Homework 5 of Math 2000 Engineering Mathematics I

Hand to TA in tutorial in 7-8/10/2019.

Read class notes and examples before you start.

1. Consider the function f (x, y) = 3x2 − xy + y 3 .

(a) Find the rate of change of f at P (1, 2) in the direction v = 3i + 4j.

(b) In what direction (unit vector) does f decrease at (1,2) at the maximum
rate? What is this maximum rate of change?
(c) In what directions is the rate of change of f at P equal to zero? Your answer
should be a pair of opposite unit vectors.
2. Let S(h, r) be the production cost of an conical object, in which r is the base
radius, and h is the height. Suppose that S(h, r) is proportional to total surface
area of this object, more precisely
√
S(h, r) = r r2 + h2 + r2 .

Originally, the dimension of the object is given by (r, h) = (3, 4). Estimate the
cost change S(3.01, 3.98).

3. Find the differential of the function z(x, y) = x2 tan−1 (y 3 ).

4. Determine the number n of points in the level curve C : 3x2 − 4y 2 = −1 at which

the the normal vector of C is perpendicular to the vector 4i + 3j.
√
5. Find an approximate value of (1.99)3 (2.98)2 + (4.01) 2
√ by using the linear (first
order) approximation of the function f (x, y, z) = x3 y 2 + z 2 .

6. Let f (x, y, z) = x2 + y 2 + z. Find the tangent plane and normal line of the level
surface S : f (x, y, z) = 9 at the point P (1, 2, 4).

7. Find the local extreme values of f (x, y) = x2 + y 2 − 4y + 9 by means of second

derivative test.

8. (a) Find all critical points of f (x, y) = y 3 + 3x2 y − 6x2 − 6y 2 .

(b) Determine all local maxima, local minima values and saddle points of f (x, y).
Definitions, Formula and Theorems 4. Chain Rule III. Let w = f (x, y, z) be differentiable, and x = x(u, v), y = y(u, v)
and z = z(u, v). Define F (u.v) = f ( x(u, v), y(u, v), z(u, v) ). Then we have
1. Definition. Let z = f (x, y) be a differentiable functions, Then one can relate
the differentials dx, dy and dz by the following formula: ∂F ∂f ∂x ∂f ∂y ∂f ∂z
(u, v) = · + · + · ;
∂f ∂f ∂u ∂x ∂u ∂y ∂u ∂z ∂u
dz = dx + dy ♠. ∂F ∂f ∂x ∂f ∂y ∂f ∂z
∂x ∂y (u, v) = · + · + · .
∂v ∂x ∂v ∂y ∂v ∂z ∂v
Remark. This formula states the change ∆z = f (a + ∆x, b + ∆y) − f (a, b) in where fx , fy , fz are evaluated at the point (x(u, v), y(u, v), z(u, v)) instead of
the output can be approximated by (x, y, z).
∂f ∂f
∆z = f (a + ∆x, b + ∆y) − f (a, b) ≈ (a, b) ∆x + (a, b) ∆y ♡ 5. Proposition. Let C : f (x, y) = k be a level curve through a point P (a, b).
∂x ∂y Suppose that ∇f (a, b) ̸= (0, 0), then
∂f ∂f (i) ∇f (a, b) is normal to the tangent line to C at (a, b).
f (a + ∆x, b + ∆y) ≈ f (a, b) + (a, b) ∆x + (a, b) ∆y.
∂x ∂y (ii) The tangent line of C at P is given by the equation

By passing to linear approximation, i.e. ignoring the higher order terms
(∆x)i (∆y)j ( e.g. (0.1)2 (0.3)3 comparing to ∆x = 0.1 and ∆y = 0.3 ), and
if one replace ∆x, ∆y by dx, dy respectively, we have the equality in ♠ instead of
approximation in ♡.
2. Let w = f (x, y, z) be a differentiable function, a = (a, b, c) be an interior point of
Dom(f ), and v ̸= 0 be a constant vectors in R3 . Define g(t) = f (a + tv).
• Definition. The linear approximation of f at (a, b, c) is given by
∇f (a, b) · (x − a, y − b) = 0, i.e. fx (a, b)(x − a) + fy (a, b)(y − b) = 0.
6. Proposition. Let S : f (x, y, z) = k be a level surface through a point P (a, b, c).
Suppose that ∇f (a, b, c) ̸= (0, 0, 0), then
(i) ∇f (a, b, c) is normal to the tangent plane to D at (a, b, c).
(ii) The tangent plane to S at P is ∇f (a, b, c) · (x − a, y − b, z − c) = 0, i.e.
fx (a, b, c)(x − a) + fy (a, b, c)(y − b) + fz (a, b, c)(z − c) = 0.

L(x, y, z) = f (a, b, c) + fx (a, b, c)(x − a) + fy (a, b, c)(y − b) + fz (a, b, c)(z − c). 7. Theorem. (Second Derivatives Test). Suppose the second partial derivatives
of f (x, y) are continuous on a disk with center (a, b), and suppose that ∇f (a, b) =
• Proposition. The rate of change g ′ (t) = dg
dt (t) = ∇f (a + tv) · v, where (0, 0) i.e. (a, b) is a critical point of f . Define the discriminant D = D(a, b) of f
at (a, b)
∇f (a + tv) = ( fx (a + tv), fy (a + tv), fz (a + tv) ) . D = D(a, b) = fxx (a, b)fyy (a, b) − [fxy (a, b)]2 .
• Definition. The directional derivative Dv f (a) of f along v is given by (a) If D > 0 and A = fxx (a, b) > 0, then f (a, b) is a local minimum;
(b) If D > 0 and A = fxx (a, b) < 0, then f (a, b) is a local maximum;
Dv f (a) = ∇f (a) · u = αfx (a, b, c) + βfy (a, b, c) + γfz (a, b, c),
(c) If D < 0, then (a, b) is a saddle point of f ;
v
where u = ∥v∥ = (α, β, γ).
(d) If D = 0, then no conclusion can be drawn.
3. Chain Rule II. Let w = f (x, y) be a differentiable function, and x = x(u, v)
and y = y(u, v). Define F (u.v) = f ( x(u, v), y(u, v) ). Then we have
∂F ∂f ∂x ∂f ∂y
= · + · ;
∂u ∂x ∂u ∂y ∂u
∂F ∂f ∂x ∂f ∂y
= · + · ,
∂v ∂x ∂v ∂y ∂v
where fx , fy are evaluated at the point (x(u, v), y(u, v)) instead of (x, y).
Worked Examples 4. Approximate the function f (x, y, z) = x2 y − 2xz + 1 near the point (1, 1, 0) by a
linear function.
1. Let f (x, y) = 2x3 − xy + xy 2 . Find the rate of change of f at (2, 3) in the direction Solution. First f (1, 1, 0) = (1)2 (1) − 2(1)(0) + 1 = 2, and
of 3i − 4j.
3i−4j
Solution. Normalize the direction as u = ||3i−4j|| = 35 i − 45 j = ( 35 , − 54 ). ∂ ∂ 2
fx (1, 1, 0) = f (1, 1, 0) = (x y − 2xz + 1)|(x,y,z)=(1,1,0)
And the gradient of f (x, y) is given by ∇f (x, y) = (6x2 − y + y 2 , −x + 2xy). ∂x ∂x
Then the directional derivative Du f (2, 3) = ∇f (2, 3) · u = 10. = (2xy − z)|(x,y,z)=(1,1,0) = 2(1)(1) − 2(0) = 2;
∂
2. Find a vector in which the function f (x, y) = 2x3 − xy + xy 2 is increasing at (1, 2) fy (1, 1, 0) = f (1, 1, 0) = (x2 )|(x,y,z)=(1,1,0) = 1
∂x
at the maximal rate. ∂
Solution. Recall that Du f (a, b) is maximal if and only if u is parallel to ∇f (a, b). fz (1, 1, 0) = f (1, 1, 0) = (−2x)|(x,y,z)=(1,1,0) = −2.
∂z
The direction should be parallel to
The linear approximation L(x, y, z) should have the same value and partial deriva-
∇f (1, 2) = (6x2 − y + y 2 , −x + 2xy)|(x,y)=(1,2) = (8, 3). tives of f (x, y, z), so is given by

3. Determine the number n of points on the level surface L(x, y, z) = f (a, b, c) + fx (a, b, c)(x − a) + fy (a, b, c)(y − b) + fz (a, b, c)(z − c)
= 2 + 2(x − 1) + (y − 1) − 2z.
S : x2 + 2y 2 + 4z 2 + xy + 3yz = 1
5. Let f (x, y, z) = ex + yz, a = (1, 1, 1), and v = (1, 1, 0) = i + j be a direction.
at which the tangent plane is parallel to the xz-plane.
Solution. We first evaluate the gradient of f (x, y, z) = x2 + 2y 2 + 4z 2 + xy + 3yz (a) Compute h(t) = f (a + tv), and then compute the derivative h′ (0).
as follows (b) Compute the directional derivative using the formula ∇f (1, 1, 1) · v.
∇f (x, y, z) = ( 2x + y, 4y + x + 3z, 8z + 3y ).
Solution. (a) First a + tv = (1, 1, 1) + t(1, 1, 0) = (1 + t, 1 + t, 1). Then we have
If the tangent plane T of level surface
h(t) = f (1 + t, 1 + t, 1) = e1+t + (1 + t)1 = e1+t + t + 1, and
LS1 (f ) = { (x, y, z) ∈ Dom(f ) | f (x, y, z) = 1 } ( 1+t )
h′ (0) = e + 1 |t=0 = e + 1.
given by f (x, y, z) = 1 is parallel to the xz-plane (with normal direction (0, 1, 0),
then the normal direction ∇f (x, y, z) of T at the point (x, y, z) is parallel to (b) We calculate the partial derivatives of f as follows:
(0, 1, 0).
∇f (x, y, z) = ( fx (x, y, z), fy (x, y, z), fy (x, y, z) ) = (ex , z, y), and
In particular, the point (x, y, z) of S must satisfy the following conditions:
∇f (1, 1, 1) = (e, 1, 1).

By chain rule, we have

y = 2x, 3y = 8z and
the defining equation x2 + 2y 2 + 4z 2 + xy + 3yz = 1.
h′ (0) = ∇f (a) · v = ∇f (1, 1, 1) · (1, 1, 0) = (e, 1, 1) · (1, 1, 0) = e + 1.
After multiplying the last equation by 16, we have

32 = 4(2x)2 + 32y 2 + (8z)2 + 8(2x)y + 24(y)(8z)

= 4y 2 + 32y 2 + (3y)2 + 8y 2 + 72y 2 = 125y 2 .
√
So y = ± 32/125, from these we know the corresponding x and y coordinates.
In particular, there are only two points on the level surface satisfying the required
condition, i.e. n = 2.
6. The air temperature is a function T (H, P ) = P 4 +P H of the humidity H(x, y, z) = 7. Determine the nature of the critical points of f (x, y) = x3 + y 3 − 6xy.
xyz and pressure P (x, y, z) = 2x2 y + xz. Find the partial derivative ∂T
∂x at the Solution. As ∇f (x, y) = (3x2 − 6y, 3y 2 − 6x), so x2 = 2y and y 2 = 2x.
point (x, y, z) = (1, 0, 2). It follows that x4 = 4y 2 = 4 × 2x = 8x, i.e.
Solution. First, it follows from definition that P (1, 0, 2) = 2 = 0 + 1 × 2 =
0 = x4 − 8x = x(x3 − 23 ) = x(x − 2)(x2 + 2x + 4).
2, H(1, 0, 2) = 0, and
As (x2 + 2x + 4) = (x + 1)2 + 3 > 0, we have x = 0 or x = 2.
∂ So 2y = 02 or 22 , so the critical points of f are (0, 0) and (2, 2).
Px (1, 0, 2) = (2x2 y + xz)|(x,y,z)=(1,0,2) = (4xy + z)|(x,y,z)=(1,0,2) = 2,
∂x
∂ Next we need to apply the 2nd derivative test.
Hx (1, 0, 2) = (xyz)|(x,y,z)=(1,0,2) = xy|(x,y,z)=(1,0,2) = 0. A = fxx = 6x, B = fxy = −6, C = fyy = 6y, and the discriminant
∂x
∆(x, y) = AC − B 2 = fxx fyy − fxy
2
= (6x)(6y) − (−6)2 = 36(xy − 1).
Second ∂T ∂T
(P, H) = 4P 3 + H and hence when (P, H) = (2, 0), we And ∆(0, 0) = −36 < 0, ∆(2, 2) = 36(4 − 1) = 108.
∂H (P, H) = P, ∂P
have Hence (i) (2, 2) is a local minimum point of f with f (2, 2) = −8 (local minimum),
∂T and (ii) (0, 0) is a saddle point of f.
(2, 0) = P |(P,H)=(2,0) = 2,
∂H 8. (a) Find all critical points of f (x, y) = y 3 + 3x2 y − 6x2 − 6y 2 .
∂T (b) Determine all local maxima, local minima values and saddle points of f (x, y).
(2, 0) = (4P + H)|(P,H)=(2,0) = 4 × 2 + 0 = 32.
3 3
∂P
Solution. We first note that f is a polynomial, so the domain D of f (x, y) is the
We can use the chain rule entire xy-plane. Next we determine the critical point(s) of f (x, y) as follows:
∂T ∂T ∂H ∂T ∂P ∇f (x, y) = (0, 0) ⇔ ( 6xy − 12x, 3y 2 + 3x2 − 12y) = (0, 0).
(x, y, z) = · + ·
∂x ∂H ∂x ∂P ∂x
From the first equation, we have 0 = 6xy − 12x = 6x(y − 2), i.e. x = 0 or y = 2.
= P (x, y, z) · yz + (4P (x, y, z)3 + H(x, y, z))(4xy + z)
∂T (a) If x = 0, then by the second equation, we have
(1, 0, 2) = 2 × 0 + (32) × (4 × 0 + 2) = 64.
∂x 0 = 3y 2 + 3x2 − 12y = 3y(y − 4) ⇐⇒ y = 0 or y = 4.
Remark. Alternatively one can write T directly as a function of (x, y, z) and So (x, y) = (0, 0), (0, 4).
hence one can compute the partial derivative normally.
(b) If y = 2, then by the second equation, we have
T (x, y, z) = (P (x, y, z)4 + P (x, y, z) · H(x, y, z) 0 = 3y 2 + 3x2 − 12y = 3(x2 − 4) ⇐⇒ x = 2 or x = −2.
2 4 2
= (2x y + xz) + (2x y + xz)xyz
So (x, y) = (2, 2), (−2, 2).
∂T ∂ ( )
(x, y, z) = (2x2 y + xz)4 + (2x2 y + xz)xyz
∂x ∂x Critical points of f (x, y) are: (x, y) = (0, 0), (0, 4), (2, 2) and (−2, 2).
∂ ( ) (b) Finally, we can apply the second order derivative test to determine the nature
= (2x2 y + xz)4 + (2x3 y 2 z + x2 yz 2 ) of these 4 critical points.
∂x
∂ A = fxx (x, y) = 6y − 12, B = fxy = 6x, C = fyy (x, y) = 6y − 12, and
= 4(2x2 y + xz)4−1 · (2x2 y + xz) + (6x2 y 2 z + 2xyz 2 ) D = AC − B 2 = (6y − 12)2 − 36x2 .
∂x
= 4(2x2 y + xz)3 (4xy + z) + 6x2 y 2 z + 2xyz 2 • At (x, y) = (0, 0), we have (A, B, C, D) = (−12, 0, −12, 144). As A < 0 and
∂T 3 D > 0, the function f has a local maximum at (x, y) = (0, 0) with f (0, 0) = 0.
(1, 0, 2) = 4(2) (2) = 64.
∂x • At (x, y) = (0, 4), we have (A, B, C, D) = (12, 0, 12, 144). As A > 0 and D >
0, the function f has a local minimum at (x, y) = (0, 4) with f (0, 4) = −32.
• At (x, y) = (2, ±2), we have D = −144 < 0, so these 2 points are the saddle
points of the function f (x, y).