HOMEWORK 4 SOLUTIONS

All questions are from Vector Calculus, by Marsden and Tromba Question 1: 3.1.16 Let w = f (x, y ) be a function of two variables, and let x = u + v, y = u v. Show that 2w 2w 2w = . uv x2 y 2 Solution. By the chain rule, w w x w y = + = wx wy . v x v y v Thus, 2w uv = = = i.e., 2w 2w 2w = . uv x2 y 2 Question 2: 3.1.22 (a) : Show that the function g (x, t) = 2 + et sin x satises the heat equation: gt = gxx . [Here g (x, t) represents the temperature in a metal rod at position x and time t.] (b) : Sketch the graph of g for t 0. (Hint: Look at sections by the planes t = 0, t = 1, and t = 2.) (c) : What happens to g (x, t) as t ? Interpret this limit in terms of the behavior of heat in the rod. Solution. (a) : Since g (x, y ) = 2 + et sin x, then gt = et sin x, gx = et cos x, and gxx = et sin x. Therefore, gt = gxx . (b) : The graph of g is shown in Figure 1.
Date : Math 1c Practical, 2008.
1

w = (wx wy ) = wx wy u v u u u wy x wy y wx x wx y + + x u y u x u y u wxx + wxy (wyx + wyy ) = wxx wyy

HOMEWORK 4 SOLUTIONS

t=0 4 t=1 t=2 3

2 4 1 2 0 x 0 0 1 2 t 3 4 4 2

Figure 1. The graph of g at t = 0, 1, and 2. (c) : Note that

lim g (x, t) = lim (2 + et sin x) = 2

This means that the temperature in the rod at position x tends to be a constant (= 2) as the time t is large enough. Question 3: 3.2.2 Determine the second-order Taylor formula for f (x, y ) = x2 1 + y2 + 1 about x0 = 0, y0 = 0.

Solution. We rst compute the partial derivatives up through second order: fx fxy fxx fyy = = = = 2x , (1 + x2 + y 2 )2 8xy , (1 + x2 + y 2 )3 2 + (1 + x2 + y 2 )2 2 + (1 + x2 + y 2 )2 2y (1 + x2 + y 2 )2 8xy fyx = (1 + x2 + y 2 )3 8x2 (1 + x2 + y 2 )3 8y 2 . (1 + x2 + y 2 )3 fy =

HOMEWORK 4 SOLUTIONS

Next, we evaluate these derivatives at (0, 0), obtaining fx (0, 0) = fy (0, 0) = 0, fxy (0, 0) = fyx (0, 0) = 0 and fxx (0, 0) = fyy (0, 0) = 2. Therefore, the second order Taylor formula is
2 f (h) = h2 1 h2 + R2 (0, h),

where h = (h1 , h2 ) and where R2 (0, h) 0 h as h 0.

Question 4: 3.2.6 Determine the second-order Taylor formula for the function f (x, y ) = e(x1) cos y expanded about the point x0 = 1, y0 = 0. Solution. The ingredients needed in the second-order Taylor formula are computed as follows: fx fy fxx fxy fyy = 2(x 1)e(x1) cos y
2 2 2

= e(x1) sin y = 2e(x1) cos y + 4(x 1)2 e(x1) cos y

= 2(x 1)e(x1) sin y = fyx = e(x1) cos y.

Evaluating the function and these derivatives at the point (1, 0) gives f (1, 0) fx (1, 0) fxx (1, 0) fxy (1, 0) fyy (1, 0) = = = = = 1 fy (1, 0) = 0 2 fyx (1, 0) = 0 1. and

Consequently, the second order Taylor formula is 1 2 f (h) = 1 + h2 1 h2 + R2 ((1, 0), h), 2 where h = (h1 , h2 ) and where R2 ((1, 0), h) 0 h as h 0.

Question 5: 3.3.7 Find the critical points for the function f (x, y ) = 3x2 + 2xy + 2x + y 2 + y + 4. and then determine whether they are local maxima, local minima, or saddle points.

HOMEWORK 4 SOLUTIONS

Solution. Here, f = 6x + 2y + 2, x We have f f = 0, =0 x y when x = y = 1/4. Therefore, the only critical point is (1/4, 1/4). Now, 2f 2f 2f x2 (1/4, 1/4) = 6, y 2 (1/4, 1/4) = 2, and xy (1/4, 1/4) = 2, which yields D = 6.2 2 = 10 > 0. Therefore (1/4, 1/4) is a local minimum.
2

f = 2x + 2y + 1. y

Question 6: 3.3.17 Find the local maxima and minima for z = (x2 +3y 2 )e1x
2 2

y 2

(2e)(6e) = 12e > 0, and (0, 0) is a local minimum. 2f 2f 2f x2 (0, 1) = 4, y 2 (0, 1) = 12, and xy (0, 1) = 0, which yields D = (4)(12) = 24 > 0, and (0, 1) are local maxima. 2f 2f 2f x2 (1, 0) = 4, y 2 (1, 0) = 4, and xy (0, 1) = 0, which yields D = (4)(4) = 16 < 0, and (1, 0) are saddle points. Question 7: 3.3.25 Write the number 120 as a sum of three numbers so that the sum of the products taken two at a time is a maximum. Solution. Let the three numbers be x, y, z . Thus, x + y + z = 120, We want to nd the maximum value for S (x, y ) We dierentiate to get S = 2x y + 120, x S = x 2y + 120. y = xy + yz + xz = xy + (x + y )(120 x y ) = x2 xy y 2 + 120x + 120y. z = 120 x y.

2f 2 4 2 2 1x2 y 2 , and y 2 = (3 15y + 6y + x (2y 1))e 2f 2 2 1x2 y 2 . xy = 4(3y + x 4)e 2f 2f 2f (0, 0) Therefore, x2 (0, 0) = 2e, y2 (0, 0) = 6e, and xy 2

Solution. We rst locate the critical points of f (x, y ) = (x2 + 3y 2 )e1x y . 2 2 f (x, y ) = e1x y (2x(1 3y 2 x2 )i + 2y (3 x2 3y 2 )j) Thus, f (x, y ) = 0 if and only if (x, y ) = (0, 0), (0, 1), or (1, 0). To determine whether they are maxima or minima, we need to calculate the second partial derivatives. 2f 4 2 2 2 1x2 y 2 x2 = (1 + 2x 3y + x (6y 5))e

= 0, which yields D =

These vanish when x = y = 40, then z = 120 (x + y ) = 40. Therefore, when x = y = z = 40 is the only critical point. The condition 0 x 120, 0 y 120, 0 z 120 describes a cube in R3 and on the boundary of the cube (either x = 0, x = 120, y = 0, y = 120, z = 0, z = 120), S is zero. Therefore the maximum of S occurs on the interior of this cube, i.e., at a local maximum. Since x = 40, y = 40, z = 40 is the only critical point, it must be a maximum.

HOMEWORK 4 SOLUTIONS

Question 8: 3.4.2 Find the extrema of f (x, y ) = x y subject to the constraint x2 y 2 = 2. Solution. By the method of Lagrange multipliers, we write the constraint as g = 0, where g (x, y ) = x2 y 2 2 and then write the Lagrange multiplier equations as f = g . Thus, we get 1 1 x y 2
2 2

= = =

2x 2y 0.

First of all, the rst two equations imply that x = 0 and y = 0. Hence we can eliminate , giving x = y . From the last equation this would imply that 2 = 0. Hence there are no extrema. Question 9: 3.4.22 Let P be a point on a surface S in R3 dened by the equation f (x, y, z ) = 1, where f is of class C 1 . Suppose that P is a point where the distance from the origin to S is maximized. Show that the vector emanating from the origin and ending at P is perpendicular to S . Solution. We want to maximize the function g (x, y, z ) = x2 + y 2 + z 2 subject to the constraint f (x, y, z ) = 1. Suppose this maximum occurs at P = (x0 , y0 , z0 ), then by the method of Lagrange multipliers we have the equations 2x0 2y0 2z0 = {f (x0 , y0 , z0 )}1 = {f (x0 , y0 , z0 )}2 = {f (x0 , y0 , z0 )}3

where {f (x0 , y0 , z0 )}i denotes the ith component of f (x0 , y0 , z0 ), 1 i 3. If v = (x0 , y0 , z0 ) is the vector from the origin ending at P, then these equations say that v = 2 f (x0 , y0 , z0 ). But f (x0 , y0 , z0 ) is perpendicular to S at P, and since v is a scalar multiple of f (x0 , y0 , z0 ) it is also perpendicular to S at P. Question 10: 3.4.28 A companys production function is Q(x, y ) = xy . The cost of production is C (x, y ) = 2x + 3y . If this company can spend C (x, y ) = 10, what is the maximum quantity that can be produced? Solution. We want to maximize Q subject to the constraint C (x, y ) = 10. Since both x, y 0, this imposes the condition that 0 x 5, 0 y 10/3. Thus, we wish to maximize Q on the line segment 2x +3y = 10, x 0, y 0. If the maximum occurs at an interior point (x0 , y0 ) of this segment, then Q(x0 , y0 ) = C (x0 , y0 ); that is, y0 x0 2x0 + 3y0 = = = 2 3 10.

Thus 6 + 6 = 10, = 5/6, y0 = 5/3, x0 = 5/2, Q(x0 , y0 ) = 25/6. The value of Q at the endpoints of this segment are Q(0, 10 3 ) = 0 = Q(5, 0). Consequently the maximum occurs at (5/2, 5/3) and the maximum value of Q is 25/6.