The generalized convexity ( English generalized convexity is) a generalization of the ordinary of convexity for functions and levels that are particularly in the treatment of non-convex optimization problems proves to be useful.
Φ convexity
Given is a set and the set of all mappings from to
X
≠
∅
{\ displaystyle X \ neq \ emptyset}
X
{\ displaystyle X}
R.
{\ displaystyle \ mathbb {R}}
F.
(
X
)
=
{
φ
∣
φ
:
X
→
R.
}
{\ displaystyle F (X) = \ {\ varphi \ mid \ varphi \ colon X \ to \ mathbb {R} \}}
A set is called a reference system for if and only if:
Φ
⊆
F.
(
X
)
{\ displaystyle \ Phi \ subseteq F (X)}
X
{\ displaystyle X}
∀
φ
∈
Φ
,
λ
≥
0
:
λ
φ
∈
Φ
{\ displaystyle \ forall \ varphi \ in \ Phi, \ lambda \ geq 0: \ lambda \ varphi \ in \ Phi}
∀
φ
∈
Φ
,
c
∈
R.
:
φ
+
c
∈
Φ
{\ displaystyle \ forall \ varphi \ in \ Phi, c \ in \ mathbb {R}: \ varphi + c \ in \ Phi}
Φ-convex function
An (extended) real-valued function is called -convex if and only if there is a set such that
f
:
X
↦
R.
¯
{\ displaystyle f \ colon X \ mapsto {\ overline {\ mathbb {R}}}}
Φ
{\ displaystyle \ Phi}
Φ
0
⊂
Φ
{\ displaystyle \ Phi _ {0} \ subset \ Phi}
f
(
x
)
=
sup
φ
∈
Φ
0
φ
(
x
)
{\ displaystyle f (x) = \ sup _ {\ varphi \ in \ Phi _ {0}} \ varphi (x)}
applies.
Φ-convex set
A set is called -convex if and only if there is a set and for each
one exists such that
A.
⊆
X
{\ displaystyle A \ subseteq X}
Φ
{\ displaystyle \ Phi}
Φ
0
⊆
Φ
{\ displaystyle \ Phi _ {0} \ subseteq \ Phi}
φ
∈
Φ
0
{\ displaystyle \ varphi \ in \ Phi _ {0}}
a
φ
{\ displaystyle a _ {\ varphi}}
A.
=
⋂
φ
∈
Φ
0
{
x
∈
X
:
φ
(
x
)
≤
a
ϕ
}
{\ displaystyle A = \ bigcap _ {\ varphi \ in \ Phi _ {0}} \ {x \ in X: \ varphi (x) \ leq a _ {\ phi} \}}
Examples
For example, if one takes the affine functions as the reference system , then the convexity coincides with the usual convexity.
Φ
=
{
φ
|
φ
(
x
)
=
⟨
v
,
x
⟩
+
c
,
v
∈
R.
n
,
c
∈
R.
}
{\ displaystyle \ Phi = \ {\ varphi \, | \, \ varphi (x) = \ langle v, x \ rangle + c, v \ in \ mathbb {R} ^ {n}, c \ in \ mathbb { R} \}}
Φ
{\ displaystyle \ Phi}
The Lipschitz continuous functions are convex to the reference system of the peak functions .
Φ
=
{
φ
|
φ
(
x
)
=
-
k
⋅
d
(
x
,
x
0
)
+
c
,
x
0
∈
X
,
c
∈
R.
}
{\ displaystyle \ Phi = \ {\ varphi \, | \, \ varphi (x) = - k \ cdot d (x, x_ {0}) + c, x_ {0} \ in X, c \ in \ mathbb {R} \}}
Φ
{\ displaystyle \ Phi}
See also
literature
<img src="https://de.wikipedia.org/wiki/Special:CentralAutoLogin/start?type=1x1" alt="" title="" width="1" height="1" style="border: none; position: absolute;">