Econ 415/615 Mathematics for Economists Lau

Problem Set VII Fall 2021

Concave/convex functions, homogeneous functions and convex sets

1) Which of the following functions is homogeneous?

a) f ( X , Y , Z ) = X aY b Z c a, b, c  0
1 1 1
b) f ( X ,Y , Z ) = 1 + X 2Y 4 Z 4
1 1 3
c) f ( X ,Y , Z ) = X Y + Z 2
2 3

1
−
−2 −2
d) f ( X , Y ) = [0.3 X + 0.7Y ] 2

2) Determine the concavity/convexity of f ( x) = xa a0

3) Prove the following 2 results for a production function y = f ( x1 ,..., xn ) , which is homogeneous
to degree r .
1

a) The function g ( x1 ,..., xn ) = [ f ( x1 ,..., xn )] is homogeneous to degree 1.

r

b) f (0,...,0) = 0

4) Assume f ( x1 ,..., xn ) and g ( x1 ,..., xn ) are homogeneous of degrees r and s respectively.

Check whether the function h( x1 ,..., xn ) defined in the following ways, is homogeneous or not.
In the case in which it is homogeneous, determine its degree of homogeneity. Show your steps.
i) h( x1 ,..., xn ) = ln x1 + ... + ln xn
ii) h( x1 ,..., xn ) = f ( x1p ,..., xnp )
iii) h( x1 ,..., xn ) = f ( x1 ,..., xn )  g ( x1 ,..., xn )

5) Prove that {( x, y) : 0  x  1,0  y  1} is a convex set.

6) Let S be the set of points {X1 , X 2 ,..., X n } satisfying the inequalities

a11 X 1 + a12 X 2 + ... + a1n X n  b1
...
am1 X 1 + am 2 X 2 + ... + amn X n  bm
X i  0, i = 1, 2,..., n

Prove that S is a convex set.

7) Determine whether the following functions are concave or convex.

[Hint: You may use any theorems discussed in class.]
a) f ( x) = 3ex + 5x4 − ln x
b) f ( x) = ln(ln(ln(ln x)))

1
8) Determine whether the following functions are concave or convex.
a) f ( x, y) = −3x2 + 2xy − y 2 + 3x − 4 y + 1
b) f ( x, y, z) = 3e x + 5 y 4 − ln z
c) f ( x, y) = x − y − x2
d) f ( x, y, z) = x2 + 2xy + y 2 + 2xz + z 2
e) f ( x, y, z) = − max[ x, y] + z
f) f ( x, y) = x2 − y
g) f ( x, y, z) = x + y + z
h) f ( x, y, z) = ax2 + by 2 + cz 2 a, b, c  0
i) f ( x, y, z) = ln( x + y + z) x+ y+z 0
1 1 1
j) f ( x, y, z ) = ln( x + y + z )
2 2 2
x, y, z  0
1
k) f ( x, y, z ) = ax 2 − by 2
a, b, y  0
l) f ( x, y, z) = e x+ y+ z x+ y+z 0

m) f ( x, y, z) = x2 + 2xy + y 2 + 3x − 4 y + 1 + 2z + z 2 + 4 yz + 4 y 2
1
n) f ( x, y, z ) = ln z − 7e + y
x 2

1
o) f ( x) = ln( x 3 + ln x) − x 2 + 4
1
p) f ( x, y, z ) = ax 2 + by 2 a, b, y  0

1 1
9) Let S  Rn be a set with the property if X1 , X 2  S , then X 1 + X 2  S . Prove that S is not
2 2
necessarily convex by providing a counter-example.

n
10) Let Si  R n i = 1,..., n be convex sets. Show that their Cartesian product S
i =1
i = S1  S2  ...  S n

is also a convex set.

11) Let C1 , C2  R m+ n be convex sets. Let C be the set of vectors x = ( y, z), y  Rm and z  Rn such
that there exist vectors z1 and z2 with ( y, z1 )  C1. ( y, z2 )  C2 and z1 + z2 = z .
Show that C is a convex set.

2
Econ 415/615 Mathematics for Economists Lau
Problem Set VII: Solution

1a) f ( X , Y , Z ) = X aY b Z c

f ( X , Y ,  Z ) = ( X )a (Y )b ( Z )c =  a+b+c X aY b Z c homogeneous to degree a + b + c

1 1 1
b) f ( X ,Y , Z ) = 1 + X 2Y 4 Z 4 not homogeneous
1 1 1

However, X 2 Y 4 Z 4 is homogeneous to degree 1.

 f ( X , Y , Z ) is homothetic as it is a monotonic transformation of a homogeneous function.

1 1 3
c) f ( X ,Y , Z ) = X 2Y 3 + Z 2

not homogeneous

1
−
d) f ( X , Y ) = [0.3 X −2 + 0.7Y −2 ] 2

1 1
− −
f ( X , Y ) = [0.3( X ) + 0.7(Y ) ]
−2 −2 2
= { (0.3 X
−2 −2 −2
+ 0.7Y )} 2

1
−
= 1 (0.3 X −2 + 0.7Y ) −2 2
= 1 f ( X , Y )

homogeneous to degree 1

2) f ( x) = xa  f '( x) = axa−1  f "( x) = a(a − 1) xa−2

f "( x)  0 [f ( x) is a strictly convex function] if a  1
f "( x)  0 [f ( x) is a strictly concave function] if a  1
f '( x) = 0 [f ( x) is a linear function] if a = 1

3
3) y = f ( x1 ,..., xn ) is homogeneous to degree r  f ( x1 ,...,  xn ) =  r f ( x1 ,..., xn )

1
a) g ( x1 ,..., xn ) = [ f ( x1 ,..., xn )] r
1 1 1
g ( x1 ,...,  xn ) = [ f ( x1 ,...,  xn )] r = [ r f ( x1 ,..., xn )] r = [ f ( x1 ,..., xn )] r =  g ( x1 ,..., xn ) QED

b) Prove that f (0,...,0) = 0

Proof:
Let f (0,...,0) = a  0 .

y = f ( x1 ,..., xn ) is homogeneous to degree r  f ( x1 ,...,  xn ) =  r f ( x1 ,..., xn )

Choose   1  f (  0,...,   0) =  r f (0,...,0) =  r a

 f (0,...,0) = a =  r a
Since a  0   r = 1   = 1 (contradiction)

 f (0) = a = 0

4) i) h( x1 ,..., xn ) = ln x1 + ... + ln xn
h( x1 ,...,  xn ) = ln  x1 + ... + ln  xn = (ln  + ln x1 ) + ... + (ln  + ln xn ) = n ln  + ln x1 + ... + ln xn

h( x1 ,..., xn ) is not homogeneous (although it is homothetic)

ii) h( x1 ,..., xn ) = f ( x1p ,..., xnp )

h( x1 ,...,  xn ) = f [( x1 ) p ,..., ( xn ) p ] = f [ p x1p ,...,  p xnp ]
= ( p ) r f [ x1 p ,..., xn p ]  f is homogeneous to degree r

h( x1 ,..., xn ) is homogeneous to degree rp

iii) h( x1 ,..., xn ) = f ( x1 ,..., xn )  g ( x1 ,..., xn )

h( x1 ,...,  xn ) = f ( x1 ,...,  xn )  g ( x1 ,...,  xn ) =  r f ( x1 ,..., xn )   s g ( x1 ,..., xn )

=  r + s f ( x1 ,..., xn )  g ( x1 ,..., xn ) =  r + s h( x1 ,..., xn )

h( x1 ,..., xn ) is homogeneous to degree r + s

4
5) Let A  {( x, y) : 0  x  1,0  y  1} and let ( x0 , y0 ) and ( x1 , y1 )  A .
Want to show: ( xˆ, yˆ )   ( x0 , y0 ) + (1 − )( x1 , y1 ) = ( x0 + (1 − ) x1 , y 0 + (1 − ) y1 )  A

Proof:
( x 0 , y 0 )  A  0  x 0 <1,0<y 0 <1 (1)
( x , y )  A  0  x <1,0<y <1
1 1 1 1
(2)
(1) : ( )(0)   x  ( )(1), ( )(0)   y  ( )(1)  0   x 0   , 0< y 0  
0 0
(3)
(2) : (1 −  )(0)  (1 −  ) x1  (1 −  )(1), (1 −  )(0)  (1 −  ) y1  (1 −  )(1)
 0  (1 −  ) x1  1 −  , 0<(1 −  ) y1  1 −  (4)

(3) + (4) 
0 + 0   x0 + (1 −  ) x1   + (1 −  )  0  xˆ  1 and 0 + 0   y0 + (1 −  ) y1   + (1 −  )  0  yˆ  1
 ( xˆ, yˆ )  A

6) Let Si  {( x1 , x2 ,...xn ) : ai1 x1 + ... + ain xn  bi }

Clearly S  S1  ...  Sm (S is defined in the problem)
Want to show: Si is a convex set.

Let ( x10 , x20 ,..., xn0 ) and ( x11 , x12 ,...x1n )  Si

Define xˆ   x0 + (1 −  ) x1  ( x10 + (1 −  ) x11 ,..., xn0 + (1 −  ) x1n )
( x10 , x20 ,..., xn0 )  Si  ai1 x10 + ... + ain xn0  bi   ai1 x10 + ... +  ain xn0   bi (1)
( x , x ,..., x )  Si  a x + ... + a x  bi  (1 −  )a x + ... + (1 −  )a x  (1 −  )bi
1
1
1
2
1
n
1
i1 1
1
in n
1
i1 1
1
in n (2)

(1) + (2)   ai1 x10 + ... +  ain xn0 + (1 −  )ai1 x11 + ... + (1 −  ) ain x1n   bi +(1 −  )bi
  ai1 x10 + (1 −  )ai1 x11 + ... +  ain xn0 + (1 −  )ain x1n  bi
 ai1[ x10 + (1 −  ) x11 ] + ... + ain [ xn0 + (1 −  ) x1n ]  bi
 ai1 xˆ1 + ... + ain xˆn  bi  xˆ  Si  Si is a convex set

Since Si is a convex set for i = 1,..., m and S  S1  ...  Sm

 S is a convex set (intersection of convex sets is also a convex set)

7a)
e x is a convex function, x 4 is a convex function, and ln x is a concave function ( − ln x is a convex function)
sum of convex functions is a convex function  f ( x) = 3e x + 5 x 4 − ln x is a convex function

7b)
ln x is an increasing concave function,
and an increasing concave function of a concave function is also a concave function
 f ( x) = ln(ln(ln(ln x))) is a concave function

5
8a) f ( x, y) = −3x2 + 2xy − y 2 + 3x − 4 y + 1
f x = −6 x + 2 y + 3 f y = 2x − 2 y − 4
f xx = −6 f yx = 2
f xy = 2 f yy = −2
f xx f xy −6 2
H = =
f yx f yy 2 −2
H1 = −6  0 H 2 = H = 8  0  negative definite  strictly concave

8b) f ( x, y, z) = 3e x + 5 y 4 − ln z
3e x strictly convex function
5 y 4 strictly convex function
ln z strictly concave function  − ln z strictly convex function
 f ( x, y, z ) = 3e x + 5 y 4 − ln z strictly convex function (sum of striclty convex functions)

8c) f ( x, y) = x − y − x2
fx = 1− 2x f y = −1
f xx = −2 f yx = 0
f xy = 0 f yy = 0
f xx f xy −2 0
H = =
f yx f yy 0 0
H1 = −2  0 or 0  0 H 2 = H = 0  0  negative semidefinite  concave
OR
x 2 convex  − x 2 concave
x − y concave
 x − y − x 2 concave (sum of concave functions)

8d) f ( x, y, z) = x2 + 2xy + y 2 + 2xz + z 2

fx = 2x + 2 y + 2z f y = 2x + 2 y fz = 2x + 2z
f xx = 2 f yx = 2 f zx = 2
f xy = 2 f yy = 2 f zy = 0
f xz = 2 f yz = 0 f zz = 2
f xx f xy f xz 2 2 2
H = f yx f yy f yz = 2 2 0
f zx f zy f zz 2 0 2
2 2
H1 = 2  0 H2 = =0 H = −8  0  nondefinite
2 2
6
8e) f ( x, y, z) = − max[ x, y] + z
max[ x, y ] convex function  − max[ x, y ] concave function
z concave function (linear function)
 − max[ x, y ] + z concave function (sum of concave function)

8f) f ( x, y) = x2 − y
fx = 2x f y = −1
f xx = 2 f yx = 0
f xy = 0 f yy = 0
f xx f xy 2 0
H = =
f yx f yy 0 0
H1 = 2  0 or 0  0 H 2 = H = 0  0  positive semidefinite  convex
OR
x 2 convex
− y convex
 x 2 − y convex (sum of convex function)

8g) f ( x, y, z) = x + y + z
fx = 1 fy =1 fz = 1
f xx = 0 f yx = 0 f zx = 0
f xy = 0 f yy = 0 f zy = 0
f xz = 0 f yz = 0 f zz = 0
f xx f xy f xz 0 0 0
H = f yx f yy f yz = 0 0 0
f zx f zy f zz 0 0 0
0 0
H1 = 0 = 0 H2 = =0 H = 0  positive semidefinite and negative semidefinite
0 0
 concave and convex
OR
f ( x, y, z ) = x + y + z is a linear function  concave and convex

7
8h) f ( x, y, z) = ax2 + by 2 + cz 2 a, b, c  0
f x = 2ax f y = 2by f z = 2cz
f xx = 2a f yx = 0 f zx = 0
f xy = 0 f yy = 2b f zy = 0
f xz = 0 f yz = 0 f zz = 2c
f xx f xy f xz 2a 0 0
H = f yx f yy f yz = 0 2b 0
f zx f zy f zz 0 0 2c
2a 0
H1 = 2a  0 H2 = = 4ab  0 H = 8abc  0  positive definite  convex
0 2b
OR
x 2 ,y 2 ,z 2 convex function  ax 2 +by 2 +cz 2 convex (sum of convex functions)

8i) f ( x, y, z) = ln( x + y + z) x+ y+z 0

ln(.) increasing concave function
x + y + z concave function (linear function is both concave and convex)
 ln( x + y + z ) is a concave function
1 1 1
8j) f ( x, y, z ) = ln( x 2 + y 2 + z 2 ) x, y, z  0
1 1 1
2 2 2
x , y , z strictly concave function
ln(.) increasing strictly concave function
1 1 1
 ln( x 2 + y 2 + z 2 ) strictly concave function (increasing strictly concave function of strictly concave function)
1
8k) f ( x, y, z ) = ax 2 − by 2 a, b, y  0
x 2 strictly convex function
1 1
y 2 strictly concave function  − y 2 strictly convex function
1
 ax 2 − by 2 strictly convex function (sum of strictly convex function)

8l) f ( x, y, z) = e x+ y+ z x+ y+z 0
x
e increasing convex function
x + y + z convex function (increasing convex function of convex function)

8
 f ( x, y, z ) = x 2 + 2 xy + y 2 + 3x − 4 y + 1 + 2 z + z 2 + 4 yz + 4 y 2
8m)
= ( x + y) 2 + ( z + 2 y) 2 + 3 x − 4 y + 2 z + 1
( x + y ) linear  convex  ( x + y ) 2 convex
( z + 2 y ) linear  convex  ( z + 2 y ) 2 convex
3 x − 4 y + 2 z + 1 linear  convex
 f ( x, y, z ) convex
1
8n) f ( x, y, z ) = ln z − 7e + y x 2

ln z strictly concave
e x strictly convex  −e x strictly concave
1
y 2 strictly concave
 f ( x, y, z ) strictly concave
1
8o) f ( x) = ln( x + ln x) − x 2 + 4
3

ln x strictly concave
1 1
x strictly concave  ( x + ln x) strictly concave function
3 3

1
ln(.) increasing strictly concave funtion  ln( x + ln x) strictly concave function 3

x 2 strictly convex function  − x 2 strictly concave function

 f ( x) strictly concave function
1
8p) f ( x, y, z ) = ax 2 + by 2 a, b, y  0
ax 2 strictly convex function
1
by 2 strictly concave function
 f ( x, y ) can be anything, hence we have to check the Ai determinants
1
f ( x, y, z ) = ax + by 2 2
a , b, y  0
f x = 2ax, f xx = 2a, f xy = 0, f xz = 0
1 3
1 − 1 −
f y = by 2 , f yx = 0, f yy = − by 2 , f yz = 0
2 4
f z = 0, f zx = 0, f zy = 0, f zz = 0
2a 0 0
f xx f xy f xz 3
1 −
A = f yx f yy f yz = 0 − by 2
0
4
f zx f zy f zz
0 0 0
3
1 −
 A1 = 2a  0, A2 = − aby 2  0, A3 = A = 0  nondefinite
2

Note that when a convex function and a concave function are added together, 3 different cases can
happen, that is why you must check the Ai's.
9
(1) When a concave function is added to a "strongly" convex function, the resulting function
can be a convex function.
For example:
f ( x ) = e5 x − e x x  0
e5 x is a convex function
−e x is a concave function
1
Note that f '( x) = 5e5 x − e x , f "( x) = 25e5 x − e x = 25e x (e 4 x − )  0 for x  0
25
 f ( x) is a convex function

(2) When a convex function is added to a "strongly" concave function, the resulting function
can be a concave function.

(3) The most common result will be a function which is neither concave nor convex.

9)
Consider S = ( x1 ,..., x n )  R n : x i  Q, i = 1,..., n where Q = rational numbers
1 1 1 j 1 j
Clearly, if X1 = ( x11 ,..., x1n ), X 2 = ( x12 ,..., x2n )  S  X1 + X 2  S x1 + x2  Q j
2 2 2 2
But S is not convex.

Consider, X1 = (0,...,0) and X 2 = (1,...,0) , then a particular convex combination

2 2 2 2
(1 − ) X1 + X2 = ( , 0,...0)  S Q
2 2 2 2

10)
n
Let X = ( x1 ,..., xn ), Y = ( y1 ,..., yn )   Si , i.e. xi , yi  Si
i =1
n
 X + (1 −  )Y = ( x1 + (1 −  ) y1 ,..., xn + (1 −  ) yn )   Si  xi + (1 −  ) yi  Si (Si convex)
i =1
n
Hence S
i =1
i is convex.

10
11)
Let ( y, z), ( y ', z ')  C , i.e. z1 , z2 with ( y, z1 )  C1, ( y, z2 )  C2 , z1 + z2 = z ,
and z1 ', z2' with ( y, z1 ')  C1 , ( y, z2 ')  C2 , z1 '+ z2 ' = z '

Consider ( yˆ , zˆ)   ( y, z) + (1 −  )( y ', z ') = ( y + (1 −  ) y ',  z + (1 −  ) z ' ) ,   (0,1)

Now, take zˆ1   z1 + (1 −  ) z1 ', and zˆ2   z2 + (1 −  ) z2 ' .

Since C1 , C2 convex  ( yˆ , zˆi ) = ( y + (1 −  ) y ',  zi + (1 −  ) zi ')  Ci , i = 1, 2

Also zˆ1 + zˆ2 =  ( z1 + z2 ) + (1 −  )( z1 '+ z2 ') =  z + (1 −  ) z '   ( y, z) + (1 −  )( y ', z ')  C

11