MATH 2023 • Spring 2015-16 • Multivariable Calculus

Problem Set #2 • Multivariable Functions, Partial Derivatives

p
1. (F) Let f ( x, y) = y − x2
(a) What is the (largest possible) domain of f ?

Solution: The domain is {( x, y) : y ≥ x2 }, the region above the parabola y = x2

(including the parabola) in R2 .

(b) Sketch the level sets f = 0, f = 1 and f = 2 in the same diagram.

Solution:
20

15

10

-4 -2 2 4

2. (F) Let
1
f ( x, y) = p
x2 + y2 − 1
(a) What is the (largest possible) domain of f ?

Solution: The domain is {( x, y) : x2 + y2 > 1}, the region outside the unit circle
x2 + y2 = 1, excluding the circle itself.

(b) Sketch the level sets f = 1, f = 2 and f = 3 in the same diagram.

Solution:
2

-2 -1 1 2

-1

-2

Page 1
MATH 2023 Multivariable Functions, Partial Derivatives Problem Set #2

1
(c) Repeat (a) and (b) for the function g( x, y) = p .
1 − x 2 − y2

Solution: The domain is {( x, y) : x2 + y2 < 1}, the region inside the unit circle
x2 + y2 = 1, excluding the circle itself.
1.0

0.5

-1.0 -0.5 0.5 1.0

-0.5

-1.0

3. (F) Compute all the first and second partial derivatives of the following functions. For
the second partials f xy and f yx , compute both and verify that they are indeed the same.
(a) f ( x, y) = y2015 + 2x2 + 2xy

Solution: (Answer only)

f x = 4x + 2y f y = 2x + 2015y2014
f xx = 4 f xy = 2
f yx = 2 f yy = 2015 × 2014y2013

2y
(b) f ( x, y) = e x

Solution: (Answer only)

2 2
f x = 2e x y xy f y = e x y x2
2 2 2 2
f xx = 2e x y y + 4e x y x2 y2 f xy = 2e x y x + 2e x y x3 y
2 2 2y
f yx = 2e x y x + 2e x y x3 y f yy = x4 e x

x
(c) f ( x, y) = x 2 + y2

Solution: (Answer only)

y2 − x 2 2xy
fx = fy = −
( x2 + y2 )2 + y2 )2
( x2
2 x3 − 3xy 2 2y y2 − 3x2



f xx = f xy = −
( x 2 + y2 )3 ( x 2 + y2 )3
2y y2 − 3x2 2x x2 − 3y2



f yx = − f yy = −
( x 2 + y2 )3 ( x 2 + y2 )3

Page 2
MATH 2023 Multivariable Functions, Partial Derivatives Problem Set #2

(d) f ( x, y) = x ln( x2 + y2 )

Solution: (Answer only)

2x2 2xy
+ log x2 + y2


fx = 2 2
fy =
x +y x2 + y2
2 x3 + 3xy2 2y y2 − x2



f xx = f xy =
( x 2 + y2 )2 ( x 2 + y2 )2
2y y2 − x2 2x x2 − y2



f yx = f yy =
( x 2 + y2 )2 ( x 2 + y2 )2

∂f
4. (FF) Compute the first partial derivative ∂x of the following functions (where x, y > 0).
y
(a) f ( x, y) = e x

Solution:
∂f ∂ (xy ) ∂ (xy ) ∂ y
= e = e · x
∂x ∂x ∂( xy ) ∂x
y y
= e(x ) · yx y−1 = yx y−1 e x

x
(b) f ( x, y) = ey

Solution:
∂f ∂ (y x ) ∂ (y x ) ∂ x
= e = e · y
∂x ∂x ∂(y x ) ∂x
x x
= e(y ) · y x ln y = yey ln y

y
(c) f ( x, y) = x e

Solution: (Answer only)

∂f y
= e y x e −1
∂x

x
(d) f ( x, y) = ye

Solution: (Answer only)

∂f x
= ye e x ln y
∂x

e
(e) f ( x, y) = x y

Solution: (Answer only)

∂f e
= y e x y −1
∂x

Page 3
MATH 2023 Multivariable Functions, Partial Derivatives Problem Set #2

e
(f) f ( x, y) = y x

Solution: (Answer only)

∂f e
= ex e−1 y x ln y
∂x

5. (F) Compute both the third-order derivatives h xyy and hyyx of the following function, and
verify that they are indeed the same.

h( x, y, z) = cos( x2 + y3 z).

Solution:

h x = −2x sin x2 + y3 z

h xy = (h x )y = −6xy2 z cos x2 + y3 z

h xyy = h xy y = 18xy4 z2 sin x2 + y3 z − 12xyz cos x2 + y3 z

hy = −3y2 z sin x2 + y3 z

hyy = hy y = −6yz sin x2 + y3 z − 9y4 z2 cos x2 + y3 z

hyyx = hyy x = 18xy4 z2 sin x2 + y3 z − 12xyz cos x2 + y3 z

 
∂ ∂f
6. (FF) Find the second derivative ∂x ∂y of each function f ( x, y ) below. [Hint: There is
a smart way to compute each of them.]
(a)
f ( x, y) = sin( x + y) cos( x − y)

Solution:
∂f
= cos( x + y) cos( x − y) + sin( x + y) sin( x − y)
∂y
= cos( x + y − ( x − y))
= cos(2y)

Here we have used the compound-angle

 formula for cos. Since this does not
∂ ∂f
depend on x we have ∂x ∂y = 0 .

Page 4
MATH 2023 Multivariable Functions, Partial Derivatives Problem Set #2

(b)
1
! 2015
sin2016 y + cos2014 y
f ( x, y) = cos( xy) + .
sin2 log(y4 + 1) + 2015

Solution: Since the second term does not depend on y, we can switch the order
of the partial derivatives taken and obtain:
   
∂ ∂f ∂ ∂f ∂
= = (−y sin( xy) + 0) = − sin( xy) − xy cos( xy)
∂x ∂y ∂y ∂x ∂y

(c)
e x +y + e x −y
f ( x, y) =
e x +y − e x −y

Solution: By factoring out e x in both numerator and denominator we obtain:

e x ey + e x e−y e x ( ey + e−y ) ey + e−y

f ( x, y) = = =
e x ey − e x e−y e x ( ey − e−y ) ey − e−y

Therefore it makes sense to switch the order of partial differentiation and calcu-
late
∂f
=0
∂x
   
∂ ∂f ∂ ∂f
Thus we also have ∂x ∂y = ∂y ∂x = 0.

∂2 f
7. (FF) Suppose that f ( x, y) is a function such that ≡ 0. Show that f can be decom-
∂x∂y
posed into the form:
f ( x, y) = F ( x ) + G (y)
where F ( x ) and G (y) are some single-variable functions.

Solution: Given that:

∂2 f
 
∂ ∂f
= = 0,
∂x ∂y ∂x∂y
∂f
we know ∂y is independent of x, and depends only on y. Therefore, it can be written
as:
∂f
= g(y)
∂y
for an arbitrary differentiable function g(y).
Now, we obtained that the y-derivative of f is g(y). To find the function f , we can
integrate g(y) with respect to y:
ˆ
f ( x, y) = g(y) dy + F ( x ).

Page 5
MATH 2023 Multivariable Functions, Partial Derivatives Problem Set #2

∂
As ∂y is a partial derivative, the integration “constant” is not really a constant but is a
quantity not depending on y. In other words, the integration “constant” is a function
F ( x ) of x.
ˆ
Since g(y) dy is also an arbitrary function of y, for simplicity we relabel it as G (y).
Therefore, we get f ( x, y) = F ( x ) + G (y).

8. (FF) Let u( x, y, z, t) be the temperature at the point ( x, y, z) at the time t. Combining with
several important laws in thermodynamics, including the Fourier’s Law and conservation
of energy, it can be derived (detail omitted) that the temperature function u( x, y, z, t)
satisfies the following equation:
 2
∂ u ∂2 u ∂2 u

∂u
=k + 2+ 2
∂t ∂x2 ∂y ∂z
where k is a positive constant depending only on the medium. This equation is known as
the heat equation.
The study of the heat equation is an important topic in physics, engineering and mathe-
matics (both pure and applied). Through solving the heat equation with an initial condi-
tion u( x, y, z, 0) = g( x, y, z), it predicts how heat diffuses for a given an initial heat profile
g( x, y, z) at time t = 0.
Your task in this problem is to verify that the following given function is a solution to the
heat equation:
x 2 + y2 + z2
 
1
ϕ( x, y, z, t) = 3 exp − .
(4πkt) 2 4kt

This particular solution ϕ represents the heat diffusion with highly concentrated heat
source at the origin (0, 0, 0) at time t = 0. As time goes by, the temperature profile
becomes more and more uniformly distributed. (In physics, this solution is also closely
related to the Dirac delta function.)
By following the outline below, show that ϕ satisfies the heat equation:
(a) Show that:
3 3 x 2 + y2 + z2
ln ϕ( x, y, z, t) = − ln(4πk ) 2 − ln t − .
2 4kt

Solution: Take ln on the function ϕ:

!
x 2 + y2 + z2

1
ln ϕ( x, y, z, t) = ln 3 exp −
(4πkt) 2 4kt
x 2 + y2 + z2
   
1
= ln + ln exp −
(4πk)3/2 t3/2 4kt
2
x +y +z 2 2
= − ln((4πk)3/2 t3/2 ) − (ln exp u = u)
4kt
x 2 + y2 + z2
= − ln(4πk)3/2 − ln t3/2 −
4kt
3 x 2 + y2 + z2
= − ln(4πk)3/2 − ln t −
2 4kt

Page 6
MATH 2023 Multivariable Functions, Partial Derivatives Problem Set #2

(b) Using (a), show that:

x 2 + y2 + z2
 
∂ϕ 3
= − ϕ.
∂t 4kt2 2t

Solution: Differentiating both sides of (a) with respect to t:

x 2 + y2 + z2
 
∂ ∂ 3/2 3
ln ϕ = − ln(4πk) − ln t −
∂t ∂t 2 4kt
d ∂ϕ 3 2 2
x +y +z ∂ 12
ln ϕ · = 0− −
dϕ ∂t 2t 4k ∂t t
x 2 + y2 + z2
 
1 ∂ϕ 3 1
=− − − 2
ϕ ∂t 2t 4k t
x 2 + y2 + z2 3
= 2
− .
4kt 2t
 2
x + y2 + z2

∂ϕ 3
= − ϕ.
∂t 4kt2 2t

(c) Using (a) again, show that:

∂2 ϕ x2
 
∂ϕ xϕ 1
=− and 2
= − 1 ϕ.
∂x 2kt ∂x 2kt 2kt

Solution: Differentiating both sides of (a) with respect to x:

x 2 + y2 + z2
 
∂ ∂ 3 3/2
ln ϕ = − ln(4πk) − ln t −
∂x ∂x 2 4kt
 2 2 2

d ∂ϕ ∂ x +y +z
ln ϕ · = 0+0−
dϕ ∂x ∂x 4kt
1 ∂ϕ 2x + 0 + 0 x
=− =−
ϕ ∂x 4kt 2kt
∂ϕ xϕ
=− .
∂x 2kt
Differentiate both sides of the above result by x:

∂2 ϕ ∂  xϕ 
= −
∂x2 ∂x 2kt
1
=− ( ϕ + xϕ x )
2kt
1  xϕ  ∂ϕ
=− ϕ−x· (from our result of )
2kt 2kt ∂x
 2 
1 x ϕ
= −ϕ
2kt 2kt
 2 
1 x
= −1 ϕ (factor out ϕ).
2kt 2kt

Page 7
MATH 2023 Multivariable Functions, Partial Derivatives Problem Set #2

(d) Hence, verify that ϕ satisfies the heat equation: ϕt = k( ϕ xx + ϕyy + ϕzz ).

Solution: Similar calculation as in (c) shows:

∂2 ϕ
 2
∂2 ϕ
  2 
1 y 1 z
= −1 ϕ and = − 1 ϕ.
∂y2 2kt 2kt ∂z2 2kt 2kt

Therefore,

k ( ϕ xx + ϕyy + ϕzz )
  2   2   2  
1 x 1 y 1 z
= k· −1 ϕ+ −1 ϕ+ −1 ϕ
2kt 2kt 2kt 2kt 2kt 2kt
 2   2   2 
1 x 1 y 1 z
= −1 ϕ+ −1 ϕ+ −1 ϕ
2t 2kt 2t 2kt 2kt 2t
 2
y2 z2

x 1 1 1
= − + − + − ϕ
4kt2 2t 4kt2 2t 4kt2 2t
 2
x + y2 + z2

3
= − ϕ
4kt2 2t
= ϕt (from part (b)).

(e) (Optional) Show that

(
∞ if ( x, y, z) = (0, 0, 0)
lim ϕ( x, y, z, t) =
t → 0+ 0 if ( x, y, z) 6= (0, 0, 0)

Solution: Since we are taking t → 0+ , we can regard t > 0.

If ( x, y, z) = (0, 0, 0), then
 2
0 + 02 + 02

1 1
ϕ( x, y, z, t) = ϕ(0, 0, 0, t) = 3 exp − = 3 .
(4πkt) 2 | 4kt (4πkt) 2
{z }
=1
1
lim ϕ(0, 0, 0, t) = lim 3 = ∞.
t → 0+ t → 0+ (4πkt) 2
If ( x, y, z) 6= (0, 0, 0), then x2 + y2 + z2 6= 0. For simplicity, denote:
x 2 + y2 + z2
A := − 6= 0
4k
then A
1 − At e− t
ϕ( x, y, z, t) = 3 e = .
(4πkt) 2 (4πk)3/2 t3/2
A
When taking t → 0+ , we regard ( x, y, z) as constants. As e− t goes to 0 much
faster than t3/2 does, we can conclude that:
A
e− t
lim 3/2 = 0.
t → 0+ t

Therefore, it concludes that lim ϕ( x, y, z, t) = 0 when ( x, y, z) 6= 0.

t → 0+

Page 8