540 Real Analysis
In=
⎧
⎪⎪⎨
⎪⎪⎩
1 · 3 · 5 ···( 2 k− 1 )
2 · 4 · 6 ···( 2 k)
·
π
2
, ifn= 2 k,
2 · 4 · 6 ···( 2 k)
1 · 3 · 5 ···( 2 k+ 1 )
, ifn= 2 k+ 1.
To prove the Wallis formula, we use the obvious inequality sin^2 n+^1 x<sin^2 nx<
sin^2 n−^1 x,x∈( 0 ,π 2 )to deduce thatI 2 n+ 1 <I 2 n<I 2 n− 1 ,n≥1. This translates into
2 · 4 · 6 ···( 2 n)
1 · 3 · 5 ···( 2 n+ 1 )
<
1 · 3 · 5 ···( 2 n− 1 )
2 · 4 · 6 ···( 2 n)
·
π
2
<
2 · 4 · 6 ···( 2 n− 2 )
1 · 3 · 5 ···( 2 n− 1 )
,
which is equivalent to
[
2 · 4 · 6 ···( 2 n)
1 · 3 · 5 ···( 2 n− 1 )
] 2
·
2
2 n+ 1
<π <
[
2 · 4 · 6 ···( 2 n)
1 · 3 · 5 ···( 2 n− 1 )
] 2
·
2
2 n
.
We obtain the double inequality
π<
[
2 · 4 · 6 ···( 2 n)
1 · 3 · 5 ···( 2 n− 1 )
] 2
·
1
n
<π·
2 n+ 1
2 n
.
Passing to the limit and using the squeezing principle, we obtain the Wallis formula.
467.Denote the integral from the statement byIn,n≥0. We have
In=
∫ 0
−π
sinnx
( 1 + 2 x)sinx
dx+
∫π
0
sinnx
( 1 + 2 x)sinx
dx.
In the first integral changexto−xto further obtain
In=
∫π
0
sinnx
( 1 + 2 −x)sinx
dx+
∫π
0
sinnx
( 1 + 2 x)sinx
dx
=
∫π
0
2 xsinnx
( 1 + 2 x)sinx
dx+
∫π
0
sinnx
( 1 + 2 x)sinx
dx
=
∫π
0
( 1 + 2 x)sinnx
( 1 + 2 x)sinx
dx=
∫π
0
sinnx
sinx
dx.
And these integrals can be computed recursively. Indeed, forn≥0 we have
In+ 2 −In=
∫π
0
sin(n+ 2 )x−sinnx
sinx
dx= 2
∫π
0
cos(n− 1 )xdx= 0 ,
a very simple recurrence. Hence forneven,In=I 0 =0, and fornodd,In=I 1 =π.
(3rd International Mathematics Competition for University Students, 1996)