Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Concrete Mathematics Exercises From 20 September: Exercise 1.2

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

Concrete Mathematics

Exercises from 20 September


Silvio Capobianco

Exercise 1.2
Find the shortest sequence of moves that transfers a tower of n disks from
the left peg A to the right peg C, using the middle peg B as a help, if direct
moves between A and B are disallowed.

Solution. For n = 1 the shortest sequence is A → B, B → C. For n = 2


it is:
1. A → B.

2. B → C.

3. A → B.

4. C → B. Note that the whole tower is on peg B now.

5. B → A.

6. B → C.

7. A → B.

8. B → C.
For the general case, observe that the strategy that solves the problem for n
disks works as follows:
1. Move the upper tower of n − 1 disks on peg C.

2. Move the n-th disk to peg B.

1
3. Move the upper tower of n − 1 disks on peg A.

4. Move the n-th disk to peg C.

5. Move the upper tower of n − 1 disks on peg C.

Let Xn be the number of moves which are necessary to solve the problem:
then X0 = 0 and Xn ≤ 3Xn−1 + 2 for every n ≥ 0, and as the nth disk can
only be moved when all the others have been moved, the converse inequality
must also hold. It is then easy to see, either by induction or by setting
Un = Xn + 1, that the unique solution is Xn = 3n − 1.

Exercise 1.3
Show that, in the previous exercise, each legal arrangement of n disks is en-
countered exactly once.
Solution. There is exactly one legal arrangement per subdivision of the n
disks in three (possibly empty) sets. There are 3n − 1 moves between dis-
placements, so there are 3n displacements reached overall. If one of these
was touched twice, then it would be possible to reduce the number of moves
by performing, the first time we reach said displacement, the chain of steps
we would have taken on the second of its occurrences: which contradicts the
result we obtained in the previous exercise.

Exercise 1.6
Some of the regions defined by n lines in the plane are infinite, while oth-
ers are bounded. What is the maximum possible number of bounded regions?

Solution. We know that the n-th line, crossing the previous n − 1, creates
at most n new regions: of those regions, two must be infinite, while the others
may be all bounded if no two lines are parallel. Then the maximum number
Bn of bounded regions of the planes determined by n satisfies the initial
conditions B1 = B2 = 0 and the recurrence equation Bn = Bn−1 + n − 2. By
setting Cn = Bn+2 we have C0 = 0 and Cn = Cn−1 + n for every n ≥ 1: then
Cn = Sn for every n ≥ 0, and Bn = Sn−2 = Sn − n − (n − 1) = Ln − 2n.

2
Exercise 1.9
Consider the following statement:
 n
x1 + . . . + x n
P (n) : x1 · · · xn ≤ ∀x1 , . . . , xn > 0 . (1)
n

This is trivially true for n = 1, and is also true for n = 2 as (x1 +x2 )2 −4x1 x2 =
(x1 − x2 )2 ≥ 0.

1. By setting xn = (x1 + . . . + xn−1 )/(n − 1), prove that P (n) implies


P (n − 1) for every n > 1.

2. Prove that, for every n ≥ 1, P (n) and P (2) together imply P (2n).

3. Explain why points 1 and 2 together imply that P (n) is true for every
n ≥ 1.

Solution. Observe that (1) expresses the following, well known fact: the
geometric mean of a finite sequence of positive numbers never exceeds their
arithmetic mean.
Point 1. Suppose P (n) is true. Then it remains true with the special
choice of xn :
x1 + . . . + xn−1
x1 · · · xn−1 ·
n−1 !n
x1 + . . . + xn−1 + x1 +...+x
n−1
n−1


n
) n
(n−1)(x +...+x )+(x +...+x
!
1 n−1 1 n−1
n−1
=
n
 n
x1 + . . . + xn−1
=
n−1
 n−1
x1 + . . . + xn−1 x1 + . . . + xn−1
= · .
n−1 n−1

As x1 , . . . , xn−1 are arbitrary and (x1 + . . . + xn−1 )/(n − 1) > 0, P (n − 1) is


true.

3
Point 2. Suppose P (n) and P (2) are both true. Then:

x1 · · · xn · xn+1 · · · x2n
 n  n
x1 + . . . + xn xn+1 + . . . + x2n
≤ ·
n n
   n
x1 + . . . + xn xn+1 + . . . + x2n
= ·
n n
 !2 n
x1 +...+xn
n
+ xn+1 +...+x
n
2n

≤  
2
 2n
x1 + . . . + xn + xn+1 + . . . + x2n
= .
2n

As x1 , . . . , x2n are arbitrary, P (2n) is true.


Point 3. For every positive integer n there exists an integer k ≥ 0 and
positive integers m0 = 2, m1 , . . . , mk = n such that, for every i < k, either
mi+1 = 2mi or mi+1 = mi − 1. (For instance, set mi+1 = 2mi until mi ≥ n,
then mi+1 = mi − 1 until mi = n.) By points 1 and 2, P (m0 ) is true, and
P (m0 ) and P (mi ) together imply P (mi+1 ) for every i < k: therefore, P (mk )
is true. As mk = n is arbitrary, the thesis follows.

You might also like