Department of Economics

November 2003
Arne Strøm

ECON3120/4120 Mathematics 2, autumn 2003

Problem solutions for seminar no. 10, 10–14 November 2003
(For practical reasons some of the solutions may include problem parts that were
not on the problem list for the seminar.)

EMEA, 7.3.8 (= MA I, 7.3.8)

(b) From f (x) = 1/x and f 0 (x) = −1/x2 we get
1 1 dx
∆y = f (x + dx) − f (x) = − =− ,
x + dx x x(x + dx)
dx
dy = f 0 (x) dx = − .
x2
(i) With x = 3 and dx = −1/10,
1
− 10 1 1
∆y = − 1

= = ≈ 0.01149,
3 3 − 10 3 · 29 87
1
− 10 1
dy = − 2
= ≈ 0.01111.
3 90
(ii) With x = 3 and dx = −1/100,
1
− 100 1 1
∆y = − 1

= = ≈ 0.001115,
3 3 − 100 3 · 299 897
1
− 100 1
dy = − 2 = ≈ 0.001111.
3 900
EMEA, 12.3.11 (= MA I, 12.2.4)
(a) F (1, 3) = 1e3−3 + 1 · 32 − 2 · 3 = 1 + 9 − 6 = 4, so the point (1, 3) does lie on
the level curve F (x, y) = 4.
The first-order partial derivatives of F are

F10 (x, y) = ey−3 + y 2 and F20 (x, y) = xey−3 + 2xy − 2.

Hence, the slope of the tangent to the level curve F (x, y) = 4 at the point (1, 3) is

F10 (1, 3) 1 + 32
y0 = − = − = −2.
F20 (1, 3) 1+6−2
The tangent therefore has the equation y − 3 = (−2)(x − 1), that is,

y = −2x + 5.

1
h03ste10 10.11.2003 24
(b) Taking the logarithm of both sides, we get

(1 + c ln y) ln y = ln A + α ln K + β ln L.

Differentiation with respect to K gives

c ∂y 1 ∂y α
ln y + (1 + c ln y) = .
y ∂K y ∂K K

If we solve this equation with respect to ∂y/∂K, we get

∂y αy
= .
∂K K(1 + 2c ln y)

In a similar fashion,
∂y βy
= .
∂L L(1 + 2c ln y)

EMEA, 12.3.10 (= MA I, 12.2.7)

Differentiating implicitly with respect to x and with respect to y gives

1 − azx0 = f 0 (y − bz)(−bzx0 ) (1)

−azy0 = f 0 (y − bz)(1 − bzy0 ) (2)

From (1) we get 1 = (a − bf 0 (y − bz))zx0 , which shows that a − bf 0 (y − bz) 6= 0 and

1
zx0 = .
a− bf 0 (y − bz)

Equation (2) then gives

−f 0 (y − bz)
zy0 = .
a − bf 0 (y − bz)
Hence,
a − bf 0 (y − bz)
azx0 + bzy0 = = 1.
a − bf 0 (y − bz)

EMEA, 12.7.5 (= MA I, 12.3.1)

(a) f (1.02, 1.99) = 3 · 1.022 + 1.02 · 1.99 − 1.992 = 1.1909.
(b) We have f10 (x, y) = 6x + y, f20 (x, y) = x − 2y, so f10 (1, 2) = 8, f20 (1, 2) = −3.
The linear approximation formula then gives

f (1.02, 1.99) ≈ f (1, 2) + 8 · 0.02 − 3 · (−0.01) = 1.19.

The error (= exact value − approximate value) is 0.0009.

2
h03ste10 10.11.2003 24
EMEA, 12.7.6 (= MA I, 12.3.2)
The linear approximation formula yields
v(1.01, 0.02) ≈ v(1, 0) + v10 (1, 0) · 0.01 + v20 (1, 0) · 0.02
= −1 − 34 · 0.01 + 13 · 0.02 = −1 − 31 · 0.02 ≈ −1.0067.

EMEA, 12.7.1 (= MA I, 12.3.3)

In both (a) and (b) we use formula (1) on page 442 (formula (3) on page 433 in
MA I) to give the approximation

f (x, y) ≈ f (0, 0) + f10 (0, 0)x + f20 (0, 0)y.

√
(a) For f (x, y) = 1 + x + y we get
1
f10 (x, y) = f20 (x, y) = √ ,
2 1+x+y
so the linear approximation to f (x, y) about (0, 0) is

f (x, y) ≈ 1 + 21 x + 12 y.

(b) For f (x, y) = ex ln(1 + y),

x ex
f10 (x, y) = e ln(1 + y) and f20 (x, y) = .
1+y
Here, f (0, 0) = f10 (0, 0) = e0 ln 1 = 0 and f20 (0, 0) = 1. That yields

f (x, y) = ex ln(1 + y) ≈ 0 + 0 · x + 1 · y = y.

EMEA, 12.7.3 (= MA I, 12.3.5)

We calculate the partial derivatives of g ∗ :
∂g ∗ 1
1/(1−β)−1
= (1 + µ)(1 + ε)α (1 + ε)α
∂µ 1−β
1
β/(1−β)
= (1 + µ)(1 + ε)α (1 + ε)α ,
1−β
∗
∂g 1
β/(1−β)
= ··· = (1 + µ)(1 + ε)α (1 + µ)α(1 + ε)α−1 .
∂ε 1−β
1 β
(Among other things, we have used that −1 = .) Hence, at the point
1−β β−1
(µ, ε) = (0, 0), we have ∂g ∗ /∂µ = 1/(1 − β) and ∂g ∗ /∂ε = α/(1 − β), and so for
small values of µ and ε, we have
1 α 1
g ∗ (µ, ε) ≈ g ∗ (0, 0) + ·µ+ ·ε= (µ + αε),
1−β 1−β 1−β
because g ∗ (0, 0) = 0.

3
h03ste10 10.11.2003 24
EMEA, 12.7.7 (= MA I, 12.3.6)
We shall use formula (3) on page 444 (formula (4) on page 434 in MA I) to find
an equation for the tangent plane.
(a) Here, ∂z/∂x = 2x and ∂z/∂y = 2y. At the point (1, 2, 5), we get ∂z/∂y = 2
og ∂z/∂x = 4, so the tangent plane at this point has the equation

z − 5 = 2(x − 1) + 4(y − 5) ⇐⇒ z = 2x + 4y − 5.

(b) From z = (y − x2 )(y − 2x2 ) = y 2 − 3x2 y + 2x4 we get ∂z/∂x = −6xy + 8x3
and ∂z/∂y = 2y − 3x2 . Thus, at (1, 3, 2) we have ∂z/∂x = −10 and ∂z/∂y = 3.
The tangent plane is given by the equation

z − 2 = −10(x − 1) + 3(y − 3) ⇐⇒ z = −10x + 3y + 3.

Exam problem 38
(a) Computing differentials, we get

v d(u2 ) + u2 dv − du = 3x2 dx + 6y 2 dy
eux d(ux) = y dv + v dy,
that is,
2uv du + u2 dv − du = 3x2 dx + 6y 2 dy
ueux dx + xeux du = y dv + v dy.

If we substitute the values x = 0, y = 1, u = 2, and v = 1, we get

4 du + 4 dv − du = 6 dy
2 dx + 0 du = dv + dy.

After a bit of calculation this yields

8 10
du = − dx + dy and dv = 2 dx − dy
3 3
at the point P . Hence, at this point
∂u 10 ∂v
= and = 2.
∂y 3 ∂x

(b) We get
8 10 8 10 2.8
∆u ≈ du = − dx + dy = − · 0.1 + · (−0.2) = − ≈ −0.933
3 3 3 3 3
and
∆v ≈ dv = 2 dx − dy = 2 · 0.1 − (−0.2) = 0.4.

4
h03ste10 10.11.2003 24
Exam problem 57
(a) Since

f (tx, ty) = (ty)3 + 3(tx)2 (ty) = t3 y 3 + 3t3 x2 y = t3 f (x, y),

f is homogeneous of degree 3. It follows that the desired constant is k = 3,

cf. Euler’s theorem.
Of course, we could also calculate directly:

xf10 (x, y) + yf20 (x, y) = x · 6xy + y(3y 2 + 3x2 ) = 3y 3 + 9x2 y = 3f (x, y).

(b) For every value of x, the function F (x, y) = y 3 + 3x2 y is strictly increasing
with respect to y, with F (x, y) → −∞ as y → −∞ and F (x, y) → ∞ as y → ∞.
It follows that the equation F (x, y) = −13 defines y as a function of x over the
entire real line.
Implicit differentiation gives

3y 2 y 0 + 6xy + 3x2 y 0 = 0,
6xy 2xy
y0 = − 2 2
=− 2 .
3x + 3y x + y2

This is the slope of the tangent to the curve at the point (x, y). With (x, y) =
(2, −1) we get y 0 = −4/5. Hence, the tangent to the curve at the point (2, −1) is
given by the equation
4 4 13
y − (−1) = (x − 2), that is, y= x− .
5 5 5
−2xy
(c) From y 0 = we get
x2 + y 2

00 (−2y − 2xy 0 )(x2 + y 2 ) − (−2xy)(2x + 2yy 0 )

y = .
(x2 + y 2 )2

At the point (2, −1) we have y 0 = −4/5 and

16 8 32


2 − 5 (4 + 1) − 4 4 − 5 ) (10 − 16) − (16 − 5 ) 78
y 00 = 2
= =− < 0.
(4 + 1) 25 125

Thus, y is a concave function of x around this point.

(d) Since y(y 2 + 3x2 ) = −13, we have (x, y) 6= (0, 0), and therefore

13
y=− < 0.
y2 + 3x2

This shows that all points on the curve lie below the x-axis.

5
h03ste10 10.11.2003 24
 In part (b) we showed that
 
0 2xy −2y
y =− 2 = · x.
x + y2 x2 + y 2

−2y
Since > 0, we have
x2+ y2

y 0 < 0 for x < 0 and y 0 > 0 for x > 0.

This means that y decreases when x increases in (−∞, 0], and increases when
x increases in [0, ∞). Hence, y attains√its least value,
√ ymin , for x = 0, and so
(ymin )3 + 0 = −13, which yields ymin = 3 −13 = − 3 13 .
(Alternatively we could try to solve the problem

minimize y subject to y 3 + 3x2 y = −13 (∗)

by Lagrange’s method. The Lagrangian is

L = y − λ(y 3 + 3x2 y + 13),

and the equations L0x = L0y = 0 give

−6λxy = 0, (1)
1 − 3λy − 3λx2 = 0.
2
(2)

We can see from equation (2) that we must have λ 6= 0. Moreover, we showed above
√
that y < 0. Hence, from (1) we get x = 0, and the constraint yields y = − 3 13 .
This is the only possible solution of the problem (∗). But it then remains to show
that it really is a solution of the problem.)

Exam problem 99
Implicit differentiation with respect to x yields

exy (y + xy 0 ) − e−xy (y + xy 0 )
y + xy 0 − = 0.
exy + e−xy
Rearranging this equation yields
y
2ye−xy + 2xe−xy y 0 = 0 =⇒ y 0 = − .
x
It then follows that
d 0 d y y0 x − y −y − y 2y
y 00 = y =− =− 2
=− 2
= 2.
dx dx x x x x

6
h03ste10 10.11.2003 24
It may be worthwhile to note that

F (x, y) = 1 + xy − ln(exy + e−xy ) = g(xy),

where
g(u) = 1 + u − ln(eu + e−u ).

The function g is strictly increasing because

eu − e−u 2e−u
g 0 (u) = 1 − = >0
eu + e−u eu + e−u
for all u. Therefore g has an inverse function, and

F (x, y) = c ⇐⇒ g(xy) = c ⇐⇒ xy = g −1 (c).

This means that a level curve for F is also a level curve for xy, and we could have
used this to find y 0 and y 00 with less work.
NOTE! In any case it is a bad idea to use the formula
 
 0 F10 F20 
1 1  0
(F20 )2 − 2F12 (F10 )2 =
 
y 00 = − 00
F11 00 0 0 00
F1 F2 + F22 F 00
F11 00 
F12
(F20 )3 (F20 )3  10 00 00 

F2 F21 F22

on page 425 in EMEA (page 426 in MA I). If we have managed to find an expression
for y 0 in terms of x and y, it is almost always better to use that expression in order
to find y 00 .

7
h03ste10 10.11.2003 24
 Extra problems:
Exam problem 89
(a) We differentiate the equation with respect to x (keeping y constant). That
yields
1 1
3z 2 ln z + z 3 = 6z 2 z10 (x, y) ln z + 2z 3 z10 (x, y),
z z
that is,
3z 2 ln z + z 2 = (6z 2 ln z + 2z 2 )z10 (x, y) (∗)

Putting x = y = z = e, we get

3e2 + e2 = (6e2 + 2e2 )z10 (e, e),

and so
4e2 1
z10 (e, e) =
2
= .
8e 2
If we differentiate (∗) with respect to x again, we get

6z ln z + 3z + 2z
1
= (12zz10 ln z + 6z 2 z10 + 4zz10 )z10 + (6z 2 ln z + 2z 2 )z11
00
z
= (12z ln z + 6z + 4z)(z10 )2 + (6z 2 ln z + 2z 2 )z11
00
.

Then, with x = y = z = e and z10 (e, e) = 1/2 in this equation, we get

1
6e + 3e + 2e = (12e + 6e + 4e) + (6e2 + 2e2 )z11
00
(e, e),
4
11e
11e = + 8e2 z11
00
(e, e),
2
which finally gives
00 11
z11 (e, e) = .
16e
(b) Differentiation with respect to x gives

3x2 F (xy) + x3 F 0 (xy)(y + xy 0 ) + exy (y + xy 0 ) = 1.

Note that it says F (xy), not F (x, y), in the problem. (F is a function of one
variable, not two.) And by the rule for differentiation of products, (xy)0 = y + xy 0 .
Then, with x = 1, y = 0, and F (0) = 0, we get

3F (0) + F 0 (0)(0 + y)0 + e0 (0 + y 0 ) = 1 ⇐⇒ F 0 (0)y 0 + y 0 = 1,

that is,
1
y0 = .
F 0 (0) +1

8
h03ste10 10.11.2003 24
Exam problem 92
2
(a) Let F (x, y) = xex y + 3x2 − 2y − 4. The formula for implicit differentiation
then yields
2 2
dy F10 (x, y) ex y + 2x2 yex y + 6x
=− 0 =− .
dx F2 (x, y) x3 ex2 y − 2
In particular, at the point (x, y) = (1, 0) we get

dy e0 + 0 + 6
=− 0 = 7.
dx e −2
(b) Differentiation with respect to z gives
dx y dy dy
e + xey + f (z) + yf 0 (z) = 0
dz dz dz
dx 
0 dx 0 dy 
g(x, y) + x g1 (x, y) + g2 (x, y) + 2z = 0
dz dz dz
By rearranging these equations we get
dx
dy
ey + xey + f (z) = −yf 0 (z)
dz dz

dx dy
g(x, y) + xg10 (x, y) + xg20 (x, y) = −2z
dz dz
This can be viewed as a linear equation system with dx/dz and dy/dz as the
unknowns. The determinant of the system is
y y
e xe + f (z)
 
 
D =  
g(x, y) + xg10 (x, y) xg20 (x, y) 
= xey g20 (x, y) − g(x, y) − xg10 (x, y) − f (z) g(x, y) + xg10 (x, y) ,

and by Cramer’s rule,

1  −yf 0 (z) xey + f (z) 

 
dx 1 0 0 y


= = −xyf (z)g2 (x, y) + 2z xe + f (z)
dz D  −2z xg20 (x, y)  D
and
ey −yf 0 (z) 
 
dy 1 
= 
dz D  g(x, y) + xg 0 (x, y)
1 −2z 
1
−2ey z + y g(x, y) + xg10 (x, y) f 0 (z) .


=
D
(Of course, instead of taking derivatives directly, we could have computed differ-
entials, but that would lead to almost exactly the same calculations.)

9
h03ste10 10.11.2003 24