Scribd
Upload
21 views

Huth - Mixed Boundary Value Problems in Potential Theory

Uploaded by

nobber898989
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
21 views

Huth - Mixed Boundary Value Problems in Potential Theory

Uploaded by

nobber898989
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

MIXED BOUNDARY VALUE PROBLEMS IN POTENTIAL THEORY

BY

J. H; HUTH’

SUMMARY

Mixed boundary value problems of potential theory are of importance in a diver-


sity of applications. Generally they can best be solved by reduction to a Riemann-
Hilbert problem, but this involves certain arbitrary constants. This paper interprets
these constants in terms of dipoles.

INTRODUCTION

Perhaps the classical example of a “mixed” boundary value problem


is afforded by the case of steady-state temperature determination when
given the values of the dependent variable only over a part of the
boundary. Over the remainder of the boundary the normal derivative
is prescribed. In many such cases the usual Fourier series method of
solution is inadequate, and the problem is best handled by reduction to
the Riemann-Hilbert case (1) .2 Other examples of importance are to be
found in the field of thermal stress analysis (2).
Although the Riemann-Hilbert approach generally yields a result in
closed form, there remains the problem of interpreting a set of undeter-
mined constants. This paper treats such a problem.

ANALYSIS

Consider a unit circle in the z = x + ir plane, and let a function 4


be prescribed on the upper half of the boundary. On the lower half let

the normal derivative 2 be given. The problem is then to solve La-

place’s equation in the interior of the circle. As a first step consider 4


as the real part of an analytic function Cpof the complex variable z.
Now since a pure Neumann problem can be solved by standard tech-
a4
niques, no loss of generality is incurred by assuming an = 0 along the
lower portion of the boundary. It is then permissible to take the
imaginary part of + equal to zero thereon.
Consequently + can be extended to the whole plane by means of a
reflection. That is, let @ be defined for jz/ > 1 in terms of @ inside the
unit circle by means of the formula:

@(z)=cp + . (1)
0
1 Consulting Engineer, Pacific Palisades, Calif.
e The boldface numbers in parentheses refer to the references appended to this paper.

I21
122 J. 11. HlrTfl [J. F. I.

Then + will be analytic in the entire plane with the exception of a semi-
circular cut corresponding to the upper half of the original boundary.
Iff(0) represents the prescribed values of 4 along the arc, then + must
satisfy a cut condition
a+ + a- = 2 f(0) (2)
where the + and - superscripts indicate values obtained by approach-
ing the cut from opposite sides. In this form the problem is that of
Reimann-Hilbert. It should be noted however that this formulation as
stated does not have a unique solution, and not every solution will
necessarily solve the original real variable problem in 4.
Let X be a solution of the homogeneous problem :
Xf + x- = 0. (3)
X can then be taken in the form

(4)

provided that the branch line for the square root is taken along the cut.
The a,‘s are complex constants and their interpretation is precisely the
object of this paper. Before proceeding in this direction, the solution
of the original Riemann-Hilbert problem will be finished for the sake of
completeness.
Equation 2 can now be written in the form :

(;)+ - ($)- = 2&.

is expressible in integral form as follows by virtue of

Cauchy’s formula for arcs.

(6)

Here t is a running variable along the arc L. Of course the complete


solution of the Riemann-Hilbert problem can be written as:
m

C a,zn
@_dzz-
2si s
L G-7
2f0)
(t - 2)
dt + ;=k. (7)

The line integral can be easily related to a contour integral around the
cut and then evaluated by residues.
To interpret and obtain the restrictions on the a,‘s it first is necessary
to recall that the objective is to have X provide a non-trivial solution of
Feb., I954.1 MIXED BOUNDARYVALUE PROBLEMS 123

the original problem in real variables with 4 = 0 on the upper half, and
a+
- = 0 on the lower half of the unit circle boundary. But this problem
an
is capable of treatment by an entirely different approach. It is well
known that Mobius transformation (3) will map a half circle with base
horizontal in the w plane into a full circle in the z plane.
Specifically, let the problem in the w plane be to find a harmonic
function vanishing on the circular part of the boundary of a half-circle,
and with vanishing normal derivative along the base. Then Mobius
transformation will result in exactly the problem to be solved by X.
It is immediately seen that a complex dipole (and only a dipole)
placed at either corner of the half-circle, together with an additive
constant potential, will satisfy the boundary conditions (4). In other
words the homogeneous problem in the w plane can be solved by the
following expression :

X(w)=&4
[
,&-; 1
+B
[ &&+;I* (8)
Here A and B are arbitrary complex strengths of the left and right di-
poles, respectively. Written out, Miibius’ transformation reads :
. w2 + 1 + 2iw
’ = - ’ ~2 + 1 - 2iw’

Combining Eqs. 8 and 9 yields the final result:

Comparing Eq. 10 with Eq. 4 yields the necessary conditions on the


a,‘~ in order for the Riemann-Hilbert solution to also solve the real
variable problem. All the a,‘~ must vanish except a0 and al. Hence
X(z) must remain finite at infinity. Further, Eq. 10 establishes two
relations between the real and imaginary parts of a0 and al, so that only
two arbitrary constants remain. Of course the same result could have
been obtained directly from (7) by referring back to the original bound-
ary conditions on 4. However the above derivation gives the physical
significance of dipoles in the w plane.

CONCLUSIONS

It has been shown that specifying 4 and 2 along the two segments
of a boundary is, in general, not sufficient to completely determine a
J. H. Hum [J. F. I.
124

harmonic 4 in the interior of the region. In addition, it is necessary to


give information concerning singularities at the junctures of the seg-
ments. These singularities have been interpreted in terms of two arbi-
trary dipoles. These concepts can, of course, be extended to more
general cases involving a larger number of boundary segments.

REFERENCES
(1) N. I. MUSKHELISHVILI,“Singular Integral Equations,” Melbourne, Aeronautical Research
Laboratories, 1949.
(2) J. H. HUTH, “Thermal Stresses in a Partially Clamped Elastic Half-Plane,” J. A#. Phys.,
Vol. 23, p. 1234 (1952).
(3) E. G. PHILLIPS, “Functions of a Complex Variable,” New York, Interscience Publishers,
Inc., 1940.
(4) V. L. STREETER, “Fluid Dynamics,” New York, McGraw-Hill Book Co., Inc., 1948.

You might also like