Scribd
Upload
50 views

Solutii Seniori

On the other hand, since these two diagonals have a common interior point they cannot belong to different sets. The claim is proved.

Uploaded by

pppttt1234
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
50 views

Solutii Seniori

On the other hand, since these two diagonals have a common interior point they cannot belong to different sets. The claim is proved.

Uploaded by

pppttt1234
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

First Selection Test Solutions

Problem
1. Given
an integer n 2, let an , bn , cn be integer numbers such that ( 3 2 1)n =

an + bn 3 2 + cn 3 4. Show that cn 1 (mod 3) if and only if n 2 (mod 3).

Solution 1. The binomial expansion of ( 3 2 1)n yields


 
 
X
X
n
n
nk
(k2)/3 n
(mod 3).
(1)
(1)
2
cn =
k
k
k2 (mod 3)

k2 (mod 3)

Since
X
k2 (mod 3)

 
 1 n
n
1

= (1 + 1)n + (1 + )n + 2 (1 + 2 )n =
2 + 2 cos(n + 2)
,
k
3
3
3

where 1 +  + 2 = 0, the condition n 2 (mod 3) may be restated as




3cn = (1)n 2n + 2 cos(n + 2)
3 (mod 9).
3
Consideration of n modulo 6 yields 3cn 3 (mod 9) if n 2 or 5 (mod 6), and 3cn 0 (mod 9)
otherwise. The conclusion follows.

Solution 2. Consider the polynomial f = (X1)n cn X 2 bn Xan Z[X]. Clearly, f ( 3 2) = 0.


Since X 3 2 is irreducible in Z[X], it follows that X 3 2 divides f in Z[X], so gn = an +bn X+cn X 2
is the remainder of the division of (X 1)n by X 3 2 in Z[X]. Write n = 3q + r, where q is a nonnegative integer and r {0, 1, 2}, to get (X 1)n = (X 3 1)q (X 1)r = (X 3 2) g + (X 1)r
in Z3 [X], and deduce thereby that gn = (X 1)r in Z3 [X]. Consequently, cn 0 (mod 3) if
r {0, 1}, and cn 1 (mod 3) if r = 2. The conclusion follows.
Problem 2. Circles and are tangent at a point P ( lies inside ). A chord AB of is
tangent to at C; the line P C meets again at Q. Chords QR and QS of are tangent to
. Let I, X, and Y be the incentres of the triangles AP B, ARB, and ASB, respectively. Prove
that P XI + P Y I = 90 .
Solution. Notice that a homothety centred at P mapping to maps C to Q, and maps the
line AB to the tangent to at Q. Thus this tangent is parallel to AB, and hence Q is the
midpoint of arc AB (not containing P ). So the points I, X, and Y lie on the segments P Q, RQ,
and SQ, respectively.
Recall that for any triangle KLM with the circumcircle and incentre J, the points K, L,
and J are equidistant from the midpoint of arc KL of not containing M . Applying this to
triangles AP B, ARB, and ASB we obtain that QA = QB = QX = QY = QI.
Since Q is the midpoint of arc AB, we get that QP A = QP B = QAB. Thus the
triangles QAC and QP A are similar, and QC QP = QA2 = QX 2 . Since QX is tangent to , it
follows that X is their point of tangency; analogously, Y is the point of tangency of QS with .
Finally, from isosceles triangles QXI and QY I we get QXI = QIX = 90 IQX/2 and
QY I = QIY = 90 IQY /2. Denoting by O the centre of , we obtain QIX + QIY =
180 XQY /2 = 180 (180 XOY )/2 = 90 + XP Y . Thus,
P XI + P Y I = XIY XP Y = (90 + XP Y ) XP Y = 90
as required.

P
R

II

X
X
X
X

YY

Remark. The relation QC QP = QA2 also follows from the inversion of pole Q interchanging
the line AB and the circle ,
Problem 3. Determine all injective functions f of the set of positive integers
P into itself satisfying
the following
condition:
If
S
is
a
finite
set
of
positive
integers
such
that
sS 1/s is an integer,
P
then sS 1/f (s) is also an integer.
Solution. We shall prove that the identity is the unique function satisfying the conditions in
the statement. Clearly, f (1) = 1, so f (n) 2 if n 2, by injectivity. We will use the following
well-known result.
Egyptian fractions theorem. For every positive rational q and positive integer N , there exists
a set {n1 , . . . , nk } of positive integers such that ni > N for every i = 1, 2, . . . , k, and
k
X
1
q=
.
ni
i=1

P Now, consider an integer n 2 and use the Egyptian fractions theorem to write 1 1/n =
sS 1/s, where S is a set of integers greater than n(n + 1), and get thereby
1=

X1
1 X1
1
1
+
=
+
+
.
n
s
n + 1 n(n + 1)
s
sS

sS

Consequently,
X 1
1
+
f (n)
f (s)
sS

and

X 1
1
1
+
+
f (n + 1) f (n(n + 1))
f (s)
sS

both are positive integers, so


1
1
1
+

f (n + 1) f (n(n + 1)) f (n)


is an integer. Since
1
1
1
1
1
1
1
1 1

<
+

<
+
+ = 1,
2
f (n)
f (n + 1) f (n(n + 1)) f (n)
f (n + 1) f (n(n + 1))
2 2
2

it follows that

1
1
1
=
+
.
f (n)
f (n + 1) f (n(n + 1))

In particular, f is strictly increasing, so f (n) n.


Finally, proceed by induction on n 2 to prove that f (n) = n. To show that f (2) = 2,
simply notice that 2/f (2) = 1/f (2) + 1/f (3) + 1/f (6) is a positive integer not exceeding 1. To
complete the proof, let f (n) = n for some n 2 and write
1
1
1
1
1
1
1
=
=
+

+
=
n
f (n)
f (n + 1) f (n(n + 1))
n + 1 n(n + 1)
n
to conclude that f (n + 1) = n + 1.
Remark. We do not need the full version of the Egyptian fractions theorem. In fact, all we need
in the solution above is the lemma below.
P
Lemma. For every integer n 2, there exists a set Sn with sSn 1/s = 1 such that n Sn ,
but n + 1, n(n + 1)
/ Sn .
Here we present a direct proof of this Lemma.
For each n {2, 3, 4, 5} one of the the sets {2, 3, 6}, {2, 4, 6, 12}, and {2, 5, 7, 12, 20, 42} fits.
Now assume that n 6 and perform the following steps, starting with the set S = {2, 3, 6}.
Step 1. Let k = max S; if k(k + 1) n then replace k with {k + 1, k(k + 1)} and repeat this
step. At the end, we arrive to a set S with k = max S such that k n < k(k + 1). If k = n then
we are done; otherwise we proceed to Step 2.
Step 2. Replace k by {n} {k(k + 1), (k + 1)(k + 2), . . . , n(n 1)} obtaining the set S 0 . Notice
0
that n + 1 k(k + 1),
 n(n + 1) > max S ; thus, if n + 1 <
 k(k + 1) then we are done. Otherwise,
replace k(k + 1) by k(k + 1) + 1, k(k + 1) k(k + 1) + 1 obtaining the desired set.
Problem 4. Let n be an integer greater than 1. The set S of all diagonals of a (4n 1)-gon
is partitioned into k sets, S1 , . . ., Sk , so that, for every pair of distinct indices i and j, some
diagonal in Si crosses some diagonal in Sj ; that is, the two diagonals share an interior point.
Determine the largest possible value of k in terms of n.
Solution. The required maximum is k = (n 1)(4n 1). Notice that |S| = 2(n 1)(4n 1).
Assume first that k > (n 1)(4n 1). Then there exists a set Si with |Si | = 1. Let Si = {d},
and assume that there are v vertices on one side of d; then the number of vertices on the other
side is 4n 3 v, and the total number of diagonals having a common interior point with d is
v(4n 3 v) (2n 2)(2n 1). Since each Sj with j 6= i contains such a diagonal, we obtain
k (2n 2)(2n 1) + 1 = (n 1)(4n 1) (n 2) (n 1)(4n 1) a contradiction.
Now it remains to construct a partition with k = (n 1)(4n 1). Let us enumerate the
vertices A1 , . . . , A4n1 consecutively; we assume that the enumeration is cyclic, thus Ai+(4n1) =
Ai . Now, for every t = 2, 3, . . . , n and every i = 1, 2, . . . , 4n 1, let us define the set St,i =
{Ai Ai+t , Ai+t1 Ai+2n }.
It is easy to see that the (n 1)(4n 1) sets St,i form a partition of S; we claim that this
partition satisfies the problem condition. Consider two sets St,i and St0 ,i0 ; by the cyclic symmetry
we may assume that i = 0. One can easily observe that a diagonal d has no common interior
points with the diagonals from St,0 if and only if its endpoints are both contained in one of the
sets
{A0 , A1 , . . . , At1 }, {At , At+1 , . . . , A2n }, {A2n , A2n+1 , . . . , A4n1 }

(recall that A4n1 = A0 ); in such a case we will say that d belongs to the corresponding set.
Now, the diagonals from St0 ,i0 cannot belong to one set since this set encompasses at most 2n
consecutive vertices. On the other hand, since these two diagonals have a common interior point
they cannot belong to different sets. The claim is proved.
Remark. The solution for a (4n 3)-gon is almost the same; one only needs to take some care
of the diagonals of the form Ai Ai+n .

You might also like