Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
EDITORS
Curtis Cooper Shing S. So
CMJ Problems CMJ Solutions
Department of Mathematics and Department of Mathematics and
Computer Science Computer Science
University of Central Missouri University of Central Missouri
Warrensburg MO 64093 Warrensburg MO 64093
email: cooper@ucmo.edu email: so@ucmo.edu
This section contains problems intended to challenge students and teachers of college mathematics.
We urge you to participate actively BOTH by submitting solutions and by proposing problems that are
new and interesting. To promote variety, the editors welcome problem proposals that span the entire
undergraduate curriculum.
Proposed problems should be sent to Curtis Cooper, either by email as a pdf, T
E
X, or Word
attachment (preferred) or by mail to the address provided above. Whenever possible, a proposed prob-
lem should be accompanied by a solution, appropriate references, and any other material that would
be helpful to the editors. Proposers should submit problems only if the proposed problem is not under
consideration by another journal.
Solutions to the problems in this issue should be sent to Shing So, either by email as a pdf,
T
E
X, or Word attachment (preferred) or by mail to the address provided above, no later than April 15,
2012.
PROBLEMS
966. Proposed by Tahani Fraiwan and Mowaffaq Hajja, Yarmouk University, Irbid,
Jordan.
Let P be a point in a plane of ABC which is not on the sidelines AB, AC, or BC and
let cevians AA
, BB
, CC
C and
A
is a midpoint of BC.
967. Proposed by Elias Lampakis, Kiparissia, Greece.
A point (x, y) on the plane with both coordinates x, y rational is called rational else,
it is called irrational. Let n 2 be a positive integer and C(Q) be the set of rational
points on the parabola C : y
2
= 4x. Prove that the tangent lines to C at every (x, y)
C(Q) {(0, 0)} intersect the curve K : y = x
n
at irrational points.
968. Proposed by Greg Oman, The University of Colorado, Colorado Springs, Col-
orado Springs, CO.
Let C(R) be the ring (under pointwise addition and multiplication) of continuous func-
tions f : R R. Suppose that S is a subring (not necessarily with 1) of C(R) which
contains only monotone functions (the monotonicity need not be strict). Must every
member of S be a constant function? Prove or provide a counterexample.
http://dx.doi.org/10.4169/college.math.j.43.1.095
VOL. 43, NO. 1, JANUARY 2012 THE COLLEGE MATHEMATICS JOURNAL 95
969. Proposed by Panagiote Ligouras, Leonardo da Vinci High School, Noci, Italy.
The lengths of the sides of the hexagon ABCDEF satisfy 3AB = BC, 3CD = DE, and
3EF = FA. Prove that
_
AF
CF
+
_
ED
AD
+
_
CB
EB
6
3
5
.
970. Proposed by Xiang-Qian Chang, Massachusetts College of Pharmacy and Health
Sciences, Boston, MA.
Let A be an n n real symmetric matrix and let (A) = n trace(A
2
) (traceA)
2
.
Show that
2
n
_
2
n
_
(A) max
1i, j n
|
i
j
|
_
2
n
(A), n even;
2
n
2
1
(A), n odd.
SOLUTIONS
Evaluating an integral limit
941. Proposed by Ovidiu Furdui, Cluj, Romania.
Let f : [a, b] [0, 1] be an increasing function which is differentiable at b with
f (b) = 1 and f
(b) = 0 and let g be a bounded function on [a, b] with lim
xb
g(x) =
L. Find the value of
lim
n
n
_
b
a
f
n
(x)g(x) dx.
Solution by Armstrong Problem Solvers, Armstrong Atlantic State University, Savan-
nah GA; Paolo Perfetti, Dipartimento di Matematica, Universit` a degli Studi di Tor Ver-
gata, Rome, Italy; and WilliamSeaman, Albright College, Reading PA (independently).
First, let f (x) = x and
g(x) =
_
_
1 if x [0,
1
2
] Q
1 if x [0,
1
2
] \ Q
0 if x (
1
2
, 1]
.
Then, for each n, the integral
_
1
0
f (x)
n
g(x) dx is not Riemann integrable.
Next, let f (x) = x and g(x) = (1
A
(x))(x 1) where A is a non-measurable
subset of [0, 1] and
A
is the characteristic function of A. Then g is non-measurable
and hence f
n
g is non-measurable.
Based on the facts above, we add the condition that
_
b
a
f (x)
n
g(x) dx exists for
each positive integer n. Let m = f
(b) and s(x) =
f (x)f (b)
xb
. Then lim
xb
s(x) = m.
Suppose m < 0. Then there exists a c [a, b) such that
3
2
m < s(c) <
1
2
m < 0, and
hence f (c) > f (b). This contradicts f being increasing. Thus, m > 0.
Let |g(x)| < M for all x [a, b]. For every > 0, since s(x) m and g(x) L
as x b
n
_
b
a
f
n
(x)g(x) dx
< nf
n
(b )M
_
b
a
dx
and f (b ) < 1,
lim
n
n
_
b
a
f
n
(x)g(x) dx
= 0.
Let us use the second term of (1) to compute the desired limit. Note that f (x) =
s(x)(x b) +1, and assume that L 0. For all x [a, b], we have
(m +)(x b) +1 < f (x) < (m )(x b) +1 (2)
and
n
_
b
b
[(m +)(x b) +1]
n
(L ) dx < n
_
b
b
f
n
(x)g(x) dx (3)
<
_
b
b
[(m )(x b) +1]
n
(L +) dx.
The bounds of n
_
b
b
f
n
(x)g(x) dx in (3) become
n
n+1
L
m
(1 r
n+1
) where r = 1
(m ) and 0 < r < 1. Thus, if L 0, then
L
m +
lim
n
n
_
b
a
f
n
(x)g(x) dx
L +
m
.
Since > 0 is arbitrary, lim
n
n
_
b
a
f
n
(x)g(x) dx =
L
f
(x)
for L 0. If L < 0, then
the inequalities in (3) change direction, but the result lim
n
n
_
b
a
f
n
(x)g(x) dx =
L
f
(x)
remains unchanged.
Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; MICHEL BATAILLE, Rouen, France; BILL
COWIESON, Fullerton C.; JAMES DUEMMEL, Bellingham WA; EUGENE HERMAN, Grinnell C.; KIM MCIN-
TURFF, Santa Barbara CA; JOEL SCHLOSBERG, Bayside NY; PRAVEEN VENKATARAMANA (home schooled
student); ALBERT WHITCOMB, Castle Shannon PA; YAJUN YANG, Farmingdale State C.; and the proposer.
VOL. 43, NO. 1, JANUARY 2012 THE COLLEGE MATHEMATICS JOURNAL 97
An equation in geometry
942. Proposed by Geoffrey Kandall, Hamden CT.
Suppose ABC is any triangle with angle bisectors AA
, BB
, and CC
meeting at the
incenter I . Let A
= IA B
, B
= IB A
, and C
= IC A
. Prove that
IA
A
+
IB
B
+
IC
C
= 1.
Solution by John Mason, Malvern PA and Richard Pefer, San Jose State University,
San Jose CA (independently).
We shall prove the following generalization: Let I be any point in the triangle
ABC and let A
, B
, and C
, B
, and C
be the intersections of
AI and B
, BI and C
, and CI and A
, respectively. Then
IA
A
+
IB
B
+
IC
C
= 1.
Using signed distance,
IA
A
=
IA
A
as {A, I, A
, A
BC,
IA
A
AB
BC
C
CI
= 1. (1)
Applying Cevas Theorem on triangle AC
IB , and
AC
concurrent at B
,
IA
A
AB
BC
C
CI
= 1. (2)
Combining (1) and (2), we have
IA
A
=
IA
A
.
Similarly,
IB
A
=
IB
B
and
IC
C
=
IC
C
. Dropping the signed distances, we now need
to show that
IA
A
+
IB
B
+
IC
C
=
IA
A
+
IB
B
+
IC
C
= 1.
Dropping perpendiculars from points A and I to side BC at X and Y, we have
IA
A
=
IY
AX
by similar triangles, and
IA
A
=
IY
AX
=
IY BC
AX BC
=
2 Area(IBC)
2 Area(ABC)
=
Area(IBC)
Area(ABC)
,
and similar results can be obtained for the ratios
IB
B
and
IC
C
. Now, we have
98 THE MATHEMATICAL ASSOCIATION OF AMERICA
IA
A
+
IB
B
+
IC
C
=
Area(IBC)
Area(ABC)
+
Area(ICA)
Area(ABC)
+
Area(IAB)
Area(ABC)
= 1.
Thus, this problem is the special case that I is the incenter of triangle ABC.
Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; MICHEL BATAILLE, Rouen, France; MI-
HAELA BLANARIU, Columbia C. Chicago; CHIP CURTIS, Missouri Southern State U.; HABIB FAR, LSCS Mont-
gomery; DMITRY FLEISCHMAN, Santa Monica CA; THE GEORGE WASHINGTON UNIVERSITY PROBLEMS
GROUP; MICHAEL GOLDENBERG, The Ingenuity Project, Baltimore Poly. Inst. and MARK KAPLAN, Towson
U. (jointly); JOHN HEUVER, Grande Prairie AB, Canada; THE IOWA STATE PROBLEM SOLVING GROUP;
PANAGIOTIS KRASOPOULOS, Athens, Greece; RICK MABRY, LSU Shreveport; JERRY MINKUS, San Francisco
CA; PETER N
UESCH, Swiss Federal Inst. of Tech., Lausanne, Switzerland; JOEL SCHLOSBERG, Bayside NY;
SANGHUN SONG, Seoul Science High School; ERCOLE SUPPA, Teramo, Italy; RONALD TIBERIO, Natick
MA; PRAVEEN VENKATARAMANA (home schooled student); MICHAEL VOWE, Therwil, Switzerland; and the
proposer.
A cross ratio problem
943. Proposed by Juan-Bosco Romero M arquez, Universidad de Valladolid, Val-
ladolid, Spain and Francisco Javier Garca Capit an, Spain.
Given the triangle ABC, let D, E be points on line BC such that lines AD and AE
are isogonal. Let the internal bisectors of angles ADB and AEC meet lines AB and
AC at F and G, respectively. Show that the points X = BG CF, Y = DG EF and
Z = DF EG are on the internal bisector l
a
of angle BAC and satisfy the cross ratio
(AXY Z) =
_
AB
AC
_
2
:
_
BD
DC
_
2
.
B
C
D
E
Z
F
G
X
Y
L
A
Solution by Joel Schlosberg, Bayside NY.
By the angle bisector theorem and the Law of Sines,
AF
FB
BL
LC
CG
GA
=
AD
DB
BA
AC
CE
EA
=
sin
ABD
sin
DAB
sin
BCA
sin
ABC
sin
CAE
sin
ECA
=
sin
ABD
sin
ABC
sin
BCA
sin
ECA
sin
CAE
sin
DAB
= 1 1 1 = 1.
VOL. 43, NO. 1, JANUARY 2012 THE COLLEGE MATHEMATICS JOURNAL 99
By Cevas theorem, AL, BG and CF are concurrent, so X = BG CF is on l
a
.
Let Y
= EF l
a
and Y
= DG l
a
. Projecting l
a
to BC through F and then pro-
jecting BC to l
a
through G,
(AXY
Z).
Therefore Y
= Y
, so Y = DG EF is on l
a
.
Since DF and EG are external angle bisectors of ADE, Z = DF EG is an ex-
center of ADE and so is on the opposite vertexs angle bisector l
a
.
By the Law of Sines,
BE
AB
BD
AB
AC
EC
AC
DC
=
sin
BAE
sin
AEB
sin
BAD
sin
ADB
sin
AEC
sin
EAC
sin
ADC
sin
DAC
=
sin
BAE
sin
DAC
sin
BAD
sin
EAC
sin
AEC
sin
AEB
sin
ADC
sin
ADB
= 1 1 1 1 = 1.
Therefore
(AXYZ) = (BCED) =
EB/EC
DB/DC
=
_
AB
AC
_
2
:
_
BD
DC
_
2
.
Also solved by MICHAEL GOLDENBERG, The Ingenuity Project, Baltimore Poly. Inst. and MARK KAPLAN,
Towson U. (jointly); RICHARD PFIEFER, San Jose State U.; ALBERT WHITCOMB, Castle Shannon PA; and the
proposers.
The bounds of a denite integral
944. Proposed by Duong Viet Thong, Ha Noi University of Science, Vietnam.
Let f : [0, 1] R be a continuously differentiable function such that
_
1
0
f (x) dx = 0
and for all x [0, 1] we have m f
(x) M. Show that
m
12
_
1
0
x f (x) dx
M
12
.
Solution 1 by James Duemmel, Bellingham WA and
Angel Plaza, University of Las
Palmas de Gran Canaria, Spain (independently).
We will prove the following generalization of this problem for any interval [a, b].
The corresponding inequalities become
m(b a)
3
12
+
a +b
2
_
b
a
f (x) dx
_
b
a
x f (x) dx
M(b a)
3
12
+
a +b
2
_
b
a
f (x) dx.
First note
_
b
a
x f (x) dx =
_
x
a +b
2
_
f (x) dx +
a +b
2
_
b
a
f (x) dx.
100 THE MATHEMATICAL ASSOCIATION OF AMERICA
Since
_
b
a
_
x
a+b
2
_
dx = 0,
_
b
a
_
x
a +b
2
_
f (x) dx =
_
b
a
_
x
a +b
2
__
f (x) f
_
a +b
2
__
dx.
By the Lagrange Mean Value Theorem, for x
_
a,
a+b
2
_
,
f (x) f
_
a +b
2
_
= f
(c)
_
x
a +b
2
_
for some c
_
x,
a+b
2
_
. Since m f
(x) M for all x [a, b],
m
_
b
a
_
x
a +b
2
_
2
dx +
a +b
2
_
b
a
f (x) dx
_
b
a
x f (x)dx M
_
b
a
_
x
a +b
2
_
2
dx +
a +b
2
_
b
a
f (x)dx
_
m
_
x
a+b
2
_
3
3
_
b
a
+
a +b
2
_
b
a
f (x) dx
_
b
a
x f (x) dx
_
M
_
x
a+b
2
_
3
3
_
b
a
+
a +b
2
_
b
a
f (x) dx
m
(b a)
3
12
+
a +b
2
_
b
a
f (x) dx
_
b
a
x f (x) dx
M(b a)
3
12
+
a +b
2
_
b
a
f (x) dx.
The desired inequality follows from the facts that a = 0, b = 1, and
_
1
0
f (x) dx = 0.
Solution 2 Richard Stephens, Columbus State University, Columbus GA.
We will prove the following generalization of this problem: For k = 0, 1, 2, . . . , let
_
1
0
x
k
f (x) dx =
k
. Then, for n = 1, 2, . . . ,
1
n +k +1
_
mn
(k +2)(n +k +2)
+(k +1)
k
_
n+k
1
n +k +1
_
Mn
(k +2)(n +k +2)
+(k +1)
k
_
.
If
0
= 0, then
mn
2(n +1)(n +2)
n
Mn
2(n +1)(n +2)
and the desired inequality can be proved by setting n = 1.
This generalization is developed in the following manner. We rst multiply the
given condition m f
(x) M by x
k+1
(1 x
n
) and integrate over [0, 1]. Then
VOL. 43, NO. 1, JANUARY 2012 THE COLLEGE MATHEMATICS JOURNAL 101
m
_
1
0
(x
k+1
x
n+k+1
) dx
_
1
0
(x
k+1
x
n+k+1
) f
(x) dx
M
_
1
0
(x
k+1
x
n+k+1
) dx.
Using integration by parts for the middle integral above yields
mn
(k +2)(n +k +2)
(k +1)
_
1
0
x
k
f (x) dx +(n +k +1)
_
1
0
x
n+k
f (x) dx
Mn
(k +2)(n +k +2)
.
Solving for
_
1
0
x
k+1
f (x) dx =
k+1
derives our general inequality stated at the begin-
ning of our solution.
We can derive many other inequalities for bounds on
k+1
in terms of
0
,
1
,
. . . ,
k
. For natural numbers p, q, and r, we multiply the given condition m f
(x)
M by x
p
(1 x
q
)
r
and integrate over [0, 1]. It follows that
m
_
1
0
x
p
(1 x
q
)
r
dx
_
1
0
x
p
(1 x
q
)
r
f
(x) dx M
_
1
0
x
p
(1 x
q
)
r
dx.
Whenever p +qr = k +2, integration by parts will generate bounds on
k+1
in terms
of some subset of
0
,
1
, . . . ,
k
. If p = q = 1 and r = k + 1, then the bounds on
k+1
will involve all of
0
,
1
, . . . ,
k
. Since
m
_
1
0
x(1 x)
k+1
dx
_
1
0
x(1 x)
k+1
f
(x) dx M
_
1
0
x(1 x)
k+1
dx;
m(2, k +2)
_
1
0
k+1
j =1
(1)
j
_
k +1
j
_
x
j +1
f
(x) dx M(2, k +2);
m(2, k +2)
_
1
0
k+1
j =1
(1)
j
_
k +1
j
_
( j +1)x
j
f (x) dx M(2, k +2),
m(2, k +2) +
k
j =0
(1)
j
_
k +1
j
_
( j +1)
j
(1)
j
(k +2)
k+1
M(2, k +2) +
k
j =0
(1)
j
_
k +1
j
_
( j +1)
j
,
where (s, t ) =
_
1
0
x
s1
(1 x)
t 1
dx is the real Beta function. When k = 0 and
0
=
0, this also proves the desired inequality.
Also solved by AJAT ADRIANSYAH, Universitas Indonesia; GEORGE APOSTOLOPOULOS, Messolonghi, Greece;
EFRAIM ARMENDARIZ and MARK DANIELS (jointly), U. of Texas at Austin; ARMSTRONG PROBLEM
SOLVERS; MICHEL BATAILLE, Rouen, France; M. BENITO,
O. CIAURRI, E. FERN
x
3
= x
3
x
3
= 0, and so (x
5
x)
k
= 0 for all integers k 3. As a result,
x
5
= x because 0 = (x
5
x) +(x
5
x)
5
+(x
5
x)
9
= x
5
x. Now,
x
2
+ x = (x
2
+ x)
5
= x
10
+5x
9
+10x
8
+10x
7
+5x
6
+ x
5
= x
2
x + x
4
+ x
3
x
2
+ x
or x
2
+ x = x
4
+ x
3
. Thus, (x
2
+ x) x(x
2
+ x) (x
4
+ x
3
) + x(x
4
+ x
3
) =
0, that is, (x x
3
) +(x x
3
) = 0 and x
3
= x follows.
(c) We have just obtained x
3
= x when n = 3, 5. Let us check that it also holds
when n = 7, that is, when x + x
7
+ x
9
= 0 for all x R. From (b), we also
have x
9
= x and so x + x
7
+ x = 0, which gives x
7
= x, and then x = x
9
=
x
2
x
7
= x
2
x = x
3
.
To complete the proof of (c), let n be an integer satisfying n 11 and n
3, 5, or 7 (mod 8) and such that x + x
9
+ x
n
= 0 for all x R. Recall that x +
x
9
+ x
n8
= x + x + x = 0 for all x R. Iterating as many times as needed,
we will reach either x + x
9
+ x
3
= 0, x + x
9
+ x
5
= 0 or x + x
9
+ x
7
= 0, in
which cases we already know that x
3
= x for all x R.
Also solved by GEORGE APOSTOLOPOULOS, Messolonghi, Greece; HOWARD BELL, Brock U.; ALBERT
WHITCOMB, Castle Shannon PA; and the proposer.
104 THE MATHEMATICAL ASSOCIATION OF AMERICA