the cirde with radius :z: tangent to the large arc, the
I tangent line TC, and the small arc with radius, say ri. (See Filure 2.)
Imagine the right triangle E F D formed by dropping a perpendicular
I EF from E, the center of the cirde with radius :z: to DC. Note that
cos LEDF
DF
DE
ri -:z:
= --- = --.
ri +:z:
Let the large semicirde have radius rand P be the midpoint of AB. Apply
the Law of COlinel to b.EPD to obtain
We have omitted the other semicirde. If it has radius r2, we can rewrite
the above as
The cover page of the January, 1983 Arbelol was intended to illustra.te
the relative values of the arithmetic mean (A), the geometric mean (g), the
Tbe January 1983 Arbelo, Cover Problem 3
o C B
IFilure 3.1
harmonic mean ('H) and the root-mean square ('R.). The diagram was not
clear, and I have taken the liberty of presenting another. (See Filure 3.)
2ab
ab = (TC)2 =(TF)(TO) = (TF) (a; b) or TF= - - = 'H.
a+b
AR= Ja +2 2
- 2 b- ='R.,
1 1
,
, ! 1
Ii 1~-2
1~=~
; ;
/~
///1_13 a
17'6 4 1
1~10
etc.
10 5 1
1~ 15 20 15 6 1
I
I The November 1983 Arbelo8 Cover Problem 5
Assume
so that
c
p
Prove: PQ - PR
and: PQ.l PRo
Jan 1984
The problem on the cover ofthe January, 198-4 Arbelo, is probably the
most provocative of all the cover problems. We can solve this problem by
using U A geometric figure may be moved in 'pace without changing ,ize or
,hape" in a judicious manner.
Now to the problem: On the side BC = a of 6.ABC we have con
structed equilateral triangle BCT and drawn PT. This is a crucial step,
and I was led to it alter a number of unsuccessful approaches. Now we have
four similar triangles: ARB, AQC, TPB, TPC. (See Figure 5.) If the
sides of 6.ABC are a, 6, c, then
IFigure 5.1
Now we rotate 6.RBP 45° clockwise about B. Then R falls on AB,
and P falls on BT. If B is held fixed and the rotated triangle is multiplied
by!, the shifted and expanded RP will coincide with AT.
u
Similarly, we rotate 6.QCP 45° counterclockwise about C and expand
by ! holding C fixed so that the shifted and similarly expanded 6.QCP
u
will coincide with 6.ACT.
Since RP and QP coincide after these transformations, they must have
been equal initially. Also, since each was rotated through 45° and were then
coincident, RP and QP must have been perpendicular to start with.
8 The March 1984 A rbelos Cover Problem
The cover on the March, 198i A rbelol has some interest because of its
supposed history. We'll sneak up to it. (For extensive generalizations and
references see Cruz Mathematicorum 6(1980) ppl85-l87.)
Suppose that, on the sides of 6.ABC, we constrllc.t6.ARB, 6.BPC
and 6.CQA and that LR + LP + LQ = 180°. (See Figure 6.) Then,the
circumcircles about 6.ARB, 6.BPC and 6.GQA intersect at a common
point OJ for, if we let 0 be the intersection of the circumcircles of 6.ARB
and 6.GQA, then
Hence
--~--------...,\Q
R I
I
II .
!I
--~~-=:-------~Q
R
IFi,ure 1.1
For the Napoleon triangle, .6.ARB, .6.BPG, and .6.CQA are equilat
eral. Hence, in .6.AZY, by the Law 0/ Co"inu,
But ZA
V3
= -32 . -·-BA 2 V3
AY = - . -AG and
2' 3 2 '
\
I
:
'
l
\ .
So
(ZY) , - (ZY)
- - , = 2(BA)(AC)
-_._....- [cos(LBAC - 60 0 ) - cos(LBAC + 60 0 )] •I
3
= ~(BA)(AC)Sin LBACsin 60 0
= ~[ABC]·.
Similarly,
, - -, 2(AB)(BC). 4 [
(X Z) - (X Z) = -J3- S10 IABC = J3 ABC]
and
since
v'i , V3 .. , --
[ABC] = -.("(XZ) - -4'(XZ) = [XYZ]- [XYZ], the difference
in area between the outer and inner Napoleon triangles equals the area
of the triangle ABC.
A First Diversion
we obtain
(2) A triangle con~i~t~ of three line~ not on it point together with the three
point~ on thue line8.
· sI'19 hi
We get somet h mg t Yd'lnerent
lr f rom "A {qUadrangle} con~i~t~
quad II'1a t eraI
of four { Pl·oints } no three on one { lin~ t } together with the {lin~St }
mes pom pom s
on them." .
Use the upper terms and you have a statement. Use the lower terms
and you have its dual. The diagrams are in Figure 9.
quadrangle quadrilateral
4 points, A, B, C, D 4 lines, a, b, e, d
6 lines, AB, AC, AD, 6 points, ab = 1, ae = 2, ad = 3,
BC, BD, CD be = 4, bd = 6, cd = 6
Note the differenee8.
IFllure 9.1
label one line without removing it. We agree to call this line the line at
infinity.
First, every line meets the line at infinity, so every line has a "point" at
infinity. Next, if two lines are coincident at the line at infinity, we shall call
these lines parallel. This permits us, for example to work with "parallelo
grams". (See Filure 10.) Quadrilateral ABCD is a parallelogram.
--'Ilr-----------#'-line at infinity
IFigure 10.1
14 The May 1984 A r~lo, Cover Problem
D~ _ _ -r ~ C
Ma 1984
,"
Sometimes it is more useful to use the projective plane rather than the
Euclidean plane in attacking a problem. Thus, for the cover problem of the
Ma.y, 1984 Arbelo8, we might use Figure 11.
The May 1984 Arbelo. Cover Problem 15
IF'lure 11.\
A B C
--.........--~--.......,~--L
"Let A, B, C be on a line L, and A', B', C' be on another line L'. Then
the intersections (AB')(A' B), (BC')(B'C), and (AC')(A'C) lie on a line."
l!!i'!~
•:
B L'
"Let A, B, C be lines on point L, and A', B', C' be lines on point L'. Then
the lines OJI (AB')(A'B), (BC/)(B'C), and (AC')(A'C) lie on & point (are
concurrent). "
IFijure 13·1
y- 0 d- 0 -d
line DS: --=--=- or dz +ay= ad
z-a O-a a
y- 0 c- 0 -c
line BP:
:r:-b = O-b = b'
or ez + by = be
y-e
line CT: z_ a = d-e
b_ ~' or (d - e)z + (a - b)y =ad - be.
Since the equation for line CT is the difference between the equations
for lines DS and BP, these three lines are concurrent.
Which is a better proof is problematic, although I would prefer the
projective (or affine) proof. Either way, they are "grown-up" proofs.
A Second Diversion 17
A Second Diversion
Most mathematicians think that geometry ended with Euclid (ea. 300
B.C.) or perhaps Pappus (ea. 300 A.D.). We are going to look at something
invented, I am told, in 1831 by L. J. Magnus - inver.!ion.
Briefly, let 0 be a circle with radius r. (See Figure 14.) If tangents are
, drawn from a point P and if line TU intersects OP at pi, we say that
i
J
points P, pI are inverse.!. Notice that op x OP' =
r 2 • We say 0 is the
I center of inversion.
I
j
It is not hard to see that a circle tangent to the cirde of inversion
and passing through the center of inversion inverts into a straight line, PAt
tangent to the cirde 0 at A. In fact, using similar triangles, one can show
18 A Second Diversion
that any circle passing through the center of inversion inverts into a line
P A and conversely.
All we need now is one more result-a eirde cutting the eirde of in
version at rightangles inverts into itself. (See Figure IS.) Remember that
if a tangent OT and a secant OPP' are drawn, then (OT)2 = (OP)(OP').
P'
IFigure 15·1
It is not hard to see some advantage of inversion. Tangent curves
remain tangent after inversion, for instance. Also, the inverse of a figure
may turn out to be simpler than the original.
-A B c
, IFilure 16·1
,diametersAB and AC pass through the center of inversion, they invert into
paralleUines. (See Fisure 16.)
- .
Circles 0 1 and O 2 will then invert into equal circles tangent to the two
lines. Now just loolung shows that 02H2 is2d 2 • And in general, using
center A and a radius equal to the tangent from A to 0", we can similarly
show that O"H" = ndn.
You can't tell if inversion will help you solve a problem if you know
nothing about inversion. And you don't really have to know the fine points
of inversion. You should know it exists, at least. For further applications
of inversion, see Yaglom, Geometric Tran8formation8 MAA NML8, p8 or
Courant and Robbins What i8 Mathematic8? Oxford Univ Press, NY, 1941,
ppl.40-16.4.
The November 1984 Arbelo.. Cover Problem 21
1\
Nov 1984
A
----~
J
j.
I
C Q
IFIgure 17 ·1
22 The January 1986 Arbelo" Cover Problem
,
\ '
II I
I
I
II y
I
\.
I
x y
IFigure 18·1
The cover problem on the January, 1985 Arbelol has many segments,
but is not difficult.
In Figure 18, all the angles we need are labeled. An angle marked 12
means that it is the sum of angles 1 and 2; an angle marked 1123 is the
sum of angles 1 + 1 + 2 + 3, etc.
Triangle ZEF is (by symmetry) isosceles. Now, from anyone of the
six similar triangles, (see 6CXY), 111223 = 180°. But LCAE+ LEFC =
1123 + 123 = 111223 = 180°. Hence C, A, E, Fare concyclie. By synune
IFigure 19.1
What I did then was to construct a small square near B with a side
on BC and a vertex on AB. Then let P be the intersection of AC with the
line through B and the vertex of the square inside 6.ABC. The point R
is on BC with PR.l BC, and Q is on AB with QP II BC. From similar
triangles, we easily find that PQ = P R and we have three vertices of the
desired square.
To make it look mysterious, we can expand the square until P R equals
altitude AH. Now we have a construction where we construct AP equal to
and perpendicular to altitude AH and draw BP intersecting the side AC at
a vertex of the desired square. (This is a simple application of homothetic
figures.)
The March 1985 Arbelo8 Cover Problem 25
I
!
!
i
I E c
I'
A~--------""7""-----~
Mar 1985
For the cover problem of the March, 1985 Arbelo8, we can proceed as
follows (See Filur. 20.) :
A
The May 1985 Arbelol Cover Problem
Ma 1985
The May 1986 Arbelol Cover Problem 27
~
a b
IFigure 2C]
The cover problem for the May, 1986 Arbelol, which uses simple results
to obtain a complicated result, is quite pretty. Recall that the area of a
triangle with perimeter 2. and inradius r equals r•. *
Let (J = LDAH = LHDC. In ~AHD we see that
Adding, rl + r'J + ra = H D.
There is another solution, depending on a rare datum about right
triangles. (See Filure 21.)
* This can easily be proved by partitioning the triangle into three trian
gles using the line segments from the center of the cir~e to the vertices of
the given triangle. The altitude of each of the three triangles is the inradius
of the original triangle.
28 The May 1986 Arbelos Cover Problem
A rare datum: In any right triangle, the sum 0/ the leg, minus the
hypotenu,e equal, the diameter 0/ the in,cribed circle. In Figure
21, the lines shown with equal letters are equal. The assertion is
that
(a + r) + (b + r) - (a + b) = 2,..
2rl =AH+HD-DA
2r, = C H + H D - DC
2ra = (DC + DA) - (AH + CH)
2(rl + r, + ra) = 2HD
so that rl + r, + ra = H D.
Another Diversion
IFilure 22·1
One diagram I found relaxing and instructive is Filure 22: t::.ABC
is a triangle inscribed in a circle with center O. Altitudes AU, BV, CW
have been drawn intersecting at orthocenter H. Points U, V, Ware joined
forming the orthic triangle UVW.
I have extended AU, BV, CW as shown and added some dotted lines.
Now we gaze and find, in order, the following results:
(1) From right triangles ABV and ACW,
LABV = 90 0
= LACW.
- LBAC
VH = CH or BH·HV=CH·HW =AH·HU.
=
Similarly AC· BH 2R· UW and AB· CH 2R· UV. Adding,=
we have
BC· AH + AC . BH + AB . C H = 2Rp
where p = perimeter(~UVW). However, we may write this as
BC· AU - BC . HU + AC . BV - AC . HV + AB . CW -AB . HW
=2Rp
3· 2[ABC] - 2[ABC] = 2Rp.
This means 2[ABC] = Rp, and the perimeter of the orthie triangle
[ABC]
equals 2 -- .
R
I stop here, although I did get more. Perhaps the reader might try it
on other diagrams. It's better than contemplating one's navel.
DF-----*----~
B-"::;~""""L.-----------=-
P
c
Given: PQR is the orthic triangle of ABC.
(Le., AP, BQ, CR are altitudes.)
AX, BY, CZ are perpendicular to
QR,RP,PQ respectively.
Prove: AX, BY, CZ are concurrent.
Nov 1985
With the help of our "another diversion", ~esult (6), we can just toss
oft' a solution to the cover problem on the November, 1985 Arbelo8. It shows
that AX, BY and CZ are parts of radii of the circumcircle ofAABC.
i
I
i'
I
34 The J...uary 1986 Ariel•• Cover ptb1em
c
c
&. -------.JI E
A o
Given: ABCDE is inscribed in the semicircle
on AE with unit radius OA.
AS-a, BC-b, CD-e, DE-d.
Prove: a 2 + b 2 + c 2 + d 2 + abc + bcd < ...
Jan 1986
With the cover problem on the January, 1986 Arbelo", we are back to .;
complicated mathematics.
c-.---__
A&....-------------~E
1 o 1
IFi,ure 24.'
The January 1986 Arbelol Cover Problem 35
Z2 = a 2 + b2 - 2ab cos B
y2 = c2 + d 2 _ 2cd cos D
Thus,
4 = a 2 + b2 + c2 + d 2 - 2(abeos B + cd cos D)
= a 2 + b2 + c2 + d 2 + aby + cdz
> a 2 + b2 + c 2 + d 2 + abc + cdb.
One can also show that
i
,.
!
36 The March 1986 A rbelo8 Cover Problem
'"
Mar 1986
I'
IFigure 25·1
,
i
I
r
38 The May 1986 Arbelo8 Cover Problem
l'
c r
B
Ma 1986
In the case ofthe cover problem on the M41/, 1986 Arbelo8, there is a
hint-e:onstrud the triangle on the given lines first.
p~~~----*::'--T
l!!lure 26·1
Construd LAPT = 60° and LBPT = 60°. Let APBC be the circum
circle of 6.APB. Then since LABC and LAPC both subtend the same
arc, LABC = LAPC = 60°. Similarly, LBAC = LBPC = 60°, so 6.ABC
is the desired equilateral triangle. (See Figure 26.)
Now a little trigonometry.
!.
-
10 . 0 v'3
=SlR60 =
20
I
i AP 2 ' AP = v'3'
, ,
I
6 .
-=slR60° =v'3
BP= ~
BP 2 '
2 400 144 20 12 °
(AB) = - + - -2· - . -.cos120
3 3 V3 v3
400 144 240 784
=3- +3- +3- = 3 .
3 V".
For a generali.ationto tetrahedra and simplices, see Cruz Mathemati
corum, 13(1987), pplfO-lf,j.
I
,
, I
I'
40 Still Another Diversion
B c
IFigure 27 ·1
.. CQ [BOC] AR [COAl
Now let ullook at the cover problem for the September, 1986 Arbelo,.
We merely compare areas of trianglel with equal altitudes.
Since AX II BY, [XBY] = [ABY]. Similarly, BY II CZ impliel
[ZBY] = [CBY]. It followl that [XBY] + [ZBY] = [ABC].
Next since XA II ZC, [ZXG] = [ZAC], so
A
IFisure 28·1
• For example,
BA = 1 - CA = ....-;-
-:;;-;;- AC + -.._ CA- = __--"'
BC - __- AB- +.-.- CB
OA - -:;;;-;
DC DC.AD DA CD AC DC CA
I.
I
I·
.
B ~--E----""'--'C
Nov 1986
Now we can get to the cover page for the November, 1986 Arbelol.
Since AH is a diameter of the circle on ADHI, sin A = ~~. (Why?)
The cover problem for the January, 19S7Arbelo, comes from KVANT,
a Russian secondary school journal in mathematics and physies which is
excellent for those who can read Russian.
We will sh~w that P lies on diagonal AC by showing that LAPK +
LKPC = 180°. Since angles AMK and APK subtend equal ares they
are equal. Since angles K N C and K PC are opposite angles of an in
scribed quadrilateral, they are supplementary. Since AM II NC, LAM K =
LKNC. Thus,
I.
I
j
I
"
As for the cover problem for on the March, 1987 Arbelo6, we have
two triangles on the same base Be, inscribed in the same circle, but with
separate orthocentei•.
i·
I
IFigure 29·1
46 The May 1987 A rbelol Cover Problem
B x C
"
B H X c
I';Iure 30·1
,
I
Epilogue
I•
I admit that the cover problems have varied from very .imple to diffi
cult. My excu.e for presenting them is that I found them fun, and I hoped
that my readers would find them fun. We got out of step when proponents
of the late "new mathematics" started fussing with the distinction between
number and numeral, greater and larger, and so on. I may be wrong, of
course, and this nicety in expression is valuable. Perhaps the time will come.
when, instead of teaching, "the dog runs after the cat", we will become pre
cise and teach, "cani, familia", pur,ue, feli, domelticu,." It may be that
if we teach reading this way, English will eventually reach that happy state
in which mathematic. finds itself today.
Anyway, I hope you like the problems and solutions.
Samuel L. Greitzer