MAT389 Fall 2016, Problem Set 6

Trigonometric and hyperbolic functions

6.1 Show that eiz = cos z + i sin z for every complex number z.
Hint: start from the right-hand side and work your way towards the left-hand side.

Simply using the definition of the sine and cosine functions in terms of the exponential yields
eiz + e−iz eiz − e−iz
cos z + i sin z = +i = eiz .
2 2i

6.2 Show that sin z = sin x cosh y + i cos x sinh y. Deduce from it the formula

| sin z |2 = sin2 x + sinh2 y

Note: I outlined the proof of the first formula in class. Here I am asking you to finish the
calculation.

First of all, notice that

ez + e−z ez − e−z
cosh z + sinh z = + = ez ,
2 2
much as in the last problem. Then,

eiz − e−iz ei(x+iy) − e−i(x+iy) eix e−y − e−ix ey

sin z = = =
2i 2i 2i
(cos x + i sin x)(cosh y − sinh y) − (cos x − i sin x)(cosh y + sinh y)
=
2i
= sin x cosh y + i cos x sinh y

Now that we have separated sin z in its real and imaginary parts, its modulus is easy to compute:

| sin z |2 = sin2 x cosh2 y + cos2 x sinh2 y

= sin2 x(1 + sinh2 y) + (1 − sin2 x) sinh2 y
= sin2 x + sinh2 y

where we have used the fundamental trigonometric and hyperbolic identities

sin2 z + cos2 z = 1, cosh2 z − sinh2 z = 1

I mentioned in class that there is an alternative derivation of the formula sin z = sin x cosh y +
i cos x sinh y using the identity

sin(z1 ± z2 ) = sin z1 cos z2 ± cos z1 sin z2

and the relationship between trigonometric and hyperbolic functions,

sinh z = −i sin(iz), cosh z = cos(iz),

Here it is:

sin z = sin(x + iy) = sin z cos(iy) + cos x sin(iy) = sin x cosh y + i cos x sinh y.

6.3 Deduce from the formula | sin z |2 = sin2 x + sinh2 y that | sin z | ≥ | sin x |.

Since sinh2 y ≥ 0,
| sin z |2 = sin2 x + sinh2 y ≥ sin2 x.
Taking the (positive!) square root,

| sin z | ≥ | sin x | ≥ sin x.

6.4 Use Problem 5.11 to check that sin z = sin z̄ and cos z = cos z̄.
Note: as in Problem 5.11, you can interpret this geometrically.

eiz − e−iz eiz − e−iz e−iz̄ − eiz̄ eiz̄ − e−iz̄

 
sin z = = = = = sin z̄
2i −2i −2i 2i
eiz + e−iz eiz + e−iz e−iz̄ + eiz̄ eiz̄ + e−iz̄
 
cos z = = = = = cos z̄
2 2 2 2

Here is an illustration of the geometric significance of the statement sin z = sin z̄: points in the
red (resp., green) regions in the z-plane get mapped to the red (resp., green) regions in the w-plane.
The picture can also be interpreted as an illustration of the identity sin(−z) = − sin z.

y v

sin z
z •
•

−π/2 π/2 w = sin z −1 1

• • x • • u

• •
−z z̄ • •
sin(−z) sin z̄
 6.5 Use the identities

sin(z1 ± z2 ) = sin z1 cos z2 ± cos z1 sin z2 ,

cos(z1 ± z2 ) = cos z1 cos z2 ∓ sin z1 sin z2 ,

and the relationship between trigonometric and hyperbolic functions,

sinh z = −i sin(iz), cosh z = cos(iz)

to deduce expressions for sinh(z1 ± z2 ) and cosh(z1 ± z2 ).

For the hyperbolic sine, we have

   
sinh(z1 ± z2 ) = −i sin i(z1 ± z2 ) = −i sin(iz1 ) cos(iz2 ) ± cos(iz1 ) sin(iz2 )
 
= −i i sinh z1 cosh z2 ± i cosh z1 sinh z2
= sinh z1 cosh z2 ± cosh z1 sinh z2 .

For the hyperbolic cosine,

 
cosh(z1 ± z2 ) = cos i(z1 ± z2 ) = cos(iz1 ) cos(iz2 ) ∓ sin(iz1 ) sin(iz2 )
= cosh z1 cosh z2 ∓ (i sinh z1 )(i sinh z2 )
= cosh z1 cosh z2 ± sinh z1 sinh z2 .

6.6 The functions sin z and cos z are 2π-periodic:

sin(z + 2π) = sin z, cos(z + 2π) = cos z

Moreover, shifting their argument by π changes their sign:

sin(z + π) = − sin z, cos(z + π) = − cos z

What do these identities imply for sinh z and cosh z?

The functions sinh z and cosh z are 2πi-periodic:

sinh(z + 2πi) = −i sin[i(z + 2πi)] = −i sin(iz − 2π) = −i sin(iz) = sinh z

cosh(z + 2πi) = cos[i(z + 2πi)] = cos(iz − 2π) = cos(iz) = cosh z

Shifting the arguments of these function by πi results in a sign change:

sinh(z + πi) = −i sin[i(z + πi)] = −i sin(iz − π) = i sin(iz) = − sinh z

cosh(z + πi) = cos[i(z + πi)] = cos(iz − π) = − cos(iz) = − cosh z

These periodicity (and anti-periodicity) conditions can be easily read off in the following pictures.
 y

−3π/2 −π/2 π/2 3π/2 w = sin z −1 1

• • • • x • • u

−2π −π w = cos z −1 1
• • •0 •π 2π
•x • • u

3πi/2 v
•

πi/2 •i
•
w = sinh z
x u
−πi/2
•
• −i

−3πi/2
•
 y
•2πi
v

•πi

w = cosh z −1 1
•0 x • • u

−πi
•

−2πi
•

6.7 Let f (z) = tanh z = sinh z/ cosh z. Find the domain of holomorphicity of f (z), as well as all of
its zeroes.

Since f (z) is a quotient of entire functions, it is holomorphic away from the zeroes of the
denominator, which occur at z = (2k + 1)πi/2 for k ∈ Z.

The zeroes of f (z) are those of the numerator. We saw in class that the solutions to the equation
sinh z = 0 are those complex numbers of the form z = πki for k ∈ Z.

• 3πi/2

πi •

• πi/2
0
• x

• −πi/2

−πi •

• −3πi/2
6.8 Find all roots of the equations
(i) cosh z = 1/2, (ii) sinh z = i, (iii) cosh z = −2, (iv) cosh z = i.

Hint: notice that sinh z and cosh z are linear combinations of ez and e−z . Letting w = ez (so
e−z = w−1 ) turns the equations above into quadratic equations for w.

(i) After the change of variables w = ez , the equation

ez + e−z 1
cosh z = = ⇐⇒ ez + e−z = 1
2 2
√
becomes w2 − w + 1 = 0, whose solutions are w = (1 ± i 3)/2 = e±iπ/3 . For the plus sign, we
obtain π 
z=i + 2πk , k ∈ Z.
3
For the minus sign, it is  π 
z = i − + 2πk , k ∈ Z.
3
y

• 7πi/3
5πi •

• πi/3
x
πi •

• −5πi/3
−7πi •

(ii) Performing the substitution w = ez as in (i), we obtain the equation w2 − 2iw − 1 = 0, with
has a unique solution: w = i. Thus,
π 
z=i + 2kπ .
2
 y
• 9πi/2

• 5πi/2

• πi/2
x

• −3πi/2

• −7πi/2

(iii) We take w = ez yet again to obtain w2 + 4w + 1 = 0. The solutions to the latter are
√ √ √
w = −2 ± 3 = (2 ∓ 3)eiπ (since 2 > 3) and hence
√
z = Log(2 ∓ 3) + (2k + 1)πi, k ∈ Z.

√ √
log(2 − 3) + 3πi • • log(2 + 3) + 3πi

√ √
log(2 − 3) + πi • • log(2 + 3) + πi

x
√ √
log(2 − 3) − πi • • log(2 + 3) − πi

√ √
log(2 − 3) − 3πi • • log(2 + 3) − 3πi

(iv) With w = ez , we get w2 − 2iw + 1 = 0, with solutions

( √
√ (1 + 2)eiπ/2
w = i(1 ± 2) = √
( 2 − 1)e−iπ/2
√
(careful: 2 > 1). Thus,
 √ π 
Log(1 + 2) + i
 + 2kπ , k ∈ Z
2
z= √  π 
Log( 2 − 1) + i − + 2kπ , k ∈ Z

2
 y

√
• log(1 + 2) + 3πi/2

√
log( 2 − 1) + πi •
√
• log(1 + 2) + πi/2
√
log( 2 − 1)
• x
√
• log(1 + 2) − πi/2

√
log( 2 − 1) − πi •
√
• log(1 + 2) − 3πi/2

The transformation w = sin z

The next two problems work out the images of vertical and horizontal lines in the z-plane under
the transformation w = sin z—which I outlined in class. Once you know how those work, you can
find the image of any (possibly infinite) rectangle in the z-plane with sides parallel to the real and
imaginary axes—for example, that in Problem 6.10 below—not only under w = sin z, but also under
other transformations like w = cos z, w = sinh z and w = cosh z (after all, these three are related to
sin z by translations and rotations).

Recall that, under the transformation w = sin z,

− the inverse image of the real axis v = 0 consists of the real axis y = 0 and the lines x = π/2 + kπ,
where k ∈ Z; and
− the inverse image of the imaginary axis u = 0 consists of the lines x = kπ, where k ∈ Z.

6.9 Show that the image of the line given by x = c1 for some fixed 0 < c1 < π/2 under the
transformation w = sin z is a branch of the hyperbola given by the equation
 2  2
u v
− = 1.
sin c1 cos c1
Pay close attention to the orientation: if you move along the line x = c1 from −∞ to +∞ in
y, how do you traverse that hyperbola branch? Show also that, if π/2 < c1 < 0, the image of
x = c1 is the other branch of the same hyperbola.
Hint: remember that sin z = sin x cosh y + i cos x sinh y.

See Figure 1. Let z(t) = c1 + it, t ∈ R, be a parametrization of the given line. Then,

w(t) = sin z(t) = sin(c1 + it) = sin c1 cosh t + i cos c1 sinh t

 y v

−π/2 π/2 w = sin z −1 1

• • x • • u

Figure 1: Problems 6.9 and 6.10

Notice that
u(t) Re w(t) v(t) Im z(t)
= = cosh t, = = sinh t.
sin c1 sin c1 cos c1 cos c1
The fundamental hyperbolic identity cosh2 t − sinh2 t = 1 then implies the equation
 2  2
u(t) v(t)
− = 1.
sin c1 cos c1

If 0 < c1 < π/2, then sin c1 > 0. Since cosh t is always positive too, we have u(t) = sin c1 cosh t > 0.
That is, the image of the line x = c1 is contained in the branch sitting in the half-plane u > 0. On
the other hand, π/2 < c1 < 0 implies sin c1 < 0 and u(t) = sin c1 cosh t < 0—landing then in the
other branch of the same hyperbola.

That the images of these lines are the entirety of the branches above follows from the fact that,
as t moves from −∞ to +∞, so does sinh t; hence v(t) = cos c1 sinh t sweeps the whole real line, from
−∞ to +∞ (since cos c1 > 0 for −π/2 < c1 < π/2).

6.10 Show that the image of the line segment given by

π π
− ≤x≤ , and y = c2
2 2
for some fixed c2 > 0 under the transformation w = sin z is the top half of the ellipse with
equation
 2  2
u v
+ = 1.
cosh c2 sinh c2
What happens if c2 < 0? And if x is outside the interval (−π/2, π/2)?
 See Figure 1. Let z(t) = t + ic2 , t ∈ [−π, π], be a parametrization of the line segment given. Then

w(t) = sin z(t) = sin(t + ic2 ) = sin t cosh c2 + i cos t sinh c2

is a parametrization of its image under the transformation w = sin z. Notice that

u(t) Re w(t) v(t) Im z(t)

= = sin t, = = cos t.
cosh c2 cosh c2 sinh c2 sinh c2

The fundamental trigonometric identity sin2 t + cos2 t = 1 then implies the equation
 2  2
u(t) v(t)
+ = 1.
cosh c2 sinh c2

Thus the image of the segment is contained in the ellipse described by the equation above. To
see that the image is the top half of the ellipse, notice that c2 > 0 implies sinh c2 > 0 and
v(t) = cos t sinh c2 > 0 for t in the interval (−π/2, π/2). On the other hand, if c2 < 0 then sinh c2 < 0
and v(t) = cos t sinh c2 < 0.

As t moves outside of the interval [−π/2, π/2], the 2π-periodicity (and the π-antiperiodicity) of
w = sin z implies that we keep circling around the ellipse.

6.11 Find a conformal transformation w = f (z) that takes the semi-infinite strip 0 < x < π/2, y > 0
onto the upper half-plane H.
Hint: start by considering the image of the domain given under Z = sin z. Do you know of
a conformal transformation w = g(Z) that takes the resulting domain to the entire upper
half-plane?
Note: by a conformal transformation I simply mean a trasformation that is conformal at all
points of the set of interest.

From the last three problems we deduce that the image of the given semi-infinite strip in the
z-plane under Z = sin z is the (open) first quadrant in the Z-plane, and that the transformation
is conformal at every point in it—since it is holomorphic and its derivative vanishes nowhere on it.
Now compose with the transformation w = Z 2 , which takes the first quadrant to the whole upper
half-plane. The latter is also conformal everywhere on the first quadrant, and hence the composed
function w = sin2 z is conformal on the strip.
 y Y

Z = sin z

x X

w = sin2 z w = Z2

The logarithm
6.12 For each of the following complex numbers z find the following:
(1) all logarithms of z (i.e., log z as a multivalued function);
(2) the principal logarithm Log z of z.
√
(i) z = −ei, (ii) z = 1 − i, (iii) z = −1 + i 3.

(i) For the multivalued logarithm, we have

π  π 
log(−ei) = Log e − i + 2πk = 1 − i + 2πk .
2 2
The principal logarithm is the above value for k = 0:
πi
Log(−ei) = 1 −
2
√
(ii) Writing 1 − i = 2 e−iπ/4 , we have
√  π  1  π 
log(1 − i) = Log 2 + i − + 2πk = Log 2 + i − + 2πk .
4 2 4
Once again, the principal value of the logarithm is the one for k = 0:
1 πi
Log(1 − i) = Log 2 −
2 4
 √
(iii) We have −1 + i 3 = 2e2πi/3 . Hence,
√ 
 
 2π
log −1 + i 3 = Log 2 + i + 2πk
3

and √ 
 2πi
Log −1 + i 3 = Log 2 +
3

6.13 Find the image under w = log(π/2) z of the wedge {z ∈ C× | 0 < Arg z < π/4}.

For z ∈ C× ,

log(π/2) z = Log |z| + i arg(π/2) z, where π/2 ≤ arg(π/2) z < π/2 + 2π = 5π/2.

The complex numbers with principal argument between 0 and π/4 are those whose arg(π/2) lie
between 2π (= 0 + 2π) and 9π/4 (= π/4 + 2π). It follows that the image of the wedge under log(π/2)
is the horizontal strip {w ∈ C | 2π < Im w < 9π/4}.

y v
9πi/4
•
•
2πi
w = log(π/2) z

x u

6.14 Determine the location of the branch cut of the function

w = f (z) = Log(1 − iz),

as well as the discontinuity of f (z) across it.

Hint: break up the transformation above as Z = 1 − iz followed by w = Log Z, as we did in
class.

Since Z = 1 − iz is an affine linear transformation, we can invert it to get z = iZ − i. The branch

point of w = Log Z is located at Z = 0, which gives z = −i. The branch cut of w = Log Z lies along
the negative real axis in the Z-plane, and that corresponds to those points in the imaginary axis of
the z-plane that are to the south of −i.
 We can also write some formulas to see the location of the branch cut. In the Z-plane, it is located
on the points Z = reiπ = −r with r ∈ R>0 (remember that Log = log(−π) ). Using the formulas for
the affine linear transformations above, we find that the branch but in the z-plane lies on the points
z = −i − ir with r ∈ R>0 .

y Y

Z = 1 − iz
0
x X
z = iZ − i
−2πi
−i

−2πi

6.15 Consider the transformation

1+z
w = f (z) = Log
1−z
(i) Determine the inverse image under the transformation
1+z
Z=
1−z
of the branch cut of w = Log Z. Use this information to determine where f (z) is holomor-
phic.
(ii) Find a pair (α1 , α2 ) for which the following two conditions hold:
− the function
g(α1 ,α2 ) (z) = log(α1 ) (1 + z) − log(α2 ) (1 − z);
is holomorphic exactly where f (z) is.
− f and g(α1 ,α2 ) coincide along the line segment joining the two branch points.
These two conditions imply, in fact, that f and g coincide everywhere. This follows from
the Principle of Analytic Continuation, which we will talk about—hopefully—in Chapter
5.
Hint: once you have identified the line segment joining the two branch points, you can explicitly
calculate the values of f and g(α1 ,α2 ) in terms of some real logarithms.
Note: the functions f and g(α1 ,α2 ) are branches of the multivalued function 2 argtanh z.

(i) The inverse of the given transformation is

Z −1
z=
Z +1
 It is of the form
az + b
Z= , a, b, c, d ∈ R, ad − bc > 0
cz + d
Remember that these preserve the upper half-plane—and hence the real line too. Hence the
inverse image by Z = (1 + z)/(1 − z) of the branch cut of w = Log Z must be contained in the
real line of the z-plane.
It is clear that the Z = 0 corresponds to z = −1. On the other hand, the negative real axis
in the Z-plane can be thought as the part of the real axis in the Z-plane between Z = 0 and
Z = ∞. But Z = ∞ corresponds to z = 1. Thus the branch cut in the z-plane can be one of
the two following things:
− the line segment between z = −1 and z = 1 along the real axis, or
− the “line segment” the same two points going the other way around—through z = ∞.
We see that it is, in fact, the second of these possibilities by looking at another point: Z = −1
comes from z = ∞. So it looks like
1+z
w = Log
1−z
has two branch points and two branch cuts! Or maybe we should just view it as one branch cut
connecting two branch points. In any case, that second branch point comes from Z = ∞—a
fact that we could interpret as saying that w = Log Z has a branch point at ∞ (it’s just that
we usually only draw the finite complex plane).
y Y

1+z
−2πi Z=
1−z
−1 1 0
x X
Z −1
−2πi z= −2πi
Z +1

(ii) In order to have the branch cut of g(α1 ,α2 ) (z) emanating from z = −1 going in the negative real
direction (as it does for w = f (z)), we can take α1 = −π. Similarly, we want the branch cut
starting at z = 1 to lie along the positive real axis; we can take also α2 = −π (the argument of
the logarithm is 1 − z, so the branch cut gets rotated as in Problem 6.14).
Let us check that the second of the conditions in the statement holds for this choice. Between
the two branch points, z = −1 and z = 1, both f (z) and g(−π,−π) (z) are holomorphic. As z
moves from z = −1 to z = 1, Z = (1 − z)/(1 + z) moves along the real axis from Z = 0 to
Z = ∞. But the principal branch of the logarithm takes real values on the positive real axis,
and so we have
1+x
f (x) = Log , −1 < x < 1.
1−x
 where Log is just the real logarithm.
On the other hand, both 1 + z and 1 − z are positive real for z on that segment, and so
Log(1 + z) and Log(1 − z) are also real on it, and
g(−π,−π) (x) = Log(1 + x) − Log(1 − x), −1 < x < 1
which coincides with f (x) above.

6.16 Find the image of the (open) unit disc D under the transformation w = f (z) in the last problem.

To compute the image of the unit circle |z| = 1 under the transformation Z = (1 + z)/(1 − z), it
is enough to take three points on said circle. We can take, for example, z = −1, z = 1 and z = i.
The first goes to Z = 0, the second to Z = ∞, and the third to (1 + i)/(1 − i) = i. Thus, the image
of |z| = 1 in the Z-plane is the imaginary axis, and the unit disc |z| < 1 goes to the right-hand plane
Re Z > 0 (z = 0 is sent to Z = 1).

On the other hand, w = Log Z maps the right half-plane in the Z-plane onto the strip −π/2 <
Im w < π/2. Hence the latter is also the image of the unit disc in the z-plane under the composed
transformation.

−2πi
−1 1
x
−2πi
1+z v
w = Log
1−z

πi/2
•

1+z
Z= u
1−z

Y
•
−πi/2

w = Log Z

0
X
−2πi