See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323296523

Euler's Sine Product Formula: An Elementary Proof

Article in The College Mathematics Journal · March 2018

DOI: 10.1080/07468342.2018.1419703

CITATIONS READS
6 13,486

1 author:

David Salwinski

2 PUBLICATIONS 18 CITATIONS

SEE PROFILE

All content following this page was uploaded by David Salwinski on 21 August 2020.

The user has requested enhancement of the downloaded file.

Mathematical Assoc. of America College Mathematics Journal 49:2 Euler’s Sine Product Formula page 1

Euler’s Sine Product Formula: An Elementary Proof

David Salwinski

David Salwinski is an undergraduate student studying

mathematics and physics at the University of Toronto. In addition to
taking courses Dave works as a teaching assistant and delivers
math seminars to younger students. His academic interests include
analysis, quantum physics and philosophy, while in his spare time
he enjoys biking, reading and watching documentaries. This is
Dave’s first publication.

In 1734 Leonhard Euler conjectured that the sine function could be expressed as an infinite
product,
∞ 
x2
Y 
sin x = x 1− 2 2
k=1
k π

for all real x. This identity became known as Euler’s sine product formula. Euler obtained it
by factoring sin x into a product over its roots as one would a polynomial. A full description
of his intuitive approach is found in [3, pp. 214–216]. Though he never managed to prove it
rigorously, Euler did provide compelling arguments that his sine product formula was indeed
correct (some are discussed in [8]) and explored its numerous consequences. Most notably he
used it to solve the Basel problem, showing that
∞
X 1 π2
2
=
k=1
k 6

and thereby answering a ninety year old open question that had stumped his contemporaries.
His original derivation is found in [5] and it is summarized in [2, p. 47], [3, pp. 216–217], and
[10, p. 19].
Euler’s sine product formula remained a conjecture for a hundred years before it was finally
proved rigorously by Karl Weierstrass using techniques from complex analysis. Many other
mathematically rigorous proofs have been discovered since then; see [1, 4, 7, 13], [9, pp. 19–
20], or [12, pp. 142–144] for six different approaches. Nearly all proofs however rely either on
complex analysis or use nontrivial results from real analysis, making them inaccessible to, say,
first year university students. The aim of this article is to present a truly elementary proof of
Euler’s sine product formula which makes use only of the simplest techniques of calculus and
can be understood by anyone studying mathematics at the university level. Our method will
also give a similar product formula for cosine and is easily adapted to prove analogous product
formulas for the hyperbolic sine and cosine.

The proofs we will present are reminiscent of a classic argument found in most introductory
calculus texts: the derivation of Wallis’s product formula for π by comparing certain definite

VOL. 49, NO. 2, MARCH 2018 THE COLLEGE MATHEMATICS JOURNAL 1

Mathematical Assoc. of America College Mathematics Journal 49:2 Euler’s Sine Product Formula page 2

integrals involving powers of sine or cosine. We will therefore begin by presenting this proof.
Our approach is very similar to the one outlined in [11, p. 395].
Theorem 1 (Wallis’s product).

2·2 4·4 6·6 (2n) · (2n) π

lim · · ··· =
n→∞ 1·3 3·5 5·7 (2n − 1) · (2n + 1) 2
Proof. We begin by establishing a reduction formula. Suppose n ≥ 2 is an integer. Integrating
the following by parts we have
Z π Z π
2 π 2
cos n−1
t cos t dt = cos n−1
t sin t 0
2
− (n − 1) cosn−2 t (− sin2 t) dt
0 0
Z π
2
= 0 + (n − 1) cosn−2 t (1 − cos2 t) dt
0
Z π Z π
2 2
n−2
= (n − 1) cos t dt − (n − 1) cosn t dt,
0 0

which upon rearrangement gives the reduction formula

π π
n−1
Z 2
Z 2
cosn t dt = cosn−2 t dt, n ≥ 2. (1)
0 n 0

Repeated application of this formula leads to two possible conclusions. For even exponents we
have
Z π2 Z π
2n − 1 2
cos2n t dt = cos2(n−1) t dt
0 2n 0
Z π
2n − 1 2n − 3 1 2
= · ··· cos0 t dt,
2n 2n − 2 2 0
and so
π n
π 1 3 2n − 1 π Y 2k − 1
Z 2
cos2n t dt = · · ··· = , n ≥ 0, (2)
0 2 2 4 2n 2 k=1 2k

while for odd exponents we obtain

π n
2 4 2n 2k
Z 2 Y
cos2n+1 t dt = · ··· = , n ≥ 0. (3)
0 3 5 2n + 1 k=1 2k + 1

Note that in the case that n = 0 the above products are defined to be one.

Now if our exponent n is large, the integrals (2) and (3) will be close in size. More precisely,
observe that the inequality
Z π Z π Z π
2 2 2
2n+2 2n+1
0< cos t dt ≤ cos t dt ≤ cos2n t dt, n≥0 (4)
0 0 0

2  THE MATHEMATICAL ASSOCIATION OF AMERICA

Mathematical Assoc. of America College Mathematics Journal 49:2 Euler’s Sine Product Formula page 3

leads to
Z π
2
cos2n+1 t dt
2n + 1
≤ Z0 π ≤ 1, n≥0
2n + 2 2
2n
cos t dt
0

when we make use of (1). Substituting the products in (2) and (3) for the integrals and inter-
twining the fractions when dividing gives the inequality
2n + 1 2 2·2 4·4 (2n) · (2n)
≤ · · ··· ≤ 1, n ≥ 1, (5)
2n + 2 π 1·3 3·5 (2n − 1) · (2n + 1)
from which Wallis’s product follows by taking limits as n → ∞.
Formulas (2) and (3) have another interesting consequence: combined they give us a good
estimate on the size of the integrals we were working with.
Corollary 1.
r π r
π 1 π 1
Z 2
n
√ ≤ cos t dt ≤ √ , n≥1 (6)
2 n+1 0 2 n
π
Proof. Beginning with equation (5) above, we multiply through by 2
and shift the denominators
in the product to obtain
π 2n + 1 2·2 4·4 (2n) · (2n) π
≤ · ··· (2n + 1) ≤ ,
2 2n + 2 3·3 5·5 (2n + 1) · (2n + 1) 2
which implies
 2
π 1 2 4 2n π 1
≤ · ··· ≤ .
2 2n + 2 3 5 2n + 1 2 2n + 1
Taking square roots and substituting the integral from (3) gives us
√ Z π2 √
π 1 2n+1 π 1
√ ≤ cos t dt ≤ q , n ≥ 0. (7)
2 n+1 0 2 n + 12

Now, (7) was ultimately obtained by performing manipulations to (4). If instead we begin
with the similar inequality
Z π2 Z π2 Z π2
2n+1 2n
0< cos t dt ≤ cos t dt ≤ cos2n−1 t dt, n ≥ 1
0 0 0

and proceed in an analogous manner dividing the integrals and making use of (2) and (3), we
will arrive at
√ Z π2 √
π 1 π 1
q ≤ cos2n t dt ≤ √ , n ≥ 1. (8)
2 n+ 21 0 2 n

The details are left to the reader. Combining (7) and (8) completes the proof.

VOL. 49, NO. 2, MARCH 2018 THE COLLEGE MATHEMATICS JOURNAL 3

Mathematical Assoc. of America College Mathematics Journal 49:2 Euler’s Sine Product Formula page 4

We have one more proof to get through before moving on to the sine product formula. The
lemma below is more than an interesting result in itself– it will prove critical in the analysis that
follows.
Lemma. Suppose f is a Riemann integrable function and |f (t) − f (0)| ≤ M t for some pos-
itive constant M on [0, π2 ]. It follows that
Z π
2
f (t) cosn t dt
0
lim Z π = f (0). (9)
n→∞ 2
n
cos t dt
0

Proof. We begin by combining the inequality t ≤ tan t with the identity 1 + tan2 t = sec2 t,
1
both of which are valid on [0, π2 ), to produce the inequality cos t ≤ √ , valid on [0, π2 ].
1 + t2
Now let n ≥ 3 be an integer. We have
Z π Z π Z π
2 2 2
f (t) cosn t dt − f (0) cosn t dt ≤ |f (t) − f (0)| cosn t dt
0 0 0
π
t
Z 2
≤M n dt
0 (1 + t2 ) 2
π
−M 1 2
= n
n − 2 (1 + t2 ) 2 −1 0

M
≤ .
n−2
Dividing through by (6) gives the inequality
Z π
2
f (t) cosn t dt r √
0 2 n+1
π − f (0) ≤ M, n ≥ 3, (10)
π n−2
Z 2
n
cos t dt
0

from which the conclusion follows by taking limits as n → ∞.

It should be noted that our hypotheses on f in the above lemma are too strong: (9) in fact
holds for any improper-Riemann integrable function as long as it is continuous at zero. See
[6, pp. 64–66] for a proof of this stronger result. Nevertheless our hypotheses are satisfied by
a wide range of functions, and our proof gives us the useful estimate (10). For instance, if
f is continuous on [0, π2 ] and differentiable on (0, π2 ) with |f 0 (t)| ≤ M , then f satisfies our
hypotheses since applying the mean value theorem will give |f (t) − f (0)| ≤ M t.

At last we are ready to establish our main results. The proof of Euler’s sine product formula
(11) along with its counterpart the cosine product formula (12) will be of the same flavor as
that of Wallis’s product: we will obtain and repeatedly apply a reduction formula found by
integrating a power of cosine over the interval [0, π2 ], but this time our integral will include the

4  THE MATHEMATICAL ASSOCIATION OF AMERICA

Mathematical Assoc. of America College Mathematics Journal 49:2 Euler’s Sine Product Formula page 5

extra factor cos(2xt). The intuition behind working with this new integral is twofold. First,
cos(2xt) has a simple antiderivative, and antidifferentiating it twice leaves it unchanged apart
from a constant factor. We therefore expect it to cooperate when we integrate by parts twice in
the hopes of deriving a reduction formula. Second, including the variable x guarantees that if an
elegant reduction formula is obtained it will be more general than (1), as that result corresponds
to the special case of x = 0.
Theorem 2 (Trigonometric product formulas). For all real x we have
∞ 
x2
Y 
sin(πx) = πx 1− 2 , (11)
k=1
k
∞ 
4x2
Y 
cos(πx) = 1− . (12)
k=0
(2k + 1)2

Proof. Suppose n ≥ 2 is an integer and x is any real number except 0 or ± n2 . Integrating by

parts twice we have
Z π2 π Z π2
n cosn t sin(2xt) 2 n
cos t cos(2xt) dt = + cosn−1 t sin t sin(2xt) dt
0 2x 0 2x 0
π
n cosn−1 t sin t cos(2xt) 2
=0+ −
2x 2x 0
π
!
1
Z 2
n−2 2 n
 
+ (n − 1) cos t (− sin t) + cos t cos(2xt) dt
2x 0
Z π2
n
(n − 1) cosn−2 t (cos2 t − 1) + cosn t cos(2xt) dt
 
= 2
4x 0
Z π2 Z π
n2 n n(n − 1) 2
= 2 cos t cos(2xt) dt − cosn−2 t cos(2xt) dt,
4x 0 4x2 0

which after rearrangement gives the reduction formula

Z π2 −1 Z π
n−1 4x2

2
cosn t cos(2xt) dt = 1− 2 cosn−2 t cos(2xt) dt, n ≥ 2. (13)
0 n n 0

Notice that this formula also holds in the initially excluded case of x = 0, and it reduces to (1)
as already noted.

As in the proof of Wallis’s product we need to consider the cases of even and odd exponents
separately. For even exponents our reduction formula is
Z π2 −1 Z π
2n − 1 x2

2
2n
cos t cos(2xt) dt = 1− 2 cos2(n−1) t cos(2xt) dt, n ≥ 1,
0 2n n 0

which after temporarily excluding the case of x an integer and repeatedly applying results in
Z π2 n n  −1 Z π
2n
Y 2k − 1 Y x2 2
cos t cos(2xt) dt = 1− 2 cos2·0 t cos(2xt) dt
0 k=1
2k k=1
k 0

VOL. 49, NO. 2, MARCH 2018 THE COLLEGE MATHEMATICS JOURNAL 5

Mathematical Assoc. of America College Mathematics Journal 49:2 Euler’s Sine Product Formula page 6

n n  −1
π Y 2k − 1 Y x2 sin(πx)
= 1− 2 .
2 k=1 2k k=1 k πx

Replacing the first product with the integral in (2) and rearranging we arrive at the suggestive
formula
Z π2
n cos2n t cos(2xt) dt
x2
Y  
0
sin(πx) = πx 1− 2 Z π2 , x ∈ R, n ≥ 1. (14)
k=1
k 2n
cos t dt
0

Notice the temporarily-excluded case of x an integer is recovered after the rearrangement.

At this point all that’s left to do is to take a limit as n → ∞ of the above. The partial product
on the right simply tends to the infinite product, while the limit of the ratio of integrals can be
found using the Lemma. Set f (t) = cos(2xt) and observe that, since this function is differen-
tiable everywhere and has derivative bounded by M = 2|x|, it satisfies the hypotheses of the
Lemma by the remarks made following its proof. Thus, the ratio of integrals tends to f (0) = 1,
and the proof of Euler’s sine product formula is complete.

As you might expect, the product formula for cosine is proved by exhaustively applying the
reduction formula (13) with an odd exponent. Temporarily excluding the case of x a half-integer
we compute
π −1 Z π
4x2

2n
Z
2 2
2n+1
cos t cos(2xt) dt = 1− 2
cos2(n−1)+1 t cos(2xt) dt
0 2n + 1 (2n + 1) 0
n n  −1 Z π
Y 2k Y 4x2 2
= 1− 2
cos2·0+1 t cos(2xt) dt.
k=1
2k + 1 k=1
(2k + 1) 0

π
cos(πx)
Z 2
Now cos t cos(2xt) dt = , as the reader is invited to verify. Substituting this
0 1 − 4x2
along with (3) into the above and rearranging leads us to the following analogue of (14):
Z π2
n cos2n+1 t cos(2xt) dt
4x2
Y  
0
cos(πx) = 1− 2 Z π2 , x ∈ R, n ≥ 1 (15)
k=0
(2k + 1) 2n+1
cos t dt
0

Taking a limit as n tends to infinity and applying the Lemma with f (t) = cos(2xt) as before
proves (12).
Some impressive results are obtained by substituting particular values of x into the sine and
cosine product formulas. For example, setting x = 41 in the cosine product we find that
∞   Y ∞
π

Y 1 (4k + 1)(4k + 3)
cos 4 = 1− 2
= ,
k=0
4(2k + 1) k=0
(4k + 2)(4k + 2)

which may be rewritten as

√ 2 · 2 6 · 6 10 · 10
2= · · ···
1 · 3 5 · 7 9 · 11

6  THE MATHEMATICAL ASSOCIATION OF AMERICA

Mathematical Assoc. of America College Mathematics Journal 49:2 Euler’s Sine Product Formula page 7

1
Similarly, setting x = 4
in the sine product formula proves that

π 4 · 4 8 · 8 12 · 12
√ = · · ···
2 2 3 · 5 7 · 9 11 · 13
Multiplying these together we recover Wallis’s product, which could also be obtained by set-
ting x = 12 in the sine product. The reader is invited to find more infinite products in Exercise 1.

We now return to the final step in the proof of the Lemma and make use of inequality (10) to
derive some bounds on the partial products of (11) and (12).
Corollary 2 (Bounds on the partial products).
q
1
2|x| n + 2 sin(πx)
1− √ ≤ n   ≤ 1, x ∈ R, x ∈
/ Z, n ≥ 2
π n−1 Y x2
πx 1− 2
k=1
k
√
2|x| n + 1 cos(πx)
1− √ 1 ≤ Y n   ≤ 1, / Z + 21 , n ≥ 1
x ∈ R, x ∈
π n− 2 4x2
1−
k=0
(2k + 1)2

Proof. Setting f (t) = cos(2xt) and M = 2|x| in inequality (10) gives us

Z π
2
cosn t cos(2xt) dt √ √
0 2|x| 2 n + 1
Z π2 −1 ≤ √ , n ≥ 3.
n
π n−2
cos t dt
0

Now because cosine is bounded above by one the quantity within the absolute value is always
nonpositive. Dropping the absolute value therefore results in
Z π
2
√ √ cosn t cos(2xt) dt
2|x| 2 n + 1 0
1− √ ≤ Z π2 ≤ 1, n ≥ 3.
π n−2 n
cos t dt
0

The proof is completed by considering the cases of even and odd n and substituting (14) and
(15), respectively.
We see from Corollary 2 that, except in the case of the roots, the nth partial product always
over-approximates sine and cosine in magnitude. The approximations are not very good when
x is comparable in size to n but improve considerably the smaller x is. This is illustrated in
Figure 1. In the case of |x| ≤ 21 better bounds are possible; see Exercise 2.

VOL. 49, NO. 2, MARCH 2018 THE COLLEGE MATHEMATICS JOURNAL 7

Mathematical Assoc. of America College Mathematics Journal 49:2 Euler’s Sine Product Formula page 8

-2 -1 1 2

-1

-2

  
x2 x2
Figure 1. A plot of sin(πx) and πx(1 − x2 ) 1 − 4 1− 9 .

Those familiar with complex analysis will realize that, by analytic continuation, both the sine
and cosine product formulas must in fact hold for all complex numbers, and that substituting ix
in them will prove two analogous product formulas for the hyperbolic sine and cosine. With a
simple modification to the integral we used in our proof of Theorem 2 however, we can arrive
at the hyperbolic products without needing to use complex analysis. The method is outlined in
the theorem below; most of the details are left to the reader.
Theorem 3 (Hyperbolic product formulas). For all real x we have
∞ 
x2
Y 
sinh(πx) = πx 1+ 2 , (16)
k=1
k
∞ 
4x2
Y 
cosh(πx) = 1+ . (17)
k=0
(2k + 1)2

Proof. Suppose n ≥ 2 is an integer and x is any real number except zero. Integrating by parts
twice gives
π π π
n2 n(n − 1)
Z Z Z
2 2 2
cosn t cosh(2xt) dt = − cosn t cosh(2xt) dt + cosn−2 t cosh(2xt) dt,
0 4x2 0 4x2 0

which, following a rearrangement, results in a reduction formula analogous to (13):

π −1 Z π
n−1 4x2
Z 
2 2
n
cos t cos(2xt) dt = 1+ 2 cosn−2 t cosh(2xt) dt, n ≥ 2. (18)
0 n n 0

As before, we will repeatedly apply this reduction formula in the cases of even and odd expo-
nents. In the case of even exponents we have, for n ≥ 1,
π −1 Z π
2n − 1 x2
Z 
2 2
cos2n t cosh(2xt) dt = 1+ 2 cos2(n−1) t cosh(2xt) dt
0 2n n 0

8  THE MATHEMATICAL ASSOCIATION OF AMERICA

Mathematical Assoc. of America College Mathematics Journal 49:2 Euler’s Sine Product Formula page 9

n n  −1 Z π
Y 2k − 1 Y x2 2
= 1+ 2 cos2·0 t cosh(2xt) dt
k=1
2k k=1
k 0

n n  −1
π Y 2k − 1 Y x2 sinh(πx)
= 1+ 2 ,
2 k=1 2k k=1 k πx

which leads to the familiar-looking expression below upon rearranging and making use of (2).
Z π
2
n  cos2n t cosh(2xt) dt
x2
Y 
0
sinh(πx) = πx 1+ 2 Z π2 , x ∈ R, n ≥ 1. (19)
k=1
k 2n
cos t dt
0

Taking a limit as n → ∞ proves the first of our two products. The reader is invited to verify that
cosh(2xt) indeed satisfies the hypotheses of the Lemma and that the ratio of integrals tends to
one.
Following the same procedure with (18) evaluated at an odd exponent leads to the formula
Z π
2
n  cos2n+1 t cosh(2xt) dt
4x2
Y 
0
cosh(πx) = 1+ Z π2 , x ∈ R, n ≥ 1, (20)
k=0
(2k + 1)2 2n+1
cos t dt
0

from which (17) follows. The details are left to the reader.
As before, we may use the estimate (10) to derive upper and lower bounds on the partial
products of (16) and (17).
Corollary 3.
q
1
2 πx n+


sinh(πx) 4 sinh 2 2
1≤ n   ≤1+ √ , x ∈ R, x 6= 0, n ≥ 2,
Y x 2 π π n−1
πx 1+ 2
k=1
k

√
cosh(πx) 4 sinh2 πx n+1
1≤ n  ≤ 1 + √ 2 , x ∈ R, n ≥ 1.
4x2 π π n − 12
Y 
1+
k=0
(2k + 1)2

Proof. Since cosh(2xt) is convex, the inequality

4 sinh2 πx


cosh(πx) − 1 2
1 ≤ cosh(2xt) ≤ 1 + π t = 1 + t
2
−0 π

holds for all t ∈ [0, π2 ], and for all x. We may rewrite it using absolute value as

4 sinh2 πx


| cosh(2xt) − 1| ≤ 2
t, t ∈ [0, π2 ], x ∈ R.
π

VOL. 49, NO. 2, MARCH 2018 THE COLLEGE MATHEMATICS JOURNAL 9

Mathematical Assoc. of America College Mathematics Journal 49:2 Euler’s Sine Product Formula page 10

Making use of estimate (10) now gives

Z π
2
cosn t cosh(2xt) dt
√ √
0
4 sinh2 πx
2 2 n+1
Z π2 −1 ≤ √ , n ≥ 3,
n
π π n−2
cos t dt
0

from which both our inequalities follow by dropping the absolute value, considering the cases
of even and odd n, and calling upon (19) and (20), respectively.

Exercises.
1. Use the product formulas proved in Theorems 2 and 3 to evaluate each of the following.
3 · 3 6 · 6 9 · 9 12 · 12
(a) · · · ···
2 · 4 5 · 7 8 · 10 11 · 13
    
1 1 1 1
(b) 1+ 1+ 1+ 1+ ···
2·4 4·6 6·8 8 · 10
    
1 1 1 1
(c) 1− 2 1− 2 1− 2 1 − 2 ···
2 3 4 5
    
1 1 1 1
(d) 1− 4 1− 4 1− 4 1 − 4 ···
2 3 4 5
Generalize each result as much as you can.
Can you evaluate (c) without needing to use any of the product formulas?

2. The bounds on the partial products we derived in Corollaries 2 and 3 are far from optimal.
Better bounds are possible when x is small in magnitude, specifically when |x| ≤ 21 .

(a) Begin with the inequality cos t ≤ cos(2xt) ≤ 1, valid on the interval [0, π2 ] whenever
|x| ≤ 21 . Multiply through by either cos2n t or cos2n+1 t and integrate over the interval.
Identities (14) and (15) can now be used to replace the middle terms. Conclude that
2n + 1 sin(πx)
≤ n  ≤ 1, |x| ≤ 12 , x 6= 0, n ≥ 1,
2n + 1 x2
Y 
πx 1− 2
k=1
k
2n + 2 cos(πx)
≤ n   ≤ 1, |x| < 21 , n ≥ 0.
2n + 3 Y 4x2
1−
k=0
(2k + 1)2

(b) Using a similar argument, prove that

sinh(πx) 2n + 1
1≤ n  2
 ≤ , |x| ≤ 12 , x 6= 0, n ≥ 1,
Y x 2n
πx 1+ 2
k=1
k

10  THE MATHEMATICAL ASSOCIATION OF AMERICA

 Mathematical Assoc. of America College Mathematics Journal 49:2 Euler’s Sine Product Formula page 11

cosh(πx) 2n + 2
1≤ n  2
 ≤ , |x| < 21 , n ≥ 0.
Y 4x 2n + 1
1+
k=0
(2k + 1)2

Hint: First show that cos t ≤ cos t cosh(2xt) ≤ 1 on [0, π2 ] whenever |x| < 12 .

3. We have already mentioned that the sine and cosine product formulas are equivalent to
their hyperbolic counterparts; each is obtained from the other by substituting imaginary
numbers. It’s a little more surprising that the sine and cosine product formulas are them-
selves equivalent. Prove this as follows:

(a) Using the double angle formula sin(2x) = 2 sin x cos x, derive the cosine product
formula assuming the sine product formula.

(b) Now begin with the cosine product formula and substitute x − 21 for x. Working with
the partial products, factor each of the resulting terms as a difference of squares, reindex
the two products, and with the help of Theorem 1 arrive at the sine product formula.

Acknowledgment. The author wishes to thank the reviewers for their many helpful comments and
suggestions.

Summary. We provide a proof of Euler’s sine product formula using simple techniques of calculus and
illustrate how our method can be used to prove similar product formulas for cosine as well as hyperbolic
sine and cosine. We also derive bounds on the partial products and explore some consequences of these
formulas.

References

1. Ciaurri, O. (2015). Euler’s product expansion for the sine: An elementary proof. Am. Math. Monthly 122: 693–695.
2. Dunham, W. (1999). Euler: The Master of Us All. Washington, DC: Mathematical Association of America.
3. Dunham, W. (1990). Journey Through Genius: The Great Theorems of Mathematics. New York: Wiley.
4. Eberlein, W. F. (1977). On Euler’s infinite product for the sine. J. Math. Anal. Appl. 58: 147–151.
5. Euler, L. (1740). De summis serierum reciprocarum. Comm. acad. sci. Petrop. 7: 123–134.
6. Furdui, O. (2013). Limits, Series, and Fractional Part Integrals. New York: Springer.
7. Holst, L. (2012). A proof of Euler’s infinite product for the sine. Am. Math. Monthly 119: 518–521.
8. McKinzie, M., Tuckey, C. (1997). Hidden lemmas in Euler’s summation of the reciprocals of the squares. Arch. Hist.
Exact Sci. 51: 29–57.
9. Melnikov, Y. A. (2011). Green’s Functions and Infinite Products: Bridging the Divide. Basel: Birkhäuser.
10. Polya, G. (1954). Mathematics and Plausible Reasoning Vol 1: Induction and Analogy in Mathematics. Princeton, NJ:
Princeton Univ. Press.
11. Spivak, M. (2008) Calculus. 4th ed. Houston, TX: Publish or Perish.
12. Stein, E. M., Shakarchi, R. (2003). Complex Analysis. Vol 2. Princeton, NJ: Princeton Univ. Press.
13. Walter, W. (1982). Old and new approaches to Euler’s trigonometric expansions. Am. Math. Monthly 89: 225–230.

VOL. 49, NO. 2, MARCH 2018 THE COLLEGE MATHEMATICS JOURNAL 11

View publication stats