Mathematics Magazine

ISSN: 0025-570X (Print) 1930-0980 (Online) Journal homepage: https://www.tandfonline.com/loi/umma20

Problems and Solutions

To cite this article: (2020) Problems and Solutions, Mathematics Magazine, 93:1, 71-78, DOI:
10.1080/0025570X.2020.1685297

To link to this article: https://doi.org/10.1080/0025570X.2020.1685297

Published online: 22 Jan 2020.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=umma20
 PROBLEMS
LES REID, Editor EUGEN J. IONAŞCU, Proposal Editor
Missouri State University Columbus State University

RICHARD BELSHOFF, Missouri State University; EYVINDUR ARI PALSSON, Virginia Tech;
CODY PATTERSON, Texas State University; ROGELIO VALDEZ, Centro de Investigación en
Ciencias, UAEM, Mexico; Assistant Editors

Proposals
To be considered for publication, solutions should be received by July 1, 2020.

2086. Proposed by David M. Bradley, University of Maine, Orono, ME.

Let f (k) denote the largest integer that is a divisor of nk − n for all integers n. For
example, f (2) = 2 and f (3) = 6. Determine f (k) for all integers k > 1.
2087. Proposed by Florin Stanescu, Şerban Cioiculescu School, Găeşti, Romania.
Consider the sequence defined by x1 = a > 0 and
 
x1 + x2 + · · · + xn−1
xn = ln 1 + for n ≥ 2.
n−1
Compute limn→∞ xn ln n.
2088. Proposed by Mircea Merca, University of Craiova, Romania.
Let n and t be nonnegative integers. Prove that

2n
Ft
(−1)k Ftk F2tn−tk = − F2tn ,
k=0
L t

where Fi denotes the ith Fibonacci number and Li denotes the ith Lucas number.
Math. Mag. 93 (2020) 71–78. doi:10.1080/0025570X.2020.1685297 
c Mathematical Association of America
We invite readers to submit original problems appealing to students and teachers of advanced
undergraduate mathematics. Proposals must always be accompanied by a solution and any relevant
bibliographical information that will assist the editors and referees. A problem submitted as a
Quickie should have an unexpected, succinct solution. Submitted problems should not be under
consideration for publication elsewhere.
Proposals and solutions should be written in a style appropriate for this Magazine.
Authors of proposals and solutions should send their contributions using the Magazine’s
submissions system hosted at http://mathematicsmagazine.submittable.com. More detailed in-
structions are available there. We encourage submissions in PDF format, ideally accompanied by
LATEX source. General inquiries to the editors should be sent to [email protected].

71
72 MATHEMATICS MAGAZINE
2089. Proposed by Rick Mabry, LSU Shreveport, Shreveport, LA.
Let A1 , A2 , . . . , An be the vertices of a convex n-gon in the plane. Identifying the
indices modulo n, define the following points: Let Bi and Ci be vertices on Ai Ai+1
such that
Ai Ai+1
Ai Bi = Ci Ai+1 < ,
2
and let Di be the intersection of Bi−1 Ci and Bi Ci+1 . Prove that

n
Bi Di
= 1.
i=1
D i Ci

2090. Proposed by Gregory Dresden, Washington & Lee University, Lexington, VA.
Recall that a matching of a graph is a set of edges that do not share any vertices. For
example, C4 , the cyclic graph on four vertices (i.e., a square), has seven matchings: the
empty set, single edges (four of these), or pairs of opposite edges (two of these).
Let Gn be the (undirected) graph with vertices xi and yi , 0 ≤ i ≤ n − 1, and edges
{xi , xi+1 }, {xi , yi }, and {yi , xi+1 }, 0 ≤ i ≤ n − 1, where the indices are to be taken
modulo n. For example, G4 is shown below. Determine the number of matchings of
Gn .
VOL. 93, NO. 1, FEBRUARY 2020 73
Quickies
1097. Proposed by George Stoica, Saint John, New Brunswick, Canada.
Let z1 , . . . , zn ∈ C with |zi | = 1. Show that there exists ω ∈ C with |ω| = 1 such that
|(ω − z1 ) . . . (ω − zn )| ≥ 2, and this result is the best possible, namely 2 cannot be
replaced by any larger number.
1098. Proposed by Oniciu Gheroghe, Botoşani, Romania.
In the convex quadrilateral ABCD, ∠BAD ∼ = ∠BCD both with measure 60◦ . The
diagonal AC bisects ∠BAD. Prove that m(∠BDA) = 2m(∠BCA).

Solutions

2061. Proposed by Florin Stanescu, Şerban Cioiculescu School, Găeşti, Romania.

Three complex numbers a, b, c satisfy
|a| = |b| = |c| = 1 and a 3 + b3 + c3 = 2abc.
Prove that a, b, c are vertices of an isosceles triangle on the complex plane.

Solution by José Heber Nieto, Universidad del Zulia, Maracaibo, Venezuela.

We shall prove more generally that, if
|a| = |b| = |c| = 1 and a 3 + b3 + c3 = rabc
for any real number r, −1 ≤ r ≤ 3, then a, b, c are the vertices of an isosceles tri-
angle in the complex plane. Note that we are allowing for the possibility of degenerate
triangles where vertices coincide.
A triangle is isosceles if and only if two of its central angles are congruent. There
are three possibilities:
a b b c c a
= , = , or = .
b c c a a b
Therefore the triangle is isosceles if and only if
(a 2 − bc)(b2 − ca)(c2 − ab) = 0, i.e.,
a 4 bc + ab4 c + abc4 = a 3 b3 + b3 c3 + a 3 c3 .

If a 3 + b3 + c3 = rabc then
1 1 1 r
+ 3 + 3 = a 3 + b3 + c3 = rabc = .
a3 b c abc
Multiplying both sides by (abc)3 we obtain
a 3 b3 + b3 c3 + a 3 c3 = r(abc)2 = a 4 bc + ab4 c + abc4
as desired.
74 MATHEMATICS MAGAZINE
Now suppose

a 3 + b3 + c3 = rabc

and, for example, c/a = a/b. Then

 3  
b c 3 rbc
1+ + = 2 = r.
a a a
Since b/a and c/a are conjugate, we may put
b c
= eφi , and = e−φi .
a a
Then

e3φi + e−3φi = r − 1, i.e.,

r −1
cos 3φ = .
2
This shows that −1 ≤ r ≤ 3. Note that degenerate triangles with

φ = 0, (e.g., a = b = c = 1) and φ = π, (e.g., a = 1, b = c = −1)

can occur when r = −1 or 3.
In the original problem, cos 3φ = 1/2, so
π 5π 7π
φ= , , or ,
9 9 9
and the angles of the triangle formed by a, b, and c are
π π π    
5π 5π 4π 7π 7π 2π
, , , , , , or , , .
18 18 9 18 18 9 18 18 9

Also solved by Robert A. Agnew, Hafez Al-Assad (Syria), Michel Bataille (France), Cal Poly
Pomona Problem Solving Group, Robert Calcaterra, Adam Cofmann, Bruce E. Davis, Robert L.
Doucette, George Washington University Problems Group, Kyle Gatesman, Michael Goldenberg
& Mark Kaplan, Eugene A. Herman, Walther Janous (Austria), Stephen Kaczkowski, Koopa Tak
Lun Koo (Hong Kong), Omran Kouba (Syria), Kee-Wai Lau (Hong Kong), Hyomin Park (Korea),
Theophilus Pedapolu, Michael Reid, Ivan Retamoso, Leonel Robert & Charlotte Ochanine, Randy
K. Schwartz, Daniel Vacaru (Romania), Lawrence R. Weill and the proposer. There was one
incomplete or incorrect solution.

2062. Proposed by Enrique Treviño, Lake Forest College, Lake Forest, IL.
For every positive integer n, let f (n) denote the number of occurrences of the digit
2 in the sequence 1, 2, . . . , n of integers written in base 10. (For instance, f (25) = 9
because the digit 2 appears once in 2, 12, 20, 21, 23, 24, 25 and twice in 22.)
(i) Find a positive integer n such that f (n) = n.
(ii) Are there infinitely many solutions to f (n) = n?
VOL. 93, NO. 1, FEBRUARY 2020 75
Solution by Cassandra DeBacco (student) and Mark Capsambelis, Riverview High
School, Oakmont, PA.
Let n ∈ N. For all integers between 0 and 10n − 1, the digit 2 appears in each of the n
decimal places 1/10 of the time, so
f (10n ) = n10n−1 .
In particular, if n = 10, then

f 1010 = 10 · 109 = 1010 ,
which answers (i).
For (ii), note that for all n > 100,
f (10n ) = n10n−1 > 10n+1 .
So if n > 100 and k ∈ N such that 10n ≤ k < 10n+1 , then
f (k) ≥ f (10n ) > 10n+1 > k
since f is non-decreasing. Therefore, there are no solutions of f (n) = n for n > 100,
so there are only a finite number of solutions.
Also solved by Mohammed Aassila (France), Ulrich Abel (Germany), Hafez Al-Assad (Syria),
Armstrong Problem Solvers, Brian Beasley, Virginia Faulkner Bolton, David Stone and John
Hawkins, Robert Calcaterra, Bill Cowieson, Robert L. Doucette, Kyle Gatesman, George Wash-
ington University Problems Group, Eugene A. Herman, Kelly Jahns, Andrea McCormack, José
Heber Nieto (Venezuela), Northwestern University Math Problem Solving Group, Charlotte Ocha-
nine, Moubinool Omarjee (France), Timothy Prescott, Michael Reid, Arnold Saunders, Randy K.
Schwartz, Allen Schwenk, James Swenson, Mark Wildon, and the proposer. There were 3 incom-
plete or incorrect solutions.

2063. Proposed by Ovidiu Furdui and Alina Sı̂ntămărian, Technical University of Cluj-
Napoca, Cluj-Napoca, Romania.
Evaluate

∞ 
∞
(−1)n+k−1
.
n=0 k=1
(n + k)2

Solution by Northwestern University Math Problem Solving Group, Evanston, IL.

The answer is ln 2.
Let SN be the following partial sum:

N 
∞
(−1)n+k−1
SN = .
n=0 k=1
(n + k)2

For each fixed N we have that SN is absolutely convergent, so we can rearrange

terms and rewrite the sum as SN = SN
+ SN

, where (writing n + k = j ):
 (−1)n+k−1  N
(−1)j −1  (−1)j −1
N
SN
= = j = ,
n+k≤N
(n + k)2 j =1
j2 j =1
j
n≥0, k≥1

∞
(−1)j −1
SN

= (N + 1) .
j =N+1
j2
76 MATHEMATICS MAGAZINE
The sum of the alternating series SN

can be bound by its first term, so we have

1
|SN

| < →0 as N → ∞.
N +1
Hence

∞ 
∞
(−1)n+k−1 
∞
(−1)j −1
= lim SN
+ lim SN

= + 0 = ln 2 .
n=0 k=1
(n + k)2 N→∞ N→∞
j =1
j

Also solved by Ulrich Abel (Germany), Farrukh Rakhimjanovich Ataev (Uzbekistan), Michel
Bataille (France), Necdet Batir (Turkey), Khristo Boyadzhiev, Brian Bradie, Robert Calcaterra,
Hongwei Chen, Bill Cowieson, Bruce Davis, Robert L. Doucette, Saumya Dubey, John N. Fitch,
Kyle Gatesman, Subhankar Gayen (India), George Washington University Problems Group, Tom
Goebeler & Wendy Sun, Russelle Guadalupe (Philippines), GWstat Problem Solving Group, Eu-
gene A. Herman, Walther Janous (Austria), Kee-Wai Lau (China), Pedro Acosta De Leon, Carl
Libis, José Heber Nieto (Venezuela), Moubinool Omarjee (France), Emily Owen, Hyomin Park
(Korea), Sumanth Ravipati, Edward Schmeichel, Randy K. Schwartz, Albert Stadler (Switzer-
land), Robert W. Vallin, Michael Vowe, Mark Wildon (UK), Lienhard Wimmer (Switzerland),
John Zacharias, Yijie Zhu (China), and the proposer. There were 3 incomplete or incorrect solu-
tions.

2064. Proposed by Ioan Băetu, Botoşani, Romania.

Characterize those integers n ≥ 2 such that the ring Zn of integers modulo n has a
subset F that is a field under the operations of addition and multiplication induced
from Zn . [Note that the unity i of such a field F need not be the unity 1 of Zn .]

Solution by Michael Reid, University of Central Florida, Orlando, FL.

The ring Zn contains a subring F that is a field if and only if n is not powerful, i.e.,
there is a prime p such that n is divisible by p but not by p2 .
First suppose that p is a prime number such that p|n but p 2  n. Write n = pm, so
p  m. By the Chinese remainder theorem, Zn ∼ = Zp × Zm , and the subset Zp × {0} is
isomorphic to Zp , which is a field.
Conversely, suppose Zn contains a field F . Then F is finite, so its characteristic is
a prime number p. The additive group (F, +) has exponent p, so it is contained in the
p-torsion subgroup of (Zn , +). Denote this p-torsion subgroup by B. Since (Zn , +)
is a cyclic group, every subgroup, in particular B, is also cyclic. Moreover, |B| is
either p or 1, depending on whether p divides n or not. Hence, p|n, so B has order p,
and F coincides with B. Write n = mp e , where e ≥ 1 and p  m. Then the p-torsion
subgroup of Zn is precisely


B = 0, mp e−1 , 2mp e−1 , . . . , (p − 1)mp e−1 .
If e ≥ 2, then every element of F = B is nilpotent, because
 2 
amp e−1 = a 2 mp e−2 mp e ≡ 0 mod n.

This is a contradiction, so we must have e = 1, and thus p 2  n.

Also solved by Paul Budney, Robert Calcaterra, Bill Cowieson, and the proposer. There was
one incomplete or incorrect solution.

2065. Proposed by Su Pernu Mero, Valenciana GTO, Mexico.

Let Q be a cube centered at the origin of R3 . Choose a unit vector (a, b, c) uniformly
at random on the surface of the unit sphere a 2 + b2 + c2 = 1, and let  be the plane
VOL. 93, NO. 1, FEBRUARY 2020 77
ax + by + cz = 0 through the origin and normal to (a, b, c). What is the probability
that the intersection of  with Q is a hexagon?

Solution by Bill Cowieson, Fullerton College, Fullerton, CA.

The probability is
6 arcsin(1/3)
1− ≈ 0.350959.
π
Call a unit vector (a, b, c) “good” if  ∩ Q is a hexagon and “bad” otherwise. A
vector is good if and only if  intersects all six sides of Q, therefore the regions of
good and bad unit vectors share a boundary that consists of those unit vectors for which
 contains a vertex of Q, i.e., those which are orthogonal to the vector from the origin
to some vertex. Without loss of generality, let Q have vertices
(1, 1, 1), (1, 1, −1), (1, −1, 1), . . . , (−1, −1, −1),

so this boundary consists of those unit vectors (a, b, c) which satisfy either

a + b + c = 0, a + b − c = 0, a − b + c = 0, or − a + b + c = 0.
By symmetry, it suffices to find the probability for (a, b, c) chosen from the positive
octant a, b, c > 0, where the good/bad boundary equations are a = b + c, b = c +
a, and c = a + b. These partition the positive octant of the sphere into 4 spherical
triangles: a central triangle of good vectors

H = {(a, b, c) ∈ R3 : a 2 + b2 + c2 = 1, a < b + c, b < c + a, c < a + b},

which has vertices

√ √ √ √  √ √
2 2 2 2 2 2
, ,0 , , 0, , and 0, , ,
2 2 2 2 2 2

and three congruent triangles of bad vectors. The desired probability is Area(H )/(π/2)
On the unit sphere, the area of a spherical triangle is the sum of the three vertex
angles minus π. Let
√ θ be√ the common angle at each vertex of H , so Area(H ) = 3θ −
π. The angle at ( 2/2, 2/2, 0) is between the planes x = y + z and y = x + z,
which is that between the normal vectors (−1, 1, 1) and (−1, 1, −1), so cos θ = 1/3,
   
1 1
Area(H ) = 3 arccos − π = π/2 − 3 arcsin ,
3 3
and the probability that a random plane through the center of a cube makes a hexagon
is
Area(H ) 6 arcsin(1/3)
=1− .
π/2 π

Also solved by Elton Bojaxhiu (Germany) & Enkel Hysnelaj (Australia), Robert L. Doucette,
John N. Finch, George Washington University Problems Group, J.A. Grzesik, Kidefumi Kat-
suura & Edward Schmeichel, Peter McPolin (Northern Ireland), Charlotte Ochanine, Randy K.
Schwartz, Yawen Zhang (student), and the proposer. There were 4 incomplete or incorrect solu-
tions.)
78 MATHEMATICS MAGAZINE
Answers
Solutions to the Quickies from page 73.

A1097. Define P (z) = (z − z1 ) . . . (z − zn ). For any complex number ω, we have

1 
n

P ω · e2ikπ/n = ωn + (−1)n z1 . . . zn .
n k=1

Choose ω = − (z1 . . . zn )1/n . Then, by the formula above,

 n 
1    1 n
  
  P ω · e2ikπ/n  .
2 = |ωn + (−1)n z1 . . . zn | =  P ω · e2ikπ/n  ≤
n  n
k=1 k=1

Therefore there exists k ∈ {1, . . . , n} for which

  
P ω · e2ikπ/n  ≥ 2.

Taking zj = e2ij π/n , we have

|P (ω)| = |ωn − 1| ≤ |ω|n + 1 = 2
for all ω with |ω| = 1, so 2 cannot be replaced by any larger number.
A1098. Denote the circles that circumscribe BCD and ABD by C1 with center
O and C2 with center O
, respectively. One version of the law of sines states that
a/ sin A = b/ sin B = c/ sin C = 2R, where R is the radius of the circle circumscrib-
ing the triangle. Since ∠BAD ∼= ∠BCD, we conclude that C1 and C2 are congruent.
The smaller arcs between B and D on C1 and C2 both have measure
2m(∠BAD) = 2m(∠BCD) = 120◦ .
Therefore OBO
and ODO
are equilateral, so O lies on C2 (and O
lies on C1 ).
Now O is the midpoint of arc BOD since AC is the angle bisector of ∠BAD. Hence we
have
m(∠BDA) = m(∠BOA) = 2m(∠BCO) = 2m(∠BCA).