Geometry and Trigonometry 657
So we have to prove that
∏n
k= 0
1 +bk
1 −bk
≥nn+^1.
The inequality from the statement implies
1 +bk≥
∑
l =k
( 1 −bl), k= 0 , 1 ,...,n.
Also, the conditionak∈( 0 ,π 2 )implies− 1 <bk<1,k= 0 , 1 ,...,n, so the numbers
1 −bkare all positive. To obtain their product, it is natural to apply the AM–GM inequality
to the right-hand side of the above inequality, and obtain
1 +bk≥nn
√∏
l =k
( 1 −bl), k= 0 , 1 ,...,n.
Multiplying all these inequalities yields
∏n
k= 0
( 1 +bk)≥nn+^1 n
√√
√√∏n
l= 0
( 1 −bl)n.
Hence
∏n
k= 0
1 +bk
1 −bk
≥nn+^1 ,
as desired.
(USA Mathematical Olympiad, 1998, proposed by T. Andreescu)
666.If we multiply the denominator and the numerator of the left-hand side by cost, and
of the right-hand side by cosnt, we obtain the obvious equality
(
eit
e−it
)n
=
eint
e−int
.
667.Using the de Moivre formula, we obtain
( 1 +i)n=
[√
2
(
cos
π
4
+isin
π
4
)]n
= 2 n/^2
(
cos
nπ
4
+isin
nπ
4
)
.
Expanding( 1 +i)nand equating the real parts on both sides, we deduce the identity from
the statement.