Stony Brook University MAT 341 Fall 2011 Homework Solutions, Chapter 4, part 1.

4.2 # 6 Solve the potential equation in the rectangle 0 < x < a, 0 < y < b, subject to the boundary conditions u(a, y ) = 1, 0 < y < b, and u = 0 on the rest of the boundary. SOLUTION: The boundary conditions are homogeneous with respect to y so when we separate variables u(x, y ) = X (x)Y (y ), X Y Y X + = 0, = =c X Y Y X (with boundary conditions Y (0) = Y (b) = 0 and X (0) = 0) the constant c should be negative, i.e. 2 . In that case the general solution for Y is Y (y ) = a cos(x) + b sin(x). 2 u = X Y + XY = 0, The boundary conditions on Y force rst a = 0 and then = n-th eigenfunction is sin ny . b
n : b

the

)2 X with general soThe n-th corresponding X -equation is X = ( n b lution n n Xn (x) = An cosh x + Bn sinh x. b b The boundary condition X (0) = 0 forces An = 0. Since these conditions are homogeneous, and the equation is linear, we can use superposition to write

u(x, y ) =
1

Bn sinh

nx ny sin b b

and use the boundary condition u(a, y ) = 1 to determine the coecients {Bn }. In fact

u(a, y ) =
1 na b

Bn sinh

ny na sin =1 b b

means that Bn sinh is the n-th Fourier sine coecient of the function equal to 1 on [0, b], i.e. the n-th Fourier coecient of the square wave: 4 if n is odd, 0 if n is even. So if n is even, then n Bn = 1 4 n sinh na b 1

and Bn = 0 if n is odd. 4.3 # 1 Solve the problem consisting of the potential equation on the rectangle 0 < x < a, 0 < y < b with the given boundary conditions. Two of the three are very easy if a polynomial is subtracted from u. (a).
u (0, y ) x

= 0; u = 1 on the rest of the boundary.

SOLUTION. Follow hint, set v (x, y ) = u(x, y ) 1. Then 2 v = 2 u = v 0 and the boundary conditions become x (0, y ) = 0; u = 0 on the rest of the boundary. The solution to this problem is clearly the constant function v = 0, so the solution to the given problem is the constant function u(x, y ) = 1. (b).
u (0, y ) x

= 0;

u (a, y ) x

= 0; u(x, 0) = 0; u(x, b) = 1.

. Then Then 2 v = SOLUTION. Follow hint, set v (x, y ) = u(x, y ) y b v 2 u = 0 and the boundary conditions become x (0, y ) = 0; u (a, y ) = x 0; u(x, 0) = 0; u(x, b) = 0. The solution to this problem is clearly the constant function v = 0, so the solution to the given problem is the constant function u(x, y ) = y . b (c).
u (x, 0) x

= 0; u(x, b) = 0; u(0, y ) = 1; u(a, y ) = 0.

SOLUTION. The condition u (x, 0) = 0 is equivalent to u(x, 0) = C , x a constant. Now the problem splits into two problems: u = u1 + u2 where 2 u1 = 0, u1 (0, y ) = u1 (a, y ) = 0, u1 (x, 0) = C, u1 (x, b) = 0 2 u2 = 0, u2 (x, 0) = u1 (x, b) = 0, u2 (0, y ) = 1, u2 (a, y ) = 0. These each can be solved by the method of 4.2 # 6. To simplify the calculations switch to v1 (x, y ) = u1 (x, b y ) and v2 (x, y ) = u2 (a x, y ) and then switch back. 4.4 # 4 Solve the potential problem in the slot 0 < x < a, 0 < y , for each of these sets of boundary conditions. (a.) u(0, y ) = 0, u(a, y ) = 0, 0 < y ; u(x, 0) = 1, 0 < x < a.

SOLUTION: This is a u1 -type problem (homogeneous x-boundary conditions) as on p. 279. So when we separate to get Y X = =c X Y as usual. we should take c = 2 ; the X -eigenfunctions are then sin nx a ny ny The Yn equation then has general solution An e a + Bn e a . The requirement that solutions be bounded as y forces An = 0. Note how the choice of exponential solutions rather than hyperbolictrigonometric simplies the calculation. The other choice is also legitimate, but will involve more work. The general solution is then

u(x, y ) =
0

Bn sin

nx ny e a . a

The coecients Bn are determined by the boundary condition u(x, 0) = 1 , 0 < x < a: nx Bn sin = 1. a 0 This is the sine series for the square wave: Bn = is even.
4 n

if n is odd, 0 if n

(b.) u(0, y ) = 0, u(a, y ) = ey , 0 < y ; u(x, 0) = 0, 0 < x < a. SOLUTION. Here the boundary condition in y is homogeneous, so we separate as X Y = =c X Y with c positive, e.g. c = 2 , and we solve for Y rst. The general Y solution is Y = a cos y + b sin y . The boundary condition u(x, 0) = 0 forces a = 0. The corresponding X -equation is X = 2 X with general solution X = A() cosh x + B () sinh x, and the general u(x, y ) solution is

u(x, y ) =

(A() cosh x + B () sinh x) sin y d.

The coecient functions A() and B () are determined by the boundary conditions u(0, y ) = 0, u(a, y ) = ey :

u(0, y ) =

A() sin y d = 0 3

means that A() corresponds to the Fourier sine integral of the zero function, so by uniqueness A() 0. Then

u(a, y ) =

B () sinh a sin y d = ey

means that B () sinh a is the Fourier sine integral of ey so 2 2 y e sin y dy = 0 1 + 2 (integrate by parts or use table of integrals) and B () sinh a = B ( ) = 1 2 . 1 + 2 sinh a

(c.) Similar to (b.), with an easier integral. Make life simpler by setting v (x, y ) = u(a x, y ) so v (0, y ) = 0 and v (a, y ) = f (y ); switch back to u to nish. 4.4 # 5 Solve the potential problem in the slot 0 < x < a, 0 < y , for each of these sets of boundary conditions. (a.)
u (0, y ) x

= 0, u(a, y ) = 0, 0 < y ; u(x, 0) = 1, 0 < x < a.

SOLUTION. In this case the x-problem is homogeneous, so we separate as Y X = = 2 X Y and solve for X rst. The general solution is X (x) = a cos x + b sin x; the boundary conditions translate to X (0) = 0, X (a) = 0. The rst 1) , so the n-th X -eigenfunction forces b = 0; the second forces = (2n2 a (2n1)x is cos 2a . The corresponding Yn equation is
Yn =(

(2n 1) 2 ) Yn 2a

1)y 1)y ) + bn exp( (2n ). Note choice with solution Yn = an exp( (2n 2a 2a of basis for solutions. The requirement that solutions be bounded as y forces an = 0. The general solution is then

u(x, y ) =
0

(2n 1)y (2n 1)x 2a cos bn e 2a 4

where the {bn } are determined by the boundary condition

u(x, 0) =
0

bn cos

(2n 1)x = 1. 2a

1)x The functions cos (2n for n = 1, 2, 3, ... are an orthogonal family 2a on [0, a] with a a (2n 1)x dx = , cos2 2a 2 0 so

bn =

2 a

a 0

cos

2 (2n 1)x a 2a (2n 1)x dx = sin |0 2a a (2n 1) 2a

4 n = 1, 3, 5, ... (2n 1) 2a 2 (2 n 1) sin = . = 4 a (2n 1) 2 n = 2, 4, 6, ... (2n 1) (0, y ) = 0, u(a, y ) = ey , 0 < y ; u(x, 0) = 0, 0 < x < a. Here the (b.) u x y -boundary condition is homogeneous, so we separate as X Y = = 2 X Y and solve for Y rst. The general solution is Y (y ) = a cos y + b sin y ; the boundary condition translates to Y (0) = 0, which forces a = 0. The X equation is then X = 2 , with general solution X = A() cosh x + X B () sinh x. The general solution for u is

u(x, y ) =

(A() cosh x + B () sinh x) sin y d,

where the coecient functions are determined by the boundary conditions using Fourier integrals. Namely:
u (0, y ) = (B () sin y d = 0 x 0 means that B () is the Fourier sine integral for the zero function; by uniqueness B () 0 so B () 0. Then

u(a, y ) =

A() cosh a sin y d = ey

means that A() cosh a is the Fourier sine integral for ey , i.e. A() cosh a = 2
0

ey sin y dy =

2 1 + 2

(the same integral as in 4.b.) so A() = 1 2 . 2 1 + cosh a 1, 0 < y < b , 0 b<y

(c.) u(0, y ) = 0, u(a, y ) = f (y ) = x < a.

u (x, 0) y

= 0, 0 <

SOLUTION: Here the y -conditions are homogeneous, so we separate = 2 , X = 2 , and solve for Y rst. The general solution and set Y Y X is Y = a cos y + b sin y ; the boundary condition at y = 0 translates to Y (0) = 0, which forces b = 0, with no condition on . The corresponding X equation has general solution X (x) = A() cosh x + B () sinh x, leading to

u(x, y ) =

(A() cosh x + B () sinh x) cos y d

where A() and B () are determined by the boundary conditions:

u(0, y ) =

A() cos y d = 0

gives A() as the cosine integral of the zero function, so A() 0. Then u(a, y ) = B () sinh a cos y d = f (y )
0

gives B () sinh a = and so B ( ) = 2

f (y ) cos y dy =

b 0

cos y dy =

2 sin b

2 sin b . sinh a

4.5 # 1 Solve the potential equation in the disc 0 < r < c if the boundary condition is v (c, ) = ||, < < pi. SOLUTION. As described in 4.5, the potential equation 2 v = 0 leads, via writing v (r, ) = R(r)Q(), to Q = 2 Q with general solution Q() = a cos + b sin ; since Q must be periodic of period 2 for the function to be well-dened on the disc, must be an integer: n = 0, 1, 2, 3, ... (negative integers dont give new solutions). The corresponding Rn must satisfy the Cauchy-Euler equation; the solutions are Rn (r) = rn , Rn (r) = rn . The second solution blows up at r = 0 and is not useful. The solution to the problem is then

v (r, ) = a0 +
1

(an cos n + bn sin n)rn ,

where the coecients an , bn are determined from the initial conditions by Fourier analysis:

v (c, ) = a0 +
1

(an cos n + bn sin n)cn = ||.

So a0 , an cn and bn cn are the coecients of the Fourier series of f () = ||, < < . This f is an even function, so the sine coecients are zero, and 1 a0 = d = 0 2 4 n odd 2 an c n = cos n d = n2 0 0 n even (note that || = on [0, ]). Finally v (r, ) = 4 2 rn cos n . n2 c n

odd

4.5 #4 Same as Exercise 1 with boundary condition v (c, ) = f () = 1 < < 0 . 1 0<<

SOLUTION. Same as Exercise 1, except here f is odd, so the cosine coecients are zero, and bn c n = So v (r, ) = 2
0

sin n d =

2 cos n| 0 = n

4 n n odd . 0 n even

odd

1 n r sin n. ncn

4.5 #5 Find the value of the solution at r = 0 for the problems of Exercises 1 and 4. , and SOLUTION. When r = 0 the solution of Exercise 1 gives v = 2 the solution of Exercise 4 gives v = 0.