Team Play Solutions which may be rewritten as

1 1 1
+ ··· + + .
Part i: The equation for the case n = 3 is ak − n ak − 2 ak − 1
1 1 1 Since ak is a solution of (∗) this expression equals 0, hence (n + 1) − ak is
1−x + 2−x + 3−x = 0. The graph will have
asymptotes at x = 1, x = 2, x = 3, and also a solution. But k < ak < k + 1, so n − k < (n + 1) − ak < n − k + 1.
the x-axis, as shown at right. (The y-axis Hence this solution must be an−k . In other words, (n + 1) − ak = an−k ,
is not drawn.) Apparently the solutions or ak + an−k = n + 1, as claimed. Adding together these equalities for
a1 and a2 are located within the intervals all values of k from 1 to n − 1, we discover that
1 < a1 < 2 and 2 < a2 < 3. 2(a1 + a2 + · · · + an−1 ) = (n + 1)(n − 1),
To solve we multiply through by (1 − x)(2 − x)(3 − x) to obtain
meaning that the sum of the solutions is 12 (n2 − 1).
(6 − 5x + x2 ) + (3 − 4x + x2 ) + (2 − 3x + x2 ) = 0.
Part iv: Let ak be any solution other than the largest. We wish to
√
This simplifies to 3x2 − 12x + 11 = 0, which has solutions a1 = 2 − 3 show that ak+1 − ak > 1. Since the graph of the left-hand side of (∗)
√ 3
and a2 = 2 + 3 is always increasing, it suffices to show that plugging in ak + 1 gives a
3 , using the quadratic formula.
negative value, meaning that we have not yet reached the zero in the
Part ii: Subtracting 1 from each term on the left-hand side, and thus
next interval. We find that
subtracting n from the right-hand side, gives
1 1 1
+ + ··· +
1 2 n 1 − (ak + 1) 2 − (ak + 1) n − (ak + 1)
−1+ − 1 + ··· + − 1 = n − n,
1−x 2−x n−x 1 1 1
= − + + ··· +
which simplifies to ak 1 − ak (n − 1) − ak
1 1
x x x = − − < 0,
+ + ··· + = 0. ak n − ak
1−x 2−x n−x
where we used the fact that ak is a solution to (∗) in the last step.
Factoring out an x reveals that the given equation has the same solutions
as equation (∗), along with the solution x = 0 as well. Part v: For the sake of conserving space we outline the computation.
To begin, multiply through by (1 − x)(2 − x) · · · (n − x) in (∗) and negate
Part iii: We first observe that in general the graph of the left-hand
if necessary to see that a1 , a2 , . . . , an−1 are the n − 1 roots of
side of equation (∗) will resemble the one shown above, except that
there will be n vertical asymptotes at x = 1, x = 2, . . . , x = n. This [(x−2) · · · (x−n)] +[(x−1)(x−3) · · · (x−n)] + · · · + [(x−1) · · · (x−(n−1))].
explains why there are precisely n − 1 solutions, and further shows that Let C be the coefficient of the linear term and let D be the constant term.
1 < a1 < 2 < a2 < 3 < · · · < n − 1 < an−1 < n. We next show that if ak Using Vieta’s formulas, one may confirm that the sum of the reciprocals
is a solution, then (n + 1) − ak is also a solution. Indeed, substituting C
of the roots is equal to − D . The constant term can be written neatly
x = (n + 1) − ak into the left-hand side of equation (∗) yields in the form
 
1 1 1 1 1 1
+ + ··· + , D = n!(−1)n−1 + + ··· + = n!(−1)n−1 A.
1 − (n + 1 − ak ) 2 − (n + 1 − ak ) n − (n + 1 − ak ) 1 2 n
The linear term is more complicated, February 2012
   
n! n−2 1 1 n! n−2 1 1
C = (−1) + ··· + + · · · + (−1) + ··· + .
1 2 n n 1 n−1
1
Every fraction of the form jk with 1 ≤ j < k ≤ n appears twice in the
above sum, which means that we can rewrite C as
" 2  #
n−2 1 1 1 1
C = n!(−1) + ··· + − + ··· + 2 ,
1 n 12 n

C B
or C = n!(−1)n−2 (A2 − B). Finally, simplifying − D gives A − A.

Part vi: We have already seen that 1 < a1 . To prove that a1 < 1 + A1
it suffices to show that substituting x = 1 + A1 into the left-hand side
of (∗) gives a positive value, since that function is increasing for all x.
1
Round Two Solutions
Since 1 + A < 2, this is equivalent to showing that

1 1 1 1
1 < 1 + 1 + ··· + 1 .
(1 + A) −1 2 − (1 + A) 3 − (1 + A) n − (1 + A)

The left-hand side reduces to just A, which is 1 + 21 + · · · + n1 . Therefore

the above inequality is equivalent to
1 1 1 1 1
1+ + ··· + < 1 + 1 + ··· + 1 .
2 n 1− A 2− A (n − 1) − A

1
Subtracting every term but n over to the right-hand side and combining
corresponding pairs of fractions yields
1 1 1 1
< + + ··· + .
n A − 1 2(2A − 1) (n − 1)((n − 1)A − 1)
1 1
However, it is clear that A − 1 < n, which means that n < A−1 . Adding
more positive terms to the right-hand side only makes the inequality
more pronounced. Hence the latter inequality is true, and you’re done.
The interested reader may wish to prove that a21 + · · · + a2n−1 is
1 2
given by 12 (n − 1)(4n + 1). For a real challenge, try showing that
a2 − a1 > a3 − a2 when n ≥ 5. Greater Testing Concepts The Mandelbrot Team Play
PO Box 760 www.mandelbrot.org
c Greater Testing Concepts 2012 Potsdam, NY 13676 [email protected]