![]()
![]()
for
Section 3.3 Harmonic Functions
Let
be
a real-valued function of the two real variables x and y defined on a
domain D. (Recall that a domain is a connected open
set.) The partial differential equation
(3-26)
is known as Laplace's
equation and is sometimes referred to as the potential
equation. If
are all continuous and if
satisfies Laplace's
equation, then
is called a harmonic
function.
Harmonic functions are important in the
areas of applied mathematics, engineering, and mathematical
physics. In calculus we might have been asked to show that
the functions
and
are harmonic (and also that the functions
and
are
harmonic). Harmonic functions are used to solve problems
involving steady state temperatures, two-dimensional electrostatics,
and ideal fluid flow. In Chapter 11 we will see how
complex analysis techniques can be used to solve some problems
involving harmonic functions. For example, the
function
is
harmonic in the upper half plane and takes on the boundary
values
and
, as
shown in Figure 3.A.
![[Graphics:Images/HarmonicFunctionMod._gr_13.gif]](harmonicfunction/HarmonicFunctionMod/Images/HarmonicFunctionMod._gr_13.gif)
Figure
3.A The harmonic function
.
These pairs of functions are not chosen at random, and there is an intimate relationship between them. We begin with an important theorem relating analytic and harmonic functions.
Theorem
3.8. Let
be
an analytic function on a domain D. Then
both
and
are
harmonic functions on D. In
other words, the real and imaginary parts of an analytic function are
harmonic.
Proof. In Corollary 6.3 we will show that, if f(z) is
analytic, then all partial derivatives of u and v are
continuous. Using that result here, we see that, as
f is analytic, u
and v satisfy the Cauchy-Riemann
equations
and
.
Taking the partial derivative with respect to x
of each side of these equations gives
and
.
Similarly, taking the partial derivative of each side with respect to
y yields
and
.
The partial derivatives
are all continuous, so we use a theorem from the calculus of real
functions that states that the mixed partial derivatives are equal;
that is,
and
.
Combining all these results finally gives
, and
.
Therefore both u and v
are harmonic functions on D.
If we have a function
that is harmonic on the domain D and
if we can find another harmonic function
such that the partial derivatives for u
and v satisfy the Cauchy-Riemann
equations throughout D, then we say
that
is
a harmonic conjugate of
. It
then follows that the function
is
analytic on D.
Example
3.11. If
, then
; hence
u is a harmonic function for all z. We
find that
is also a harmonic function and that
, and
.
Therefore v is a harmonic conjugate
of u, and the function f
given by
is an analytic function.
Theorem 3.8 makes the construction of harmonic functions from known analytic functions an easy task.
Example 3.12. The
function
is
analytic for all values of z, hence it follows that
is harmonic, and
is a harmonic conjugate of u(x,y).
Figures 3.2 and 3.3 show the graphs of
these two functions. The partial derivatives
are
,
,
, and
. They
satisfy the Cauchy-Riemann equations because they are the real and
imaginary parts of an analytic function. At the point
, we
have
, and
these partial derivatives appear along the edges of the surfaces for
u and v where
. Similarly,
and
also
appear along the edges of the surfaces for u
and v where
.
![[Graphics:Images/HarmonicFunctionMod._gr_80.gif]](harmonicfunction/HarmonicFunctionMod/Images/HarmonicFunctionMod._gr_80.gif)
Figure
3.2
. Figure
3.3
.
We can use complex analysis to show easily
that certain combinations of harmonic functions are
harmonic. For example, if v
is a harmonic conjugate of u, then
their product
is a harmonic function. This can be verified directly by computing
the partial derivatives and showing that Laplace's equation
(3-26) holds, but the details are
tedious. If we use complex variable techniques instead, we
can start with the fact that
is an analytic function. Then we observe that the square
of f is also an analytic function,
which is
.
We then know immediately that the imaginary
part,
, is
a harmonic function by Theorem 3.8. Since a constant
multiple of a harmonic function is harmonic, it follows that
is harmonic. It is left as an exercise to show that if
and
are two harmonic functions that are not
related in the preceding fashion, then their product need
not be harmonic.
Theorem 3.9 (Construction of a
Conjugate). Let
be
harmonic in an
-neighborhood
of the point
. Then
there exists a conjugate harmonic function
defined
in this neighborhood such that
is an analytic function.
Proof. A conjugate harmonic function v
will satisfy the Cauchy-Riemann equations
and
. Assuming
that such a function exists, we determine what it would have to look
like by using a two-step process. First, we integrate
(which should equal
)
with respect to y and get
(3-27) ![[Graphics:Images/HarmonicFunctionMod._gr_99.gif]](harmonicfunction/HarmonicFunctionMod/Images/HarmonicFunctionMod._gr_99.gif)
where
is a function of x alone that is yet
to be determined. Second, we compute
by differentiating both sides of this equation with respect to
x and replacing
with
on
the left side, which gives
![[Graphics:Images/HarmonicFunctionMod._gr_104.gif]](harmonicfunction/HarmonicFunctionMod/Images/HarmonicFunctionMod._gr_104.gif)
It can be shown (we omit the details) that because u is harmonic, all
terms except those involving x in the last equation will cancel,
revealing a formula for
involving x
alone. Elementary integration of the single-variable
function
can then be used to discover
. We finally observe that the function v
so created indeed has the properties we seek.
The functions
and
are computed with the formulas:
, and
.
Technically we should always specify the domain of function when defining it. When no such specification is given, it is assumed that the domain is the entire complex plane, or the largest set for which the expression defining the function makes sense.
Method. Construction of the Harmonic Conjugate of u(x,y).
Execute this cell to activate the construction of the harmonic conjugate subroutine.
Example 3.13. Show
that
is
a harmonic function and find the harmonic
conjugate
.
Solution. We follow the construction process of Theorem
3.9. The first partial derivatives are
(3-28)
and
.
To verify that u is harmonic, we
compute the second partial derivatives and note
that
, so
u satisfies Laplace's Equation
(3-26). To construct
v, we start with Equation
(3-27) and the first of Equations
(3-28) to get
![[Graphics:Images/HarmonicFunctionMod._gr_118.gif]](harmonicfunction/HarmonicFunctionMod/Images/HarmonicFunctionMod._gr_118.gif)
Differentiating the left and right sides of this equation with
respect to x and using
and
Equations (3-28) on the left side
yields
![]()
which implies that
,
then an easy integration yields
, where
C is a constant. Now we substitute
and obtain the desired solution
.
Contour plots for the families of
curves
and
are
shown in Figures 3.B.
![[Graphics:Images/HarmonicFunctionMod._gr_147.gif]](harmonicfunction/HarmonicFunctionMod/Images/HarmonicFunctionMod._gr_147.gif)
Figure
3.B Contour plot for
and
.
In Chapter 11 we will show that
if
is an analytic function. Then the two families of level
curves
and
, are
orthogonal in the sense that if
is
a point in common to the two curves
and
, and if
, then
these two curves intersect orthogonally. The orthogonal
grid shown in Figure 3.C.
![[Graphics:Images/HarmonicFunctionMod._gr_157.gif]](harmonicfunction/HarmonicFunctionMod/Images/HarmonicFunctionMod._gr_157.gif)
Figure
3.C The orthogonal grid formed by
and
.
Extra Example
1. Show that
is
a harmonic function and find the harmonic
conjugate
.
Fluid Flow
Harmonic functions are solutions to many
physical problems. Applications include two-dimensional
models of heat flow, electrostatics, and fluid flow. We
now give an example of the latter.
We assume that an incompressible and
frictionless fluid flows over the complex plane and that all cross
sections in planes parallel to the complex plane are the
same. Situations such as this occur when fluid is flowing
in a deep channel. The velocity vector at the point
is
(3-29)
.
The assumption that the flow is
irrotational and has no sources or sinks implies that both the curl
and divergence vanish, that is,
and
. Hence
p and q
obey the partial differential equations
(3-30)
, and
.
Equations (3-30) are similar to the
Cauchy-Riemann equations and permit us to define a special complex
function:
(3-31)
.
Here we have
. We
can use Equations (3-30) to verify that
the Cauchy-Riemann equations are satisfied for f:
, and
.
Assuming the functions p and
q have continuous partials, Theorem
3.4 guarantees that function f
defined in Equation (3-31) is analytic,
and that the fluid flow of Equation
(3-29) is the conjugate of an analytic
function, that is,
.
In Chapter 6 we will prove that every
analytic function f has an analytic
antiderivative F; assuming
this to be the case, we can write
(3-32)
,
where
.
Theorem 3.8 tells us that
is a harmonic function. If we use the vector
interpretation of a complex number we see that the gradient of
can be written as
.
The Cauchy-Riemann equations applied to
give
; making
this substitution in the last equation yields
.
Equation (3-14) says
that
, which
by the preceeding equation and Equation
(3-32) imply that
.
Finally, from Equation
(3-29),
is the scalar potential function for the a fluid flow, so
.
The curves given by
are called equipotentials. The curves
are called streamlines and describe the path of fluid
flow. In Chapter 11 we will see that the family of
equipotentials is orthogonal to the family of streamlines.
Example 3.14. Show
that the harmonic function
is
the scalar potential function for the fluid
flow
.
Solution. We can write the fluid flow expression as
.
An antiderivative of
is
, and
the real part of
is the desired harmonic function:
.
Note that the hyperbolas
are
the equipotential curves and that the
hyperbolas
are
the streamline curves; these curves are orthogonal, as shown in
Figure 3.6.
![[Graphics:Images/HarmonicFunctionMod._gr_207.gif]](harmonicfunction/HarmonicFunctionMod/Images/HarmonicFunctionMod._gr_207.gif)
Figure 3.6 Blue
equipotential curves
,
and red streamline curves
,
for
.
Library Research Experience for Undergraduates
Download this Mathematica Notebook
Return to the Complex Analysis Project
(c) 2006 John H. Mathews, Russell W. Howell