Convex analysis

Renu M. R.

March 3, 2020
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2

Linear combination
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2

Linear combination α1 , α2 ∈ R.
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2

Linear combination α1 , α2 ∈ R. No restrictions.

The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2

Linear combination α1 , α2 ∈ R. No restrictions.

   
1 0.2
Let v1 = and v2 = . Linear combination?
0.5 3
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2

Linear combination α1 , α2 ∈ R. No restrictions.

   
1 0.2
Let v1 = and v2 = . Linear combination? R2 .
0.5 3
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2

Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2

Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination α1 + α2 = 1.
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2
Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination α1 + α2 = 1.

Affine combination for previous example?

The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2
Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination α1 + α2 = 1.

Affine combination for previous example? Line passing

through v1 and v2 .
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2

Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination α1 + α2 = 1.

Convex combination
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2

Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination α1 + α2 = 1.

Convex combination α1 + α2 = 1, 0 ≤ α1 , α2 ≤ 1.
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2
Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination α1 + α2 = 1.

Convex combination α1 + α2 = 1, 0 ≤ α1 , α2 ≤ 1.

Convex combination for previous example?

The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2
Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination α1 + α2 = 1.

Convex combination α1 + α2 = 1, 0 ≤ α1 , α2 ≤ 1.

Convex combination for previous example? Line segment

between v1 and v2 .
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2

Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination α1 + α2 = 1.

Convex combination α1 + α2 = 1, 0 ≤ α1 , α2 ≤ 1.

Conic combination
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2

Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination α1 + α2 = 1.

Convex combination α1 + α2 = 1, 0 ≤ α1 , α2 ≤ 1.

Conic combination α1 , α2 ≥ 0.
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2
Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination α1 + α2 = 1.

Convex combination α1 + α2 = 1, 0 ≤ α1 , α2 ≤ 1.

Conic combination α1 , α2 ≥ 0.

Conic combination?
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2
Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination α1 + α2 = 1.

Convex combination α1 + α2 = 1, 0 ≤ α1 , α2 ≤ 1.

Conic combination α1 , α2 ≥ 0.

Conic combination? Cone contained by the rays containing v1 ,

v2 .
The different ways of combining
vectors
Let v1 and v2 be two vectors in some vector space with R as the
field.

Let v = α1 v1 + α2 v2

Linear combination α1 , α2 ∈ R. No restrictions.

Affine combination α1 + α2 = 1.

Convex combination α1 + α2 = 1, 0 ≤ α1 , α2 ≤ 1.

Conic combination α1 , α2 ≥ 0.
Convex, affine sets and cone
Let C ⊂ Rn .
Convex, affine sets and cone
Let C ⊂ Rn .
C is an affine set if for any x1 , x2 ∈ C,
Convex, affine sets and cone
Let C ⊂ Rn .
C is an affine set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,
for α ∈ R.
Convex, affine sets and cone
Let C ⊂ Rn .
C is an affine set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,
for α ∈ R.

Examples?
Convex, affine sets and cone
Let C ⊂ Rn .
C is an affine set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,
for α ∈ R.

Examples?

C is a convex set if for any x1 , x2 ∈ C,

Convex, affine sets and cone
Let C ⊂ Rn .
C is an affine set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,
for α ∈ R.

Examples?

C is a convex set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,

for 0 ≤ α ≤ 1.
Convex, affine sets and cone
Let C ⊂ Rn .
C is an affine set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,
for α ∈ R.

Examples?

C is a convex set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,

for 0 ≤ α ≤ 1.

Examples?
Convex, affine sets and cone
Let C ⊂ Rn .
C is an affine set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,
for α ∈ R.

Examples?

C is a convex set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,

for 0 ≤ α ≤ 1.

Examples?

C is a cone if for any x1 , x2 ∈ C,

Convex, affine sets and cone
Let C ⊂ Rn .
C is an affine set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,
for α ∈ R.

Examples?

C is a convex set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,

for 0 ≤ α ≤ 1.

Examples?

C is a cone if for any x1 , x2 ∈ C, α1 x1 + α2 x2 ∈ C, for

α1 , α2 ≥ 0.
Convex, affine sets and cone
Let C ⊂ Rn .
C is an affine set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,
for α ∈ R.

Examples?

C is a convex set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,

for 0 ≤ α ≤ 1.

Examples?

C is a cone if for any x1 , x2 ∈ C, α1 x1 + α2 x2 ∈ C, for

α1 , α2 ≥ 0.

Examples?
Convex, affine sets and cone
Let C ⊂ Rn .
C is an affine set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,
for α ∈ R.

Examples?

C is a convex set if for any x1 , x2 ∈ C, αx1 + (1 − α)x2 ∈ C,

for 0 ≤ α ≤ 1.

Examples?

C is a cone if for any x1 , x2 ∈ C, α1 x1 + α2 x2 ∈ C, for

α1 , α2 ≥ 0.

Examples?

Note that all the above cases can be generalized for more than
two points.
Convex, affine and conic hulls
Let C be an arbitrary subset of Rn .
Convex, affine and conic hulls
Let C be an arbitrary subset of Rn .

Affine hull Smallest affine set that contains C is the affine hull of C (aff C)
Convex, affine and conic hulls
Let C be an arbitrary subset of Rn .

Affine hull Smallest affine set that contains C is the affine hull of C (aff C)
n
X n
X 
aff C = αi xi | xi ∈ C, αi = 1 .
i=1 i=1
Convex, affine and conic hulls
Let C be an arbitrary subset of Rn .

Affine hull Smallest affine set that contains C is the affine hull of C (aff C)
n
X n
X 
aff C = αi xi | xi ∈ C, αi = 1 .
i=1 i=1

Convex hull Smallest convex set that contains C (conv C)

Convex, affine and conic hulls
Let C be an arbitrary subset of Rn .

Affine hull Smallest affine set that contains C is the affine hull of C (aff C)
n
X n
X 
aff C = αi xi | xi ∈ C, αi = 1 .
i=1 i=1

Convex hull Smallest convex set that contains C (conv C)

n
X n
X 
conv C = αi xi | xi ∈ C, αi = 1, αi ≥ 0 .
i=1 i=1
Convex, affine and conic hulls
Let C be an arbitrary subset of Rn .

Affine hull Smallest affine set that contains C is the affine hull of C (aff C)
n
X n
X 
aff C = αi xi | xi ∈ C, αi = 1 .
i=1 i=1

Convex hull Smallest convex set that contains C (conv C)

n
X n
X 
conv C = αi xi | xi ∈ C, αi = 1, αi ≥ 0 .
i=1 i=1

Conic hull Set of all conic combinations of points in C

Convex, affine and conic hulls
Let C be an arbitrary subset of Rn .

Affine hull Smallest affine set that contains C is the affine hull of C (aff C)
n
X n
X 
aff C = αi xi | xi ∈ C, αi = 1 .
i=1 i=1

Convex hull Smallest convex set that contains C (conv C)

n
X n
X 
conv C = αi xi | xi ∈ C, αi = 1, αi ≥ 0 .
i=1 i=1

Conic hull Set of all conic combinations of points in C

n
X 
αi xi | xi ∈ C, αi ≥ 0 .
i=1
Convex, affine and conic hulls
Let C be an arbitrary subset of Rn .

Affine hull Smallest affine set that contains C is the affine hull of C (aff C)
n
X n
X 
aff C = αi xi | xi ∈ C, αi = 1 .
i=1 i=1

Convex hull Smallest convex set that contains C (conv C)

n
X n
X 
conv C = αi xi | xi ∈ C, αi = 1, αi ≥ 0 .
i=1 i=1

Conic hull Set of all conic combinations of points in C

n
X 
αi xi | xi ∈ C, αi ≥ 0 .
i=1
Examples?
Convex functions
Let C ⊂ Rn , be a convex set.
f : C → R,
is a convex function if for every x, y ∈ C, and 0 ≤ α ≤ 1,
Convex functions
Let C ⊂ Rn , be a convex set.
f : C → R,
is a convex function if for every x, y ∈ C, and 0 ≤ α ≤ 1,

f (αx + (1 − α)y) ≤ αf (x) + (1 − α)f (y)

Convex functions
Let C ⊂ Rn , be a convex set.
f : C → R,
is a convex function if for every x, y ∈ C, and 0 ≤ α ≤ 1,

f (αx + (1 − α)y) ≤ αf (x) + (1 − α)f (y)

f is strictly convex if a strict inequality holds for x 6= y and

0 < α < 1.
Convex functions
Let C ⊂ Rn , be a convex set.
f : C → R,
is a convex function if for every x, y ∈ C, and 0 ≤ α ≤ 1,

f (αx + (1 − α)y) ≤ αf (x) + (1 − α)f (y)

f is strictly convex if a strict inequality holds for x 6= y and

0 < α < 1.

f is concave if −f is convex.
Convex functions
Let C ⊂ Rn , be a convex set.
f : C → R,
is a convex function if for every x, y ∈ C, and 0 ≤ α ≤ 1,

f (αx + (1 − α)y) ≤ αf (x) + (1 − α)f (y)

f is strictly convex if a strict inequality holds for x 6= y and

0 < α < 1.

f is concave if −f is convex.

Examples?
Convex functions
Let C ⊂ Rn , be a convex set.
f : C → R,
is a convex function if for every x, y ∈ C, and 0 ≤ α ≤ 1,

f (αx + (1 − α)y) ≤ αf (x) + (1 − α)f (y)

f is strictly convex if a strict inequality holds for x 6= y and

0 < α < 1.

f is concave if −f is convex.

Examples?

How to check for convexity?