List of Books Recommended for Further study LIST OF BOOKS RECOMMENDED FOR FURTHER STUDY Agmon, Shmuel, Lectures on elliptic boundary value problems, Van Nostrand, 1965. Bers, Lipman, John, F., Schechter, M., Partial differential equations, Interscience Publishers, 1964. Carroll, R. C., Abstract methods in partial differential equations, Harper and Row, 1969. Courant, R., and Hilbert, D., Methods of mathematical physics, Interscience Publishers, Vol. I, 1953, Vol. II, 1962. Duff, G. F. D., Partial differential equations, University of Toronto Press, 1950. Fichera, Gaetano, Linear elliptic differential systems and eigenvalue problems, Lecture Notes in Mathematics, 8, Springer Verlag, 1965. Friedman, Avner, Partial differential equations of parabolic type, Prentice Hall, 1964. Partial differential equations, Holt, Rinehart and Winston, 1969. Garabedian, P. R., Partial differential equations, John Wiley and Sons, 1964. Hellwig, G., Partielle Differentialgleichungen, Teubner, 1960. lf6rmander, Lars, Linear partial differential operators, Springer Verlag, 1963. John, Fritz, 215 Plane waves and spherical means applied to partial differential equations, Interscience Publishers, 1955.
36
Embed
List of Books Recommended for Further study 215 …978-1-4615-9979... · 2017-08-27 · List of Books Recommended for Further study ... Friedman, Avner, Partial differential equations
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
List of Books Recommended for Further study
LIST OF BOOKS RECOMMENDED FOR FURTHER STUDY
Agmon, Shmuel,
Lectures on elliptic boundary value problems,
Van Nostrand, 1965.
Bers, Lipman, John, F., Schechter, M.,
Partial differential equations,
Interscience Publishers, 1964.
Carroll, R. C.,
Abstract methods in partial differential equations,
Harper and Row, 1969.
Courant, R., and Hilbert, D.,
Methods of mathematical physics,
Interscience Publishers, Vol. I, 1953, Vol. II, 1962.
Duff, G. F. D.,
Partial differential equations,
University of Toronto Press, 1950.
Fichera, Gaetano,
Linear elliptic differential systems and eigenvalue problems,
Lecture Notes in Mathematics, 8, Springer Verlag, 1965.
Friedman, Avner,
Partial differential equations of parabolic type,
Prentice Hall, 1964.
Partial differential equations,
Holt, Rinehart and Winston, 1969.
Garabedian, P. R.,
Partial differential equations,
John Wiley and Sons, 1964.
Hellwig, G.,
Partielle Differentialgleichungen, Teubner, 1960.
lf6rmander, Lars,
Linear partial differential operators,
Springer Verlag, 1963.
John, Fritz,
215
Plane waves and spherical means applied to partial differential equations,
Interscience Publishers, 1955.
LIST OF BOOKS RECOMMENDED FOR FURTHER STUDY
Lax, P. D.,
Lectures on hyperbolic partial differential equations,
Stanford University, 1963.
Lions, J. L., and Magenes, E.,
Probl~mes aux limites non homog1nes et applications,
Dunod, 1968, Vol. I, II.
Lions, J. L.,
Equations differentieles op:rationelles et probl~mes aux limites,
Springer, 1961.
Mikhlin, S. G.,
Linear equations of mathematical physics,
Holt, Rinehart and Winston, 1967.
Petrovsky, I. G.,
Lectures on partial differential equations,
Interscience Publishers, 1954.
Smirnov; V. I.,
A course of higher mathematics, Vol. IV, Translation,
Addison Wesley, 1964.
Tikhonov, A. N., and Samarskii, A. A.,
Equations of mathematical physics, Translation,
Pergamon Press, 1963.
Treves, F.,
Linear partial differential equations with constant coefficients;
existence, approximation and regularity of solutions,
Gordon and Breach, 1966.
Problems
PROBLEMS
Chapter I
Section 1
1)
2)
Solve xu -yu = ° with initial condition u(x,O) y x
f.~ arc taney/x) (Answer: u = f,\Jx-+y- ) e 1
2 Solve u +u = u with initial condition u(x,O) x y
(A () f(x-y) 1 nswer: u x,y = l-yf(x-y)
f(x) .
f(x).
3) Prove directly that the only solution of
xu +yu +u = 0 x y
that is continuously differentiable in the square I xl :S a, I yl < a
is u ~ 0. (Hint: Prove that Max u:S 0, Min u~ 0).
4) Let u(x,y) be the solution of the quasi-linear equation
5)
u +a(u)u = ° y x
with initial condition u(x,O) = f(x).
a) Prove that the characteristic line Cs through the point (s,O) is
the straight line x = s + a(f(s)).
b) Prove that for
yes) d a(f(s))
ds <0
Ux becomes infinite on Cs for t = -l/y(s).
Solve u = xuu with initial condition u(x,O) = x. y x
(Answer: Implicitly x = ue -yu 1.
Section 2
1)
2)
3 For the differential equation u = u y x
a) Find the solution with u(x,O) = 2x3/ 2
(Answer: u = 2x3/ 2 (1_ 27y)-1/2 1.
b) Prove that every integral surface consists of straight lines.
c) Prove that every solution regular for all x,y is linear.
2 2 2 For the differential equation u +u = u find x y
217
218 PROBLEMS
a) The characteristic strips
Answer:
-~ z - 1-2tzO
~ q = 1-2tzO
b) The integral surfaces through the circle x = cos s, y = sin s, z 1.
[Answer: 122
z = exp [+ (1- oJ x - +y ~ )]
c) The integral surfaces through the line x = s, y = 0, z 1.
[Answer: +y
u=e- ].
3) a) Find the first order equation satisfied by the family of planes
z = x cos a + y sin a + b with parameters a, b.
[Answer: 2 2
u + U = 1]. x Y
b) Find the general characteristic strip for the equation.
p cos a, q = sin a ]
z x cos a + y sin a + b,
° -x sin a + y cos a + c,
c) Find the conoid with singularity (~,~,S).
[Answer:
Section 3
222 (z-s) = (x-~) + (y-~) ].
1) Find the solution u(x,y,z) of xu +yu +u x y z
u(x,y,O) = f(x,y).
[Answer: -z -z z
u = f(xe ,ye )e ].
u with initial condition
2) Euler's differential equation for homogeneous functions is given by
xlu + x2u + ... + xu = a u, (a = const.). xl x2 n xn
Prove that the general solution has the form
with a suitable function ~
Problems
Chapter II
Section 1
Construct the solution through the initial manifOld]
xl = 1, z = f(~, •.. ,Xn)
1) Notation of Laurent Schwartz.
We combine n indices il, ••• ,in (each i k a nonnegative integer) into
a "multi-index" i = (il, ••• ,in). For this multi-index i we define
Similarly ai
219
For any vector S = (Sl' ••• 'Sn) with n components we define si to be the
product
Using the symbol Dk for the partial differentiation ofO~, we introduce
the" gradient vector" D = (D l , ••• ,Dn) • The general higher order differentia
tion operator is then given by
. iii Dl _ DID 2 n
- 1 2 ••. Dn
Prove the validity of the following formulae for vectors x = (xl, ••• ,xn),
y = (Yl, •.• ,Yn) and multi-indices i,j,k
a) (X+y)i = L. ~ j+k=i j! k!
b) Di(f(x)g(x)) = L. .~!k' j+k=i J. •
c) 1 = L. xi (I-Xl) (1-X2)··· (l-xn) i
(binomial theorem).
. k (DJf(x) (D g(x)) (Leibnitz rule).
if I Xli < 1, I x21 < 1 , •.. , I xnl < 1 .
d)
220 PROBLEMS
if I xlI + IX21 + .•. + IXnl < 1.
e) (Xl + x2 + .•• + X ) m __ '\' m! Xi ~ for any positive integer m.
n IiI =m i!
2) Write the formulae for power series on pp. 48, 49 in the Schwartz notation.
3) Let u(x) be the analytic function of the single variable X represented
by the power series
00 n u(x) = c I. x2
n=l n for I xl < 1
where c is a positive real constant.
a) Prove that u majorizes u2, if c is sufficiently small.
b) Express u(x) in integral form
[Answer: u(x) = -c JX log(l-S) o S
dS 1.
Section 2
1) Show that every cone with vertex at the origin is a characteristic surface
for the differential equation
(0: = const.).
2) Decompose the Laplace expression
~u=u + .•• +u xl xl xnxn
into a tangential and a normal second derivative on the sphere
2 2 + ••• + x
n = a •
ill = Tu + Nu, where
-2 Nu = a 2:: x.x. u ]
. k J K x.~ J, J
3) Find the characteristic curves for the Tricomi equation
(y> 0).
[Answer: 3/2 3x.:!: 2y = const. 1.
4) Find the ordinary differential equation for the characteristic curves for a
solution u of the minimal surface equation (13), p. 4.
Problems
~ =-dx
u u +ijl+U 2+u 2 ] x y- x Y
1 + U 2 Y
5) Write the differential equation (1), p. 54 and the characteristic condition
(39), p. 65 in the notation of L. Schwartz explained in Problem 1) of
Chapter II, Section 1.
Section 3
1) For the differential equation
(A) U -u U = ° xxyyy
a) find solutions of the form
u = f(x)g(y) ,
where f(O) = 00, f(oo) = 0, g(O)
[Answer: u = y3/3x2 J •
g' (0) °
221
b) find solutions of (A) which satisfy an additional relation of the form
(B) u = f(u) with f(O) = ° y x
with a suitable function f, and which vanish on the curve
c~" 1/3 given by f(p) = (3p) , while u is
1 2 + 3t,
1 4 2 3 x = - 6% y = "3bClo - 3Clo t, u = -2Clo t
in parameters Clo' t
2 Y = x
] .
c) show that for the solution u of the preceding question the charac-
teristic curves belonging to u as a solution of (B) form one of the
two families of characteristic curves belonging to u as a solution
of (A).
[Hint: Use f2f' = 1 J.
Section 4
1) Find the solution of u = u +u with initial conditions yy xx
2 u(x,O) = l+x ,
u (x,O) = ° in closed form, by expansion of u(x,y) according to powers y
of y.
222 PROBLEMS
[Answer: u = (1+x2) cosh Y + y sinh y ].
2) Derive the expression (38), p. 85 for Q fram the differential equation and
initial condition stated, using the methods of Chapter I.
3) Observe that the solution of the ordinary differential equation problem
u t = 1+u2 with u(O) = 0
is majorized by that of
u t =.l:... with u(O) = O. l-u
Hence deduce an upper bound for the coefficient
series expansion of tan x.
[Answer: for n> 2 ].
Chapter III
Section 1
1) Let
c n
L1U = au + bu + cu, x y L2u = du + eu + fu, x y
where a, b, c, d, e, f are constants with ae-bd I O.
in the power
a) Prove that necessary and sufficient for the two first order equations
to have a cammon solution u in a convex region is that L2f = Llg.
[Hint: Prove first for Ll , L2 of the special form where a = e = 1,
b = d = O. Then write in the general case Ll and L2 as linear
combinations of operators of this special form ].
b) Show that any solution u of the second order equation L1L2u = 0
can be represented in the form u = ~ + ~,where Ll~ = L2u2 = O.
[Hint: Use result of a) J.
c) Let L be the second order operator with constant coefficients defined
by
Problems
2)
Lu = Au + 2Bu + Cu + 2Du + 2Eu + Fu. xx xy yy x y
When can L be represented as a product L = L1L2 of first order
operators?
[Answer: When
A B D
B C E o D E F
(Use the condition for a quadratic form
222 A~ +2B~~+C +2D~~+2E~~+F~
to be a product of linear forms) 1.
Find the solution u(x, t) of x2 with initial conditions
ut = 0 for t = o.
223
u = x,
[Answer: 122124
u = x + 2 x t + 12 c t , found by using a special solution of the
inhomogeneous equation which is independent of t 1.
3) For a fixed constant c 1 0 define u and s as functions of x,y by the
implicit equations
x + ct =.J2 cos( u+s) , x-ct =.J2 sin(u-s)
near x = 0, t = 1/ c, u = 0, s = Tr /4.
a) Prove that u is a solution of
b) Show that the curves u = const. are ellipses, and find their
envelope.
c) Find u and as functions of x on the curve u = o.
Section 2
1) Let rl 2"(l
L - - - c 2' where - ot2 ox c const.
a) Prove that for Lu = Lv = 0 also
b) Prove that if Lu = Lv = 0 for a < x < b, t > 0 and u = 0 for
x.= a,b and t> 0 then
224 PROBLEMS
d b 1 2 -- f - (u v + c u v ) dx = ° for t > 0. dt a 2 t t x x
2) For a solution of the wave equation given by (29), (31), p. 100 express the
energy
in terms of the an and ~n'
3) Find the solution of the following initial boundary value problem for the wave
equation in closed form:
u _c2u = ° for x ~ 0, t ~ ° tt xx
u(o,t) h(t) for t ~ ° u(x,o) f(x) , ut (x,O) g(x) for x ~ °
where f,g,h are given functions with continuous second derivatives for non-
negative arguments, and moreover
h(O) = f(O), h'(O) = g(o) , h"(O) = c2f"(0).
Verify that the solution obtained has·continuous second derivatives even on
the characteristic line x = ct.
[~nswer: Using the expression u = ~(x+ct) + * (x-ct) for the general
solution, one finds that
u= 1 x+ct
f(x+ct)+f(x-ct) + __ f g(~) d~ 2 2c x-ct
for ° < ct < x
f(ct+x)-f(ct-x) + ~ fct+x X U = .::....>...:=.:.:~:..>.:::-"--=':L.. g(~) d~ + h(t- -)
2 2c ct-x c for ° < x < ct 1.
4) Find the solution u(x,t) of the following initial-boundary value problem
("vibration of string plucked initially at center"):
Utt-Uxx = ° for 0 < x < v, t > 0
u(o,t) = u(v,t) = ° for t > ° u(x,O) = ~ -I~ -xl, ut(x,O) ° for ° < x < v.
Problems 225
(Answer: u(x,t) 4 00 (-1) m
- L - - (cos(2m+l)t) (sin(2m+l)x) 7r m=O (2m+l)2
] .
Section 3
1) Let u(x) = u(xl, .•• ,xn) satisfy the n-dimensional Laplace equation
.6u=u + •.• +u =0 xlxl xnxn
in an open set n. Prove that u has the ~
value property: If a solid sphere with boundary S and center x is con-
tained in n, then u(x) is equal to the arithmetic average of the values of
u on S.
[Hint: Use (8), p. 103 ].
2) Let u(x,t) = u(xl ,x2,x3,t) be a solution of the 3-dimensional reduced wave
equation
3)
2 .6u + A u = 0, (A = const.)
Then v(x,t) = u(x)eimt is a solution of the wave equation
provided A = 2ill/c). Let I(x,r) denote the arithmetic average of the values
of u on the sphere of center x and radius r, where we assume that both
the sphere and its interior lie in the domain of definition of u. Prove
that u has a generalized mean value property expressed by
( sin Ar () I x,r) = ~ u x •
(Hint: Use (8), p. 103 to show that 2 rI +21 +A rI
rr r
I(x,O) = u(x) ].
0, while
Let f(x) = f(Xl, .•• ,x) be a function with spherical symmetry, that is f . n
is of the form
where r = .;; 2+ ••. +x 2 1 n
a) Prove that the Laplace Operator applied to f is given by
n-l M = cp"(r) + - cp'(r).
r
b) Find all solutions u with spherical symmetry of the n-dimensional
Laplace equation .6u = 0.
226
[Answer: u= {A log r + B
2-n A r + B
constants A, B ].
PROBLEMS
for n = 2 with suitable
for n> 2
c) Find all solutions u of the n-dimensional bi-harmonic equation
62u = 0 with spherical symmetry.
[Answer:
2 2 u _ {A + Br + Clog r + Dr log r
- 2 2-n 4-n A + Br + Cr + D r
for
for n> 2
with suitable constants A,B,C,D ].
d) Find all solutions of the 3-dimensional reduced wave equation
6u + A2u = 0 with spherical symmetry. (Compare with Problem 2).
[Answer: A cos Ar + B sin Ar
u= r
(A,B = const.) ] .
)1-) a) Prove that the most general spherically symmetric solution (" spherical
wave") of the 3-dimensional wave equation 2 Utt = c 6u has the form
u = F(r+ct) +G(r-ct)
r
(The condition G(-s) = -F(s) for s > 0 has to be imposed to make
this expression for u meaningful for r = 0.)
[Hint: For u = fer ,t) we have 2 c (rf)rr = (rf)tt ].
b) Find the solution u(x,t) of the same equation with initial data
u=cp(r),
Find the value of u(O,t).
[Answer: The solution is unique; assuming it to have spherical sym-
metry, we find from the preceding question that
u = (r+ct)p(r+ct) + (r-ct)p(r-ct) 2r
u(x,O) = cp(ct) + ct cp'(ct) ] .
c) Take in part b) the sequence of initial functions given by
) 1/4
() (l+COS r 1
cp r = +-n n n3
Prove that for the corresponding spherical wave solutions u (x,t) n
Problems 227
lim u (x,t) = ° n -4", n
uni~ormly for all x and ~or
while on the other hand
0< t <~ -€ with any ~ixed positive - c
. (If 1 l~m u 0,---)='" n c n n -4 '"
(This indicates some lack o~ continuous dependence o~ the solution o~
the 3-dimensional wave equation on earlier values. However, formula
(17), p. 105, shows that u depends continuously on values of u and
values o~ its first derivatives at earlier times.)
d) Find the solution with initial data
u(x,o) = 0, 1 for r < 1
ut(x,O) = { ° ~or r> 1
[Answer: Use the results o~ (4a), to ~ind
° ~or r > ct+a, t > ° and ~or r < ct-a, t > a/c
u(x,t) ~or r < -ct+a, ° < t < a/c 2 2
a -fr - ct) I I - - for a-ct < r < a+ct, cr t> ° ] .
Section 4
1) In Tspace dimensions find the solution u(xl ,x2 ,x3,t) = u(x,t) o~ the in-
2 homogeneous wave equation Utt-C 6u = w(x,t) with initial data
u(x,o) = ut(x,O) = ° where
~or r = <a w(x, t)
~or r> a
[Answer: Use (2), p. 110 and problem 4d), section 3, to ~ind
€,
228
u(x,t)
for r < a, a-r
0< t <c
3a2(r+ct)+(r-ct)3-2a3-2r3
12c2r for r < a, a-r < t < ~
c c
for r < a,
o for a < r,
~<t c
o < t < r-a c
-3a2(r-ct)+(r-ct)3+2a3
12c2r for a < r,
a3 ---2- for a < r, 3c r
r+a < a c
~<t<~ c c
2) The same as problem 1) with w(x,t) defined by
w(x,t) { sin mt
o
for r < a
for r > a
but determine the solution u(x,t) only for r:a < t, a < r.
[Answer: . am am c Sln - - am cos --u = ______ ~c __ ~ ______ ~c~
m3r
Section 5
sin m(t - ~) c 1.
PROBLEMS
J •
1) Prove that under appropriate regularity assumptions we have for two functions
u(x,y), v(x,y) defined in a domain D' of the xy-plane with boundary B'
the identity
2 2 II (u6v - v~ u) dxdy D'
J. (u d~: - ~v ¥n + 6u ~: - v d~~) dS .
[Hint: Apply (6), p. 118 twice J.
2) Prove the analogue of Green's second identity (6), p. 118 for functions
xl, •.• ,xn defined in a region D with boundary B:
Problems
f(u(y).6v(y) - v(y)6u(y)) dy = I(u(y) ~ - v(y) ~) dB, D B n n
where on the left we have an n-tuple, on the right an (n-l)-tuple integral,
and
y = (Yl' ••• 'Yn)' dy = volume element = dYldY2 ••• dYn'
dS (n-l)-dimensional area element of B
229
(u and v are assumed to have continuous second derivatives in the closure
of D, and D is supposed to be sufficiently regular for application of the
divergence theorem. The same type of assumptions is made in the subsequent
problems) •
3) Prove the following extension of formula (24), p. 122 for a function
u(X) = u(xl, ••• ,xn) of n independent variables in an n-dimensional region
D with boundary B: For n = 2
21Tu(x) = I (u(y) agog r - log r ~)dB + ff (log r)6u(y) dy, B n n D
for n > 2
(2-n)(l) u(x) n
a 2-n au(y) I (u(y) +n- - r 2- n +n-)dB + I l-nllu(y) dy B n n D
Here x is an interior point of D and r stands for the distance
area of the unit sphere in n dimensions.
[Hint: Follow the same arguments as in the proof of (24), p. 122 making
use of problem 3b), section 3 ].
4) Derive an analogous formula as in the preceding problem in 3 space dimensions
with the Laplacean 6 replaced by the "redteed wave operator" 6 + A.2•
[Answer: Use problem 3d), section 3.
-47fu(x) = ff (ufn - v¥n) dB + Iff V(llu+A.2u) dy, B n n D
where v=~ ]. r
5) Derive an identity analogous to that in the two preceding problems, involving
230 PROBLEMS
the bi-harmonic operator 62 in two dimensions.
[Answer: Use problems 3c), section 3, and 1), section 5.
87ru(x) = I (u¥n - 6V~ + L'iufn - v~) dS + II v62u dy, B n n D
where 2
v = r log r J.
6) The maximum principle for harmonic function in the form of Theorem 4, p. 124,
can also be proved as follows, without invoking the mean value property: Let
u be twice continuously differentiable in the open bounded set D with
boundary B, and continuous in the closure D = D+B. Let 6u = 0 in D. Let w
be any sufficiently regular function with 6w > 0 in D and € be any
positive constant. Set v = u + €W, so that ~ > O. Then v has no maximum
in D, since at least one of the numbers
Max v < Max v , IT - B
v xx
Max u < Max v - € Min w < Max v - € Min w IT -1) D -B 15"
< Max u + E Max w - E Min w B B D
or v is positive. yy
For E ~O we obtain the desired inequality Max u < Max u. 1) - B
Thus
a) Prove the analogous maximum property for solutions of the Laplace equa-
tion 6u = 0 in any number of dimensions.
b) Prove the maximum property for solutions of the two-dimensional elliptic
equation
Lu = au + 2bu + cu + 2du + 2eu = 0 xx xy yy x y
where a, b, c, d, e are continuous functions of x,y in D+B, for
which ac_b2 > 0, a > O.
[Hint: Prove first the maximum property for a solution of Lv> 0,
using that at a maximum point of v in D
2 2 vxxS + 2v S~ + v ~ < 0 for all s,~.
xy yy
Then choose v = U+€W where
Problems 231
outside D+B, and M sufficiently large J.
7) Prove that not all solutions of the reduced wave equation ~u + \2u = 0
(with \ > 0) in 3 dimensions have the maximum property.
[Hint: See problem 3d), section 3 J.
8) Show that a harmonic function u(x,y) of two real independent variables in a
domain D is an analytic function of x and y in D in the sense of p. 48.
[Hint: Show, using the estimate (28), p. 124 and Taylor's formula with error
term, that u is locally representable by power series J.
9) Show that u(x) defined in an open set has continuous derivatives of all
orders if u is a solution of
a) The Laplace equation in n-space.
The reduced wave equation in 3-space. b)
c) The bi-harmonic equation in the plane.
(assuming in each case that u has continuous derivatives of the order oc-
cur ring in the differential equation).
[Hint: Use the integral representations for u from problem 3) and differen-
tiate under the integral sign J.
10) Let u(x,y) be a harmonic function (i.e. solution of Laplace's equation
u +u = 0) in the simply connected open set D. xx yy
a) Prove that there exists a conjugate harmonic function v(x,y) in D
such that the Cauchy-Riemann equations
u =-v y x
are satisfied.
[Hint: Let (xo' yO) be a fixed point in D. For any (x,y) in D
define v by
PROBLEMS
v(x,y) (u dy - u dx) x y
where the integral is taken along any path in D joining (xo'yo) to
(x,y) ].
b) Introducing the complex valued function f = u+iv of the complex argu-
ment Z = x+iy, show that fez) has a derivative in the sense that for
a sequence zn = xn+iYn with limt Z = x+iy in D
lim n --+ 00
fez )-f(z) n Z -z n
= u (x,y) + iv (x,y) x x
independently of the manner in which xn tends to x and Yn to y.
[Hint: Apply the mean value theorem of Differential Calculus to
fez )-f(z) = (u(x ,y )-u(x,y)) + i(v(x ,y )-v(x,y) 1. n n n n n
c) Prove Cauchy's theorem: For any closed curve C in D
I f(z)dz = I (u+iv) (dx+idy) = 0 C C
[Hint: There exist functions ~(x,y), *(x,y) in D with
d~ u dx - v dy, d* = v dx + u dy ].
Section 7
1) Derive an integral representation for solutions of Laplace's equation in 3-
space, analogous to (4), p. 146, defining an appropriate Green's function.
[Answer: Define the Green's function for the domain D with boundary B by
1 G(x;~) = - 4'1Tr + w(x,~)
where x = (xl ,x2,x3), ~ = (~1'~2'S3) and r is the distance of the points
x and ~. Moreover w is for fixed x in D a harmonic function of s chosen so that G(x;~) = 0 for ~ on B. Then for u(x) harmonic in D
we can represent u in terms of its boundary values on B by the formula
u(x) = II u(~) dG£X;~) dS B n
where the variable of integration ~ ranges over the surface B with surface
Problems
element ds, and the normal derivative of G is taken with respect to the
variable !; 1.
2a) Find an expression for the Green's function for the unit sphere in 3-space,
analogous to the expression (6), p. 147.
[Answer: -41T G(x;!;) 111 =--------I ~-xl I ~-x'l I xl
where we write I xl for J x12 + x22 + x32 and hence I x- ~I for r, and
where x' is the point given by x' = xiI xl 2 1.
233
2b) Derive the 3-dimensional analogue of Poisson's integral formula (13), (130).
[Answer: u(x) = ~ ff 1-1 xl ~ f( g) dS ].
I gl =1 I x-sl
3a) Find the Green's function for the two-dimensional Laplace equation correspond-
ing to the upper half-plane.
[Answer: -l-iz Use (11), p. 148 and the mapping F = l-iz of the half plane onto
the unit circle
where x' = (xl ,-x2) ].
3b) Find the corresponding integral representation for the solution of the
Dirichlet problem for the upper half-plane: ~ u(xl ,x2) = ° for x2 > 0,
u(xl,O) = f(Xl )·
[Answer: 1 {'" 1T
] . -00
3c) Show that the preceding integral formula actually represents a solution u
of the Dirichlet problem, i~ f(s) is bounded and is continuous. (Observe
the non-uniqueness, since we can e.g. add ~ to u).
[Hint: With u defined by the integral we have
234 PROBLEMS
Split the integral, into two parts corresponding to IX1-~11 < 0 and
IX1-~11 > 0, where 0 is such that I f(~l)-f(Xl)1 < € for IX1-~11 < O.
Estimate the two parts, using that If I is bounded and show that
lim (u(xl ,X2)-f)) = O. x2 ---70
Prove also that u(xl ,x2) is harmonic J.
4) Find the Green's function for the first quadrant of the xl x2 plane.
5)
6)
[Answer:
where
X= J.
Prove that .6u(x) .6u(Xl , x2 '· .• ,xn) = 0 implies that also
.6(lxI 2- n u(x/I xl 2) = 0
for x/I xl 2
in the domain of definition of u.
On p. 150 a fundamental solution u(x,y) of the 2-dimensional Laplace equa-
tion (with "pole" (~,1'])) was characterized by the symbolic equation
.6u = o(x, y; ~,1']) = "Dirac function". This equation stands for the" concrete"
identity (19), p. 150 that is obtained by formally mUltiplying the symbolic
equation by an arbitrary function v(X,y), and integrating by parts until all
derivatives of u have been removed by the integrand. (Identity (19) has a
direct elementary meaning, since the function u behaves like log r and is
integrable, while its second derivatives are not). In a similar way (related
to the theory of "distributions" in the sense of Laurent Schwartz) we can de-
fine fundamental solutions for more general linear differential operators L.
A function u(x) = u(Xl, .•. ,xn) is called a fundamental solution for L with
pole ~ = (~l""'~n)' if it satisfies the symbolic equation LU(x) = o(x;~),
or equivalently the symbolic identity f vex) Lu(x) dx = v(~) for arbitrary
functions v. Here the lefthand side stands for the concrete eX?ression ob-
tained by remOving all derivatives of u from the integrand by formal re-
peated integration by parts. We can avoid all boundary contributions arising
from the integration by parts by restricting ourselves to arbitrary v that
Problems 235
vanish identically near the boundary of the region of integration (that is to
v "of cc:mpact support"). Then the fundamental solution u is characterized
by the identity
v(~) J u(x) LV(x) dx,
valid for all v of compact support, where Lv is the differential
expression obtained by the integration of parts (1 is the operator adjoint
to L in the sense of p. 187). The fundamental solution with pole ~ is not
unique since we can always add any "regular" solution w(x) of LW(x) = ° to u.
Find fundamental solutions with pole ~ for the following differential
operators L :
a) Lu = .6u in n dimensions
b) Lu = 6u + 2
A. u in 3-dimensions
c) Lu = 2 6u in two dimensions.
[Answers: From problem 3) section 5 with r = lx-~l denoting the distance
of the points x and ~
a)
b)
c)
cos Ar u=~
1 2 u = em- r log r 1.
7) a) Prove that the function u(x,y) defined by
for x > 0, y > ° u(x,y)
for all other (x,y)
is a fundamental solution for ?l L = dXaY with pole at the origin.
[ Hint: v(o,o) = J
° bounded set 1.
00 00
J v (x,y) dxdy for all v vanishing outside a ° xy
PROBLEMS
b) Prove that the function u(x,y) defined by
u(x,y) for y> I xl
for all other (x,y)
?l '02 L - - - - with pole at the origin. - 'Oi 'Oi is a fundamental solution for
[Hint: 1 v(O,O) = I 2 f Lv(x,y) dxdy for all v vanishing outside
y> Ixl
some bounded set ].
(Notice that the fundamental solutions for the elliptic operators L in
problem 6) all are singular only at the pole ~ itself, while those for
the hyperbolic operators in the present problem are discontinuous along
whole lines.)
Section 8
1) Let D be an open set in 3-space with (sufficiently regular) boundary B.
Let w(x) = w(xl ,x2,x3) be of class Cl in D and cO in D+B. Prove
Poisson's equation
w(x) - lj! 1 6 f If ~ dy for x in D, 'IT" DIy-xI
where again I y-xl denotes the distance of the points x and y, and dy the
element of volume.
[Hint: Proceed as in two dimensions ].
2) The gravitational attraction exerted on a unit mass located at the point
x = (xl ,x2 ,x3) by a solid D with density ~ = ~(x) is, according to
Newton's law, given by the vector
F(x) r f f f ~(y) (y-x) dy D I y-xl 3
(r = universal gravitational constant).
a) Prove that the 3 components Fl ,F2,F3 of the force F(x) have the form
F. (x) = 'o~(x) 1. X.
i 1,2,3 1.
Problems 237
where u(x), the "gravitational potential" of' D, is given by
u(x) y I JJ \-I(y) dy Diy-xi
b) Prove that the attraction F(x) exerted by D on a f'ar away unit mass
is approximately the same, as if' the total mass M of' D were con
centrated at its center of' gravity yO.
[Hint: By def'inition
M = If I \-I(y) dy, D
My0 = fff \-I(y) y dy. D
Since f'or large I xl and bounded y
-3 I y-xl
we have
I °12 ° ° (x-y - 2(x-y ). (y-y )
I °1-3 (I 0-4 x-y + ° x-y) )
I °12)-3/2 + y-y
c) Calculate the potential u and attraction F of' a solid sphere D of'
radius a with center at the origin and of' constant density \-I. Use
here that u must have spherical symmetry, must be harmonic outside D,
satisf'y Poisson's equation in D, be of' class 1 C everywhere, be regu-
lar at the origin, and vanish at infinity.
[Hint: Use problem 3), section 3, to show that u must be of the form
2 2 u(x) = A - "3 7fY \-II xl f'or I xl < a
u(x) = R for I xl > a
with suitable constants A,B. Hence from the remaining conditions
{
222 27f'Y).La -"3 7f'Y).L1 xl
u(x) = 3 4 a - 7f'Y).L r::r 3 I XI
for I xl < a
for I xl > a
238 PROBLEMS
f ~ rrwx for Ixi < a
F(x) 4 3 x for Ixi >a (compare with b) ~ J - "3 7TrJ.La j:lT
3) Let u(x,y) be of class c2 in the open bounded set D with boundary B
in the xy-plane and of class CO in D+B. Let u be a solution of the
equation
u + u + 2a(x,y)u + 2b(x,y)u + c(x,y)u = ° xx yy x y
for (x,y) in D, where the coefficient c(x,y) is negative throughout D.
Prove that if u = ° on B then u = ° in D.
[ Hint: Show that Max u SO, Min u ~ OJ.
Section 9
1) Let Lu = u - c26u = ° be the wave equation for 3 space dimensions. tt
a) Prove that the equation is invariant under reflection with respect to
the plane xl = 0, i.e., on replacing xl bu -Xl.
b) Show that if u(xl ,x2,x3,t) = u(x,t) is a solution of Lu = ° for all
x and for t > 0, with vanishing initial data for t = 0, xl < 0, then
is a solution of Lv = ° with the same initial data as f for
Xl > 0, and moreover satisfying the boundary condition v = ° for
xl = 0, t > 0.
c) For the solution v of the following boundary initial value problem
Lv = ° for
v = ° for
v = 0 for
for v = t
for
xl > 0, t > ° xl = 0, t > ° xl> 0, t = °
2 2 2 1 (Xl-I) +x2 +x3 <4'
2 2 2 1 (Xl-I) +x2 +x3 > 4' t = ° ,
Problems 239
find v for xl = 0, t > O. xl
(Answer: Use for u the solution of problem (4d), section 3 with
Section 10
to find that
5 2 222 4 + x2 + x3 - c t
2 2 3/2 2C(1+x2 +x3 )
1 a="2
for 1 2 221 2 ("2 - ct) < 1 + x2 + X3 < ("2 + ct) and
x2 ,x3' t ].
o for other
1) Let f(x) = f(xl, ..• ,xn) be continuous for all x and bounded uniformly.
Denote by K(x,t) = K(xl, ... ,xn,t) the function
K(x,t) = (4~t)-n/2 e-\xl 2/4t .
Prove that
u(x,t) = J K(x-s,t)f(s) ds
(the integration being extended over the whole n-dimensional s-space) is a
solution of the n-dimensional heat equation
Lu = ut - 6u = 0
00
of class C for t > 0, which is continuous for t ~ 0, and has the initial
values u(x,O) = f(x). (The case n = 1 is represented by formula (19),
p. 1'74).
(Hint: Proceed as in the case n = 1, using that
LK(x,t) = 0 for t > 0
lim K(x, t) = 0 uniformly for I xl > 5 for any 5 > 0 t ~O
K(x,t) > 0 for t > 0
240 PROBLEMS
f K(x,t) dx = 1 ] .
2) Consider the n-dimensional heat equation Lu = ut - 6u = o.
a) Let D be an open bounded set in x-space with boundary B, and let
u(x,t) be a solution of class c2 of Lu = 0 for x in D and
0< t ~ T, which is continuous for x in D+B and 0 ~ t~ T. Prove
that u assumes its maximum at some point (x,t) for which either x
on B or t = O.
[Hint: Compare problem 6), section 5. The maximum property follows for
functions v with Lv < 0 from the fact that at a maximum point
(x,t) with x in D and 0 < t < T we would have to have v t ~ 0,
2 I:N ~ o. Then take v = u + el xl , and let e -+ 0 ) •
b) If u(x,t) is a solution of Lu = 0 for t > 0, and is continuous and
bounded uniformly for all x and all t ~ 0, then u(x,t) never ex
ceeds the least upper bound of its initial values u(x,O).
[Hint: Fur u(x,t) ~ M for all x and t ~ 0, and u(x,O) < F for all
x, consider for any positive a,e,T the expression
2 U(x,t) = u(x,t) + e(2nt-lx\ ).
Then LU = 0 and for all sufficiently large a we have U(x,O) ~ F,
and also U(x,t) < F for I xl = a, 0 ~ t ~ T. It follows that
U(x, t) ~ F for all x and t ~ O. Let e -+ 0 ).
c) Show that the solution of the initial value problem constructed in
Problem 1) for bounded and continuous f is the only bounded solution.
3) a) Show that the bounded solution of the l-dimensional heat equation
ut = Uxx with initial data
for x > 0 u(x,O)
for x < 0
is given by
Problems 241
1 ( x u(x,t) ="2 1 + ~(- )), :f4t
where ~(s) is the "error function"
2 s _t2 ~(s) J e dt -.r; 0
[Hint: Use (19), p. 174 J.
b) Find the solution v(x,t) of the heat equation vt Vxx for x> 0,