Anda di halaman 1dari 16

Section 13.

8
C13S08.001: If f (x, y) = 3x − 7y and P (17, 39) are given, then

∇f (x, y) =  3, −7 , and thus ∇f (17, 39) =  3, −7 .

C13S08.002: If f (x, y) = 3x2 − 5y 2 and P (2, −3) are given, then

∇f (x, y) =  6x, −10y , and thus ∇f (2, −3) =  12, 30 .

C13S08.003: If f (x, y) = exp(−x2 − y 2 ) and P (0, 0) are given, then

∇f (x, y) =  −2x exp(−x2 − y 2 ), −2y exp(−x2 − y 2 ) , and thus ∇f (0, 0) =  0, 0  = 0.

C13S08.004: If f (x, y) = sin 14 πxy and P (3, −1) are given, then

1   √ √ 
∇f (x, y) = 1
4 πy cos 4 πxy,
1 1
4 πx cos 4 πxy , and so ∇f (3, −1) = 1
8π 2 , − 38 π 2 .

C13S08.005: Given f (x, y, z) = y 2 − z 2 and P (17, 3, 2), then

∇f (x, y, z) =  0, 2y, −2z , and therefore ∇f (17, 3, 2) =  0, 6, −4 .


C13S08.006: Given f (x, y, z) = x2 + y 2 + z 2 and P (12, 3, 4), then
 
x y z
∇f (x, y, z) =  ,  ,  ,
x2 + y 2 + z 2 x2 + y 2 + z 2 x2 + y 2 + z 2
 12 
and therefore ∇f (12, 3, 4) = 3 4
13 , 13 , 13 .

C13S08.007: Given f (x, y, z) = ex sin y + ey sin z + ez sin x and P (0, 0, 0), then

∇f (x, y, z) = ez cos x + ex sin y, ex cos y + ey sin z, ey cos z + ez sin x ,

and therefore ∇f (0, 0, 0) =  1, 1, 1 .

C13S08.008: Given f (x, y, z) = x2 − 3yz + z 3 and P (2, 1, 0), then

∇f (x, y, z) =  2x, −3z, 3z 2 − 3y , and so ∇f (2, 1, 0) =  4, 0, −3 .


C13S08.009: Given f (x, y, z) = 2 xyz and P (3, −4, −3), then

yz xz xy 3
∇f (x, y, z) = √ , √ , √ , and so ∇f (3, −4, −3) = 2, − , −2 .
xyz xyz xyz 2

C13S08.010: Given f (x, y, z) = (2x − 3y + 5z)5 and P (−5, 1, 3), then

∇f (x, y, z) =  10(2x − 3y + 5z)4 , −15(2x − 3y + 5z)4 , 25(2x − 3y + 5z)4 ,

1
and therefore ∇f (−5, 1, 3) =  160, −240, 400 .

C13S08.011: Given f (x, y) = x2 + 2xy + 3y 2 , P (2, 1), and v =  1, 1 , we first compute a unit vector
with the same direction as v:
v  √ √ 
u= = 12 2 , 12 2 .
|v|

Also ∇f (x, y) =  2x + 2y, 2x + 6y , so ∇f (P ) =  6, 10 . Therefore


 √ √  √
Du f (P ) = (∇f (P )) · u =  6, 10  · 12 2 , 12 2 = 8 2 .


C13S08.012: Given f (x, y) = ex sin y, P 0, 1
4π , and v =  1, −1 , we first compute a unit vector with
the same direction as v:
v  √ √ 
u= = 12 2 , − 12 2 .
|v|
 √ √ 
Also ∇f (x, y) =  ex sin y, ex cos y , so ∇f (P ) = 12 2 , 12 2 . Therefore Du f (P ) = (∇f (P )) · u = 0.

C13S08.013: Given f (x, y) = x3 − x2 y + xy 2 + y 3 , P (1, −1), and v =  2, 3 , we first compute a unit


vector with the same direction as v:
v  √ √ 
2 3
u= = 13 13 , 13 13 .
|v|

Also ∇f (x, y) =  3x2 − 2xy + y 2 , 3y 2 + 2xy − x2 , so ∇f (P ) =  6, 0 . Therefore


 √ √  12 √
Du f (P ) = (∇f (P )) · u =  6, 0  · 13
2 3
13 , 13 13 = 13 13 .

C13S08.014: Given f (x, y) = arctan(y/x), P (−3, 3), and v =  3, 4 , we can automate the computation
of Du f (P ) using Mathematica 3.0 as follows. First we find the unit vector u with the same direction as v
(remember that v.v is the way to compute v · v):

v = {3, 4};

u = v/Sqrt[v.v]

3 4
,
5 5

Then we define f and compute its gradient:

f[x , y ] := ArcTan[y/x]

{D[f[x,y], x], D[f[x,y], y]} // Simplify



y x
− ,
x2 + y 2 x2 + y 2

Then we evaluate ∇f (P ) (recall that % refers to the “last output”):

% /. {x → -3, y → 3}

2

1 1
− ,−
6 6

Now we can compute Du f (P ) = (∇f (P )) · u:

%.u

7

30

C13S08.015: Given: f (x, y) = sin x cos y, the point P 13 π, − 23 π , and the vector v =  4, −3 , we first
construct a unit vector with the same direction as v:

v 4 3
u= = ,− .
|v| 5 5

Next, ∇f (x, y) =  cos x cos y, − sin x sin y , and hence ∇f (P ) =  − 14 , 3


4 . Therefore

13
Du f (P ) = (∇f (P )) · u = − .
20

C13S08.016: Given f (x, y, z) = xy + yz + zx, the point P (1, −1, 2), and the vector v =  1, 1, 1 , we
first construct a unit vector with the same direction as v:
√ √ √ 
v 3 3 3
u= = , , .
|v| 3 3 3

Next, ∇f (x, y, z) =  y + z, x + z, x + y , and thus ∇f (P ) =  1, 3, 0 . Therefore



4 3
Du f (P ) = ∇f (P ) · u = .
3


C13S08.017: Given f (x, y, z) = xyz , the point P (2, −1, −2), and the vector v =  1, 2, −2 , we first
construct a unit vector u with the same direction as v:

v 1 2 2
u= = , ,− .
|v| 3 3 3

Next,
 
yz xz xy
∇f (x, y, z) =  ,  ,  ,
2 xyz 2 xyz 2 xyz

1 1 1 2 2 1
and hence Du f (P ) = ∇f (P ) · u = , −1, − · , ,− =− .
2 2 3 3 3 6

C13S08.018: We are given f (x, y, z) = ln(1 + x2 + y 2 − z 2 ), the point P (1, −1, 1), and the vector
v =  2, −2, −3 . The first step is to construct a unit vector u with the same direction as v:

v 2 2 3
u= = √ , −√ , −√ .
|v| 17 17 17
Next,

2x 2y 2z
∇f (x, y, z) = , ,− ,
1+x +y −z 1+x +y −z
2 2 2 2 2 2 1 + x + y2 − z2
2

and thus ∇f (P ) =  1, −1, −1 . Therefore


7
Du f (P ) = ∇f (P ) · u = √ .
17

C13S08.019: Given f (x, y, z) = exp(xyz), the point P (4, 0, −3), and the vector v =  0, 1, −1 , we first
construct a unit vector with the same direction as v:
 √ √ 
v 2 2
u= = 0, ,− .
|v| 2 2

Next, ∇f (x, y, z) = exp(xyz) yz, xz, xy , and so ∇f (P ) =  0, −12, 0 . Therefore



Du f (P ) = ∇f (P ) · u = −6 2 .


C13S08.020: Given f (x, y, z) = 10 − x2 − y 2 − z 2 , the point P (1, 1, −2), and the vector v =
 3, 4, −12 , we begin by constructing the unit vector u with the same direction as v:

v 3 4 12
u= = , ,− .
|v| 13 13 13

Then
 
x y z
∇f (x, y, z) = − , − , − .
10 − x2 − y 2 − z 2 10 − x2 − y 2 − z 2 10 − x2 − y 2 − z 2

Therefore

1 1 3 4 12 31
Du f (P ) = ∇f (P ) · u = − ,− ,1 · , ,− =− .
2 2 13 13 13 26

C13S08.021: Given f (x, y) = 2x2 + 3xy + 4y 2 and the point P (1, 1), we first compute

∇f (x, y) =  4x + 3y, 3x + 8y .

So the direction in which f is increasing


√ the most rapidly at P is ∇f (P ) =  7, 11  and its rate of increase
in that direction is | 7, 11 | = 170 .
y
C13S08.022: Given f (x, y) = arctan and the point P (2, −3), we first compute
x

y x
∇f (x, y) = − 2 , .
x + y 2 x2 + y 2

So the direction in which f is increasing the most rapidly at P is


3 2
v = ∇f (P ) = , ,
13 13

4

13
and its rate of increase in that direction is |v| = .
13

C13S08.023: Given f (x, y) = ln(x2 + y 2 ) and the point P (3, 4), we first compute

2x 2y
∇f (x, y) = , .
x2 + y 2 x2 + y 2
Therefore the direction in which f is increasing the most rapidly at P is

6 8
v = ∇f (P ) = , ,
25 25
2
and its rate of increase in that direction is |v| = .
5
1
C13S08.024: Given f (x, y) = sin(3x − 4y) and the point P 1
3 π, 4 π , we first compute

∇f (x, y) =  3 cos(3x − 4y), −4 cos(3x − 4y) .

Therefore the direction in which f is increasing the most rapidly at P is v = ∇f (P ) =  3, −4  and its rate
of increase in that direction is |v| = 5.

C13S08.025: Given f (x, y, z) = 3x2 + y 2 + 4z 2 and the point P (1, 5, −2), we first compute

∇f (x, y, z) =  6x, 2y, 8z .

Therefore the direction in which f is increasing √ the most rapidly at P is v = ∇f (P ) =  6, 10, −16  and
its rate of increase in that direction is |v| = 14 2 .

C13S08.026: We are to find the direction in which f (x, y, z) = exp(x − y − z) is increasing the most
rapidly at the point P (5, 2, 3) and its rate of increase in that direction. Such computations can easily be
carried out with computer algebra systems such as Mathematica 3.0. We first define f :

f[x , y , z ] := Exp[x - y - z]

Then we compute the gradient of f :

{D[f[x,y,z],x], D[f[x,y,z],y], D[f[x,y,z],z]}


 
ex−y−z , −ex−y−z , −ex−y−z

Then we evaluate the last output at P :

% /. {x → 5, y → 2, z → 3}

 1, −1, −1 

The last output, ∇f (P ), gives the direction in which f is increasing the most rapidly at P . Its magnitude
is the rate of increase of f in that direction:

Sqrt[%.%]

3

5

C13S08.027: We are given f (x, y, z) = xy 2 z 3 and the point P (2, 2, 2). We first compute the gradient
of f :
 
y2 z3 xyz 3 3xy 2 z 2
∇f (x, y, z) =  ,  ,  .
2 xy 2 z 3 xy 2 z 3 2 xy 2 z 3

Thus the direction in which f is increasing√ the most rapidly at P is ∇f (P ) =  2, 4, 6  and its rate of
increase in that direction is |∇f (P )| = 2 14 .

C13S08.028: Given: f (x, y, z) = 2x + 4y + 6x and the point P (7, 5, 5). We first compute the gradient
of f :
 
1 2 3
∇f (x, y, z) =  ,  ,  .
2x + 4y + 6x 2x + 4y + 6x 2x + 4y + 6x

Hence the direction in which f is increasing the most rapidly at P is


1 1 3
v = ∇f (P ) = , ,
8 4 8

and its rate of increase in that direction is |v| = 1
8 14 .

C13S08.029: Let f (x, y) = exp(25 − x2 − y 2 ) − 1. Then

∇f (x, y) =  −2x exp(25 − x2 − y 2 ), −2y exp(25 − x2 − y 2 ) ,

so at P (3, 4) we have ∇f (P ) =  −6, −8 , a vector normal to the graph of f (x, y) = 0 at the point P .
Hence, as in Example 7, an equation of the line tangent to the graph at P is −6(x − 3) − 8(y − 4) = 0;
simplified, this is 3x + 4y = 25.

C13S08.030: Let f (x, y) = 2x2 + 3y 2 − 35. Then ∇f (x, y) =  4x, 6y . Thus a vector normal to the
graph of f (x, y) = 0 at the point P (2, 3) is ∇f (P ) =  8, 18 . Hence an equation of the line tangent to the
graph at P is 8(x − 2) + 18(y − 3) = 0; that is, 4x + 9y = 35.

C13S08.031: Let f (x, y) = x4 + xy + y 2 − 19. Then ∇f (x, y) =  4x3 + y, x + 2y , so a vector normal to


the graph of f (x, y) = 0 at the point P (2, −3) is ∇f (P ) =  29, −4 . So the tangent line at P has equation
29(x − 2) − 4(y + 3) = 0; that is, 29x − 4y = 70.

C13S08.032: Let f (x, y, z) = 3x2 + 4y 2 + 5z 2 − 73. Then ∇f (x, y, z) =  6x, 8y, 10z , so a vector normal
to the graph of f (x, y, z) = 0 at the point P (2, 2, 3) is ∇f (P ) =  12, 16, 30 . Therefore the plane tangent
to the graph at P has equation

12(x − 2) + 16(y − 2) + 30(z − 3) = 0; that is, 6x + 8y + 15z = 73.

C13S08.033: Let f (x, y, z) = x1/3 + y 1/3 + z 1/3 − 1. Then


1 1 1
∇f (x, y, z) = , , ,
3x2/3 3y 2/3 3z 2/3

and thus a vector normal to the surface f (x, y, z) = 0 at the point P (1, −1, 1) is

1 1 1
∇f (P ) = , , .
3 3 3

Therefore an equation of the plane tangent to the surface at P is

1 1 1
(x − 1) + (y + 1) + (z − 1) = 0; that is, x + y + z = 1.
3 3 3

C13S08.034: Let f (x, y, z) = xyz + x2 − 2y 2 + z 3 − 14. Then ∇f (x, y, z) =  yz + 2x, xz − 4y, xy + 3z 2 ,


so a vector normal to the surface f (x, y, z) = 0 at the point P (5, −2, 3) is ∇f (P ) =  4, 23, 17 . Therefore
the plane tangent to this surface at P has equation

4(x − 5) + 23(y + 2) + 17(z − 3) = 0; that is, 4x + 23y + 17z = 25.

C13S08.035: If u and v are differentiable functions of x and y and a and b are constants, then

∂ ∂
∇ au(x, y) + bv(x, y) = au(x, y) + bv(x, y) , au(x, y) + bv(x, y)
∂x ∂y

=  aux + bvx , auy + bvy  =  aux , auy  +  bvx , bvy 

= a  ux , uy  + b  vx , vy  = a∇u(x, y) + b∇v(x, y).

C13S08.036: If u and v are differentiable function of x and y, then


∂ ∂
∇ u(x, y) · v(x, y) = (uv), (uv)
∂x ∂y

=  ux v + uvx , uy v + uvy  =  ux v, uy v  +  uvx , uvy 

= v ·  ux , uy  + u ·  vx , vy  = u(x, y) · ∇v(x, y) + v(x, y) · ∇u(x, y).

C13S08.037: If u and v are differentiable functions of x and y and v(x, y)


= 0, then
 

u(x, y) ∂ u ∂ u vux − uvx vuy − uvy


∇ = , = ,
v(x, y) ∂x v ∂y v v2 v2
 vu vu   uv uv  v∇u u∇v v∇u − u∇v
x y x y
= 2
, 2 − 2
, 2 = 2 − 2 = .
v v v v v v v2

C13S08.038: Suppose that n is a positive integer and that u is a differentiable function of x and y. Then

∂ n ∂ n
∇(u ) =
n
(u ), (u ) =  nun−1 ux , nun−1 uy 
∂x ∂y

= nun−1 ·  ux , uy  = nun−1 · ∇u.

C13S08.039: We know that v = ∇f (P ) gives the direction in which f is increasing


the most rapidly at P .
Then v is the direction in which −f is decreasing the most rapidly at P . But ∇ − f (P ) = −∇f (P ) = −v,
so that −v is the direction in which −f is increasing the most rapidly at P and, therefore, is the direction
in which f is decreasing the most rapidly at P .

7
C13S08.040: Suppose that f is a differentiable function of the three independent variables x, y, and z.
Then


Di f (x, y, z) = ∇f (x, y, z) · i =  fx , fy , fz  ·  1, 0, 0  = fx (x, y, z),

Dj f (x, y, z) = ∇f (x, y, z) · j =  fx , fy , fz  ·  0, 1, 0  = fy (x, y, z), and

Dk f (x, y, z) = ∇f (x, y, z) · k =  fx , fy , fz  ·  0, 0, 1  = fz (x, y, z).

C13S08.041: Let f (x, y) = Ax2 + Bxy + Cy 2 − D. Then

∇f (x, y) =  2Ax + By, 2Cy + Bx ,

so a vector normal to the graph of f (x, y) = 0 at the point P (x0 , y0 ) is

∇f (P ) =  2Ax0 + By0 , 2Cy0 + Bx0 .

Hence, as in Example 7, an equation of the line tangent to the graph at P is

(2Ax0 + By0 )(x − x0 ) + (2Cy0 + Bx0 )(y − y0 ) = 0;

2Ax0 x + By0 x − 2A(x0 )2 − Bx0 y0 + 2Cy0 y + Bx0 y − 2C(y0 )2 − Bx0 y0 = 0;

2Ax0 x + By0 x + Bx0 y + 2Cy0 y = 2A(x0 )2 + 2Bx0 y0 + 2C(y0 )2 ;

2(Ax0 )x + B(y0 x + x0 y) + 2(Cy0 )y = 2D;


1
(Ax0 )x + B(y0 x + x0 y) + (Cy0 )y = D.
2

C13S08.042: Let f (x, y, z) = Ax2 + By 2 + Cz 2 − D. then

∇f (x, y, z) =  2Ax, 2By, 2Cz ,

so a vector normal to the graph of f (x, y, z) = 0 at the point P (x0 , y0 , z0 ) is

∇f (P ) =  2Ax0 , 2By0 , 2Cz0 .

Therefore an equation of the plane tangent to the graph at P is

2Ax0 (x − x0 ) + 2By0 (y − y0 ) + 2Cz0 (z − z0 ) = 0;

(Ax0 )x + (By0 )y + (Cz0 )z = A(x0 )2 + B(y0 )2 + C(z0 )2 = D.

C13S08.043: The equation of the paraboloid can be written in the form

H(x, y, z) = Ax2 + By 2 − z = 0, and ∇H(x, y, z) = 2Ax, 2By, −1.

A vector normal to the paraboloid at the point P (x0 , y0 , z0 ) is n = 2Ax0 , 2By0 , −1, and hence the plane
tangent to the paraboloid at P has an equation of the form

2Ax0 x + 2By0 y − z = d.

8
But the point P also lies on the plane, and hence

d = 2A(x0 )2 + 2B(y0 )2 − z0 = 2(Ax20 + By02 ) − z0 = 2z0 − z0 = z0 .

Hence an equation of the tangent plane is 2Ax0 x + 2By0 y − z = z0 , and the result in Problem 43 follows
immediately.

C13S08.044: Because v is not a unit vector, we must replace it with a unit vector having the same direction
before we can use the formulas of this section. So we take

v 1 2 2
u = = , ,− .
|v| 3 3 3

The gradient vector of f is

∇f = (y + z)i + (x + z)j + (x + y)k,

so ∇f (1, 2, 3) = 5i + 4j + 3k. Hence

Du f (P ) = 5, 4, 3 ·  13 , 23 , − 23  = 7
3

(degrees per kilometer) for the desired range of change of temperature with respect to distance.

C13S08.045: In the solution of Problem 44 we calculated ∇f (P ) = 5, 4, 3, and the unit vector in the
direction from P to Q is

−

PQ 2 2 1
u = − = , , .
| P Q| 3 3 3

Then
2 
Du f (P ) = ∇f (P ) · u = 5, 4, 3 · 2 1
3, 3, 3 = 7

(degrees per kilometer). Hence


  
dw dw ds deg km deg
= · = 7 2 = 14
dt ds dt km min min

as the hawk’s rate of change of temperature at P .

C13S08.046: The gradient vector is

∂f ∂f
∇f = i+ j = (0.006)xi − (0.008)y j,
∂x ∂y
so
 √ 
∇f (40, 30) = (0.24)i − (0.24)j = 0.24 2 u.

The unit vector

∇f (40, 30) i−j


u = = √
|∇f (40, 3 0)| 2

9
points southeast (into the fourth quadrant); this is the direction in which the bumblebee should initially fly.
And, according to Section 13.8, the directional derivative of f in this optimal direction is

Du f (40, 30) = |∇f (40, 30)| = (0.24) 2 ≈ 0.34

degrees per unit of distance.

C13S08.047: Part (a): If W (x, y, z) = 50 + xyz, then ∇W =  yz, xz, xy , so at the point P (3, 4, 1)
we have ∇W (P ) =  4, 3, 12 . The unit vector with the same direction as v =  1, 2, 2  is

v 1 2 2
u= = , , ,
|v| 3 3 3

and so the rate of change of temperature at P in the direction of v is


1 2 2 34
∇W (P ) · u =  4, 3, 12  · , , = .
3 3 3 3

Because distance is measured in feet, the units for this rate of change are degrees Celsius per foot.
Part (b): The maximal directional derivative of W at P is |∇W (P )| = | 4, 3, 12 | = 13 and the direction
in which it occurs is ∇W (P ) =  4, 3, 12 .

C13S08.048: Part (a): If W (x, y, z) = 100 − x2 − y 2 − z 2 , then ∇W =  −2x, −2y, −2z . The unit
vector with the same direction as v =  3, −4, 12  is

v 3 4 12
u= = ,− , ,
|v| 13 13 13

so the rate of change of W in the direction of v at the point P (3, −4, 5) is


3 4 12 170
∇W (P ) · u =  −6, 8, −10  · ,− , =− .
13 13 13 13

Because distance is measured in meters, the units for this rate of change are degrees Celsius per meter.

Part (b): The maximal directional derivative of W at P is |∇W (P )| = 10 2 and the direction in which
it occurs is ∇W (P ) =  −6, 8, −10 .

C13S08.049: Part (a): Given f (x, y) = 1


10 (x
2
− xy + 2y 2 ), let

1
g(x, y, z) = z − f (x, y); then ∇g(x, y, z) =
 y − 2x, x − 4y, 10 .
10

Thus a normal to the surface z = f (x, y) at the point P 2, 1, 25 is 10 1
 −3, −2, 10 . Hence an equation
of the plane tangent to this surface at P is

3 1 2
3x + 2y − 10z = 4; that is, z= x+ y− .
10 5 5
−
Part (b): Let Q denote the point (2, 1) and R the point (2.2, 0.9). Then v = QR =  0.2, −0.1 . Thus an
approximation to f (2.2, 0.9) is
11
f (2, 1) + ∇f (Q) · v = 0.4 +  0.3, 0.2  ·  0.2, −0.1  = = 0.44.
25
56
The true value is f (2.2, 0.9) = 125 = 0.448.

10
C13S08.050: Let F (x, y, z) = 2x2 + 3y 2 − z. The equation F (x, y, z) = 0 has the paraboloid as its graph.
The vector n =  4, −3, −1  is normal to the plane. All we require is that ∇F =  4x, 6y, −1  is parallel
to n. This leads to x = 1, y = − 12 , and (because F (x, y, z) = 0) z = 11
4 . Thus an equation of the required
plane is

   
1 11
4(x − 1) − 3 y + − z− = 0; that is, 16x − 12y − 4z = 11.
2 4

C13S08.051: Let F (x, y, z) = z 2 − x2 − y 2 and G(x, y, z) = 2x + 3y + 4z + 2. Then the cone is the graph
of F (x, y, z) = 0 and the plane is the graph of G(x, y, z) = 0. At the given point P (3, 4, −5) we have

∇F (3, 4, −5) =  −6, −8, −10  and ∇G(3, 4, −5) =  2, 3, 4 .

Let P denote the plane normal to the ellipse (the intersection of the cone and the first plane) at the point
P . Then a normal to P is

 
 i j k 
 
 
 
n =  −6, −8, −10  ×  2, 3, 4  =  −6 −8 −10  =  −2, 4, −2 .
 
 
 
2 3 4

We will use instead the parallel vector  1, −2, 1 . In the usual way we find that P has Cartesian equation
x − 2y + z + 10 = 0.

C13S08.052: Let F (x, y, z) = z 2 − x2 − y 2 and G(x, y, z) = 2x + 3y + 4z + 2. Then the cone is the graph
of F (x, y, z) = 0 and the plane is the graph of G(x, y, z) = 0. Then a plane P normal to the ellipse at the
point P (x, y, z) will itself have normal vector

 
 i j k 
 
 
 
n = ∇F × ∇G =  −2x −2y 2z  =  −8y − 6z, 8x + 4z, 4y − 6x .
 
 
 
2 3 4

The condition that the line tangent to the ellipse be horizontal implies that the third component of n must
be zero, so that P is a vertical plane. Thus we obtain the three simultaneous equations that the highest and
lowest points of the ellipse must satisfy:

4y − 6x = 0,

z 2 = x2 + y 2 , and

2x + 3y + 4z + 2 = 0.

They have exactly two simultaneous solutions, and thus we discover the answers:

11

52 + 16 13
Low point: x= ≈ 2.8125338566006110,
39

26 + 8 13
y= ≈ 4.2188007849009165,
13

−8 − 2 13
z= ≈ −5.0703675169759929.
3


52 − 16 13
High point: x= ≈ −0.1458671899339443,
39

26 − 8 13
y= ≈ −0.2188007849009165,
13

−8 + 2 13
z= ≈ −0.2629658163573405.
3

C13S08.053: Let F (x, y, z) = x2 + y 2 + z 2 − r2 and G(x, y, z) = z 2 − a2 x2 − b2 y 2 . Then the sphere is


the graph of F (x, y, z) = 0 and the cone is the graph of G(x, y, z) = 0. At a point where the sphere and
the cone meet, these vectors are the normals to their tangent planes. To show that the tangent planes are
perpendicular, it is sufficient to show that their normals are perpendicular. But


∇F · ∇G =  2x, 2y, 2z  ·  −2a2 x, −2b2 y, 2z  = −4a2 x2 − 4b2 y 2 + 4z 2 = 4(z 2 − a2 x2 − b2 y 2 ) = 0

because (x, y, z) lies on the cone. Therefore the tangent planes are perpendicular at every point of the
intersection of the sphere and the cone.

C13S08.054: The equation of the ellipsoid may be written in the form

F (x, y, z) = x2 + y 2 + 2z 2 − 2 = 0,

and ∇F (x, y, z) = 2x, 2y, 4z  is normal to the ellipsoidal surface at the point (x, y, z). A normal at
P (a, b, c) is therefore n = a, b, 2c, and in general there are four points on the ellipsoid with z-coordinate
c: They are (±a, ±b, c). Thus there are four normal vectors in question, ±a, ±b, 2c, and consequently
four tangent planes, with equations ±ax ± by + 2cz = d for some constant d (the same d for all four planes).
These planes meet the z-axis where x = y = 0, and thus all four have the same z-intercept z = d/(2c).
(If c = 0, then all four tangent planes are vertical—parallel to the z-axis—and none meets the z-axis. If
a = b = 0 then there is only one tangent plane meeting the ellipsoidal surface at a point with z-coordinate
c. If exactly one of a and b is zero, then there are only two tangent planes of the sort specified in Problem
54, but the same argument shows that both meet the z-axis at the same point.)

C13S08.055: The surface is the graph of the equation G(x, y, z) = 0 where

G(x, y, z) = xyz − 1, so that ∇G(x, y, z) = yz, xz, xy .

Suppose that P (a, b, c) is a point strictly within the first octant (so that a, b, and c are all positive). Note
that abc = 1. A vector normal to the surface at P is n = bc, ac, ab, and hence the plane tangent to the
surface at P has equation

bcx + acy + abz = d

12
for some constant d. Moreover, because P is a point of the surface,

bca + acb + abc = d; that is, d = 3abc.

Hence an equation of the tangent plane is bcx + acy + abz = 3abc. The intercepts of the pyramid therefore
occur at (3a, 0, 0), (0, 3b, 0), and (0, 0, 3c). Therefore, because of the right angle at the origin, the pyramid
has volume

1 27 9
V = (3a)(3b)(3c) = abc = ,
6 6 2
independent of the choice of P , as we were to show.

C13S08.056: Part (a): Given z = f (x, y) = 500 − (0.003)x2 − (0.004)y 2 , we begin by constructing a
vector v =  −1, 1  that points northwest, then the unit vector with the same direction:
 √ √ 
v 2 2
u= = − , .
|v| 2 2

Next, ∇f (x, y) =  −(0.006)x, −(0.008)y , so the value of the gradient at your position on the hill is
v = ∇f (−100, −100) =  0.6, 0.8 . So your initial rate of climb in the direction of u is

3 √ 4 √ 1 √
Du f (−100, −100) = v · u = − 2 + 2 = 2 ≈ 0.1414213562
10 10 10
in units of feet per foot; that is, you initially climb at the rate of about 0.1414 feet upward for every foot
you travel horizontally. Your initial angle of climb is
 
1 √
arctan 2
10

radians, approximately 8◦ 2 58.081 (a gentle slope).


Part (b): If instead you head northeast, repeat the previous calculations with the new unit vector
√ √ 
2 2
u= , .
2 2

Your initial rate of climb in the new direction of u will be

3 √ 4 √ 7 √
Du f (−100, −100) = v · u = 2 + 2 = 2 ≈ 0.989949493661
10 10 10
7√
feet per foot. The initial angle of climb will be arctan 10 2 radians, approximately 44◦ 42 38.241 , an
extremly steep climb, comparable to the last 30 meters up the north face of Rabun Bald.

C13S08.057: The hill is steepest in the direction of ∇z(−100, −100) =  0.6, 0.8 . The slope of the hill
in that direction is | 0.6, 0.8 | = 1, so that your initial angle of climb would be 45◦ . The compass heading
in the direction you are climbing is

π 4
− arctan
2 3
radians, approximately 36◦ 52 11.632 .

C13S08.058: Part (a): First we compute the gradient of

13
1000
z= :
1 + (0.00003)x2 + (0.00007)y 2
1
∇z = − 2  (0.06)x, (0.14)y  .
[1 + (0.00003)x2 + (0.00007)y 2 ]
The slope of this hill at the point (100, 100, 500) in the northwest direction is


 √ √ 
2 3 7 2 2 √
[∇z(100, 100)] ·  −1, 1  = − ,− · − , = − 2 ≈ −1.4142135623731.
2 2 2 2 2

Your initial angle of descent is thus approximately 1.4142 feet per foot and the angle of descent is approxi-
mately 54◦ 44 8.2 .
Part (b): The slope of this hill at the point (100, 100, 500) in the northeast direction is


√ √  √
2 3 7 2 2 5 2
[∇z(100, 100)] ·  1, 1  = − ,− · , =− ≈ −3.535533905933.
2 2 2 2 2 2

Your initial rate of descent is thus approximately 3.5355 feet per foot and the angle of descent is approximately
74◦ 12 13.6 , a very steep descent.

C13S08.059: Given

1000
z = f (x, y) = ,
1 + (0.00003)x2 + (0.00007)y 2

we first compute

600 000 000x 1 400 000 000y


∇f (x, y) = − , − .
(100000 + 3x2 + 7y 2 )2 (100000 + 3x2 + 7y 2 )2

Hence to climb the most steeply, the initial direction should be


3 7
∇f (100, 100) = − , − ,
2 2

and the initial rate of ascent will
√ be |∇f (100, 100)| = 22 58 ≈ 3.807886553 feet per foot. The initial
angle of climb will be arctan 12 58 ≈ 1.313982409 radians, approximately 75◦ 17 8.327 . The compass
heading is
 ◦
◦ 7
270 − arctan ≈ 203◦ 11 54.926 .
3

C13S08.060: Given
 2 
x + 3y 2
z = f (x, y) = 100 exp − ,
701

we first compute
 2   2 

200 x + 3y 2 600 x + 3y 2
∇f (x, y) = − x exp − , − y exp − .
701 701 701 701

14
Part (a):
   

6000 2100 12000 2100


v = ∇f (30, 20) = − exp − , − exp −
701 701 701 701

≈  −0.427965138743, −0.855930277485 

gives the initial direction in which you should head to climb the most steeply, and if you do so, your rate
of climb will initially be |v| ≈ 0.956959142228 feet per foot. That will be at an angle of approximately
43◦ 44 24.196 from the horizontal. The initial heading will be approximately
 ◦
0.855930277485
270◦ − arctan ≈ 206◦ 33 54.184 .
0.427965138743

Part (b): If you initially head west, with direction vector u =  −1, 0 , then your initial rate of ascent will
be
 
6000 2100
v·u = exp − ≈ 0.427965138743
701 701

feet per foot, so you will initially climb at an angle of approximately 23◦ 10 9.252 from the horizontal.

C13S08.061: Let

1
f (x, y) = (3x2 − 5xy + y 2 ).
1000
Then

1
∇f (x, y) =  6x − 5y, 2y − 5x  ,
1000
 
and therefore v = ∇f (100, 100) = 1
10 , − 10
3
.
1
√ √ 
Part (a): A unit vector in the northeast direction is u = 2 2, 2 , so the directional derivative of f
at (100, 100) in the northeast direction is
  √ √  √
v·u = 1
10 , − 10
3
· 12 2, 2 = − 10
1
2.
1√
Hence you will initially be descending the hill, and at an angle of arctan 10 2 below the horizontal,
◦  
approximately 8 2 58.081 .
√ 
Part (b): A unit vector in the direction 30◦ north of east is u = 12 3 , 1 , so the directional derivative
of f at (100, 100) in the direction of u is

 1  1 √  3− 3
10 , − 10 ·2 3, 1 = − ≈ −0.06339746.
3
20
Hence you will initially be descending the hill, and at an angle of approximately 3◦ 37 56.665 .

C13S08.062: Given: The two surfaces f (x, y, z) = 0 and g(x, y, z) = 0 both pass through the point P
at which both ∇f (P ) and ∇g(P ) exist. Part (a): Suppose that the two surfaces are mutually tangent at
P . Then their tangent planes there coincide. So their normal vectors at P are parallel. Therefore

∇f (P ) × ∇g(P ) = 0.

15
To prove the converse, simply reverse the steps in this argument.
Part (b): Suppose that the two surfaces are orthogonal at P . Then their tangent planes at P are perpen-
dicular. So their normal vectors at P are perpendicular. Therefore

∇f (P ) · ∇g(P ) = 0.

To prove the converse, simply reverse the steps in this argument. Also see the solution of Problem 53.

C13S08.063: Because u = a, b and v = c, d are not collinear, neither is zero and neither is a scalar
multiple of the other. Hence, as vectors, they are linearly independent, and this implies that the simultaneous
equations

afx (P ) + bfy (P ) = Du f (P ),

cfx (P ) + dfy (P ) = Dv f (P )

have a unique solution for the values of fx (P ) and fy (P ). Thus ∇f (P ) = fx (P ), fy (P ) is uniquely
determined, and therefore so is the directional derivative

Dw f (P ) = ∇f (P ) · w

in the direction of the arbitrary unit vector w. —C.H.E.

C13S08.064: Obviously f is continuous at the origin because f (x, y) → 0 as (x, y) → (0, 0). Next,
√ √ 3
f (at, bt) − f (0, 0)
3
at + 3
bt √ √ 3
3 3
Da,b f (0, 0) = lim = lim = a+ b (1)
t→0 t t→0 t
for all a and b. Thus every directional derivative exists. For instance, with a = 1 and b = 0 we find that
fx (0, 0) = 1, and with a = 0 and b = 1 we find that fy (0, 0) = 1. Therefore ∇f (0, 0) = 1, 1.
 
But if f were differentiable at the origin, it would follow with u = 35 , 45 that

3 4 7
Du f (0, 0) = ∇f (0, 0) · u = 1 · +1· = .
5 5 5
But the calculation in Eq. (1) shows that
  3
7
Du f (0, 0) = 3 3
5 + 3 4
5 ≈ 5.561701
= .
5

Therefore f is not differentiable at the origin. —C.H.E.

16

Anda mungkin juga menyukai