Note on the generators of the S-convexity

phenomenon
I.M.R. Pinheiro∗

July 20, 2009

Abstract
In this paper, we deal with theorems which may be used to generate
S−convex functions. As a side dish, we explore the reasons for the S,
and not another letter or symbol, in the expression ‘S−convexity’.

AMS (2000): 26A51

Key-words: Analysis, Convexity, Definition, S-convexity, Generator, Exam-
ple, Examples, Dragomir, Pierce.

I. Introduction
Our sequence of presentation in this piece is:

• Definitions & Symbols;

• Generators;

• On the name: What is special about the S?

• Conclusions;

• References.
∗
Postal address: PO Box 12396, A’Beckett st, Melbourne, Victoria, Australia, 8006.
Electronic address: illmrpinheiro@gmail.com.

1
II. Definitions & Symbols
Symbols ([1])

• Ks1 stands for the set of S-convex classes of type 1;

• Ks2 stands for the set of S-convex classes of type 2;

• 0 < s ≤ 1 is the index designating one of the classes for S-convexity;

• K11 ≡ K12 and both classes are the same as the convex class of functions;

• s1 stands for the s value designating a class in Ks1 ;

• s2 stands for the s value designating a class in Ks2 ;

• New results will appear accompanied by a F and past results, which

we prove to be equivocated, will appear accompanied by a }.

Definitions ([1])

Definition 1. A function f : X− > < is told to be S−convex in the first

sense if
1
f (λx + (1 − λs ) s y) ≤ λs f (x) + (1 − λs )f (y) (1)
holds ∀x, y/{x, y} ⊂ X, ∀λ/λ ∈ [0, 1], where X ⊆ <+ .

Definition 2. A function f : X− > I is told to belong to Ks2 if the inequality

f (λx + (1 − λ)(x + δ))

≤ λs f (x) + (1 − λ)s f (x + δ) (2)

holds ∀λ/λ ∈ [0, 1]; ∀x/x ∈ X; s = s2 /0 < s2 ≤ 1; X/X ⊆ < ∧ I/I ⊆
<+ ∧ X = [a, b]; ∀δ/0 < δ ≤ (b − x).

Definition 3. A function f : X− > I is told to belong to Ks2 if the inequality

f (λx + (1 − λ)(x + δ))

1 1
≤ λ s f (x) + (1 − λ) s f (x + δ) (3)
holds ∀λ/λ ∈ [0, 1]; ∀x/x ∈ X; s = s2 /0 < s2 ≤ 1; X/X ⊆ < ∧ I/I ⊆
<− ∧ X = [a, b]; ∀δ/0 < δ ≤ (b − x).

2
Remark 1. If1 the inequality is obeyed in the reverse situation by f then f
is said to be s2 −concave.

Definition 4. f is called s2 −convex, s2 6= 1, if the graph lies below the ‘bent

chord’ between any two points of the curve for f , that is, for every compact
interval J ⊂ I, with I being the domain of f , J having boundary ∂J, and a
special function, of image curve L, L being of curvature Ψ, determined by
s2 ∈ (0, 1), we have:

Lx (s) ≥ sup(Lx (s) − f (x)) ≥ sup(Lx (s) − f (x)).

J ∂J

III. Generators
IIIa. What is in the scientific literature so far
From Dragomir and Pierce [2], page 283:
M) For 0 < s < 1, {a, b, c} ⊂ <, u ∈ <+ , take M to be:
f (u) = a, u = 0 and f (u) = bus + c, u > 0.
1) Generators for s1 −functions
If, in the model just mentioned, b ≥ 0 and c ≤ a then f ∈ Ks1 , which we
shall name situation A, that is, A will stand for set of conditions for which
the functional model M is a generator of examples of functions in Ks1 .
2) Generators for s2 −functions
2.1) In Dragomir and Pierce [2], p. 292, we read that if b > 0 and c < 0
then f 6∈ Ks2 ;
2.2) The same source, same page, also reveals that if b ≥ 0 and 0 ≤ c ≤ a
then f ∈ Ks2 .
3) Odd assertions: U. R2 .: Related problems found in the literature
(Problem Q) From the same source, [2], p. 283, we read that if A is found,
with b ≥ 0, but c < a, that is, not allowing c = a, then we have a non-
decreasing function in (0, ∞) but not necessarily in [0, ∞). This is a severely
odd remark, once there is mathematical proof, also provided in Dragomir
and Pierce ([2], p. 283), as to state that any function in Ks1 will fall into the
non-decreasing category for the (0, ∞) case.
Notice that if we impose continuity to the function, as a qualifying quality
for it to be S-convex, then such an assertion cannot ever be true, and that
is our case.

1
This remark applies to both definitions preceding it.
2
Our acronym for ‘under review’.

3
IIIb. New results and reviewed past results
F

Lemma 1. The sum of two S−convex functions is an S-convex function.

Proof. Assume f and g are S−convex (either sense) functions. We then

have:

f (ax + b(x + δ)) + g(ax + b(x + δ)) ≤ as (f + g)(x) + bs (f + g)(x + δ).

This implies

(f + g)(ax + b(x + δ)) ≤ as (f + g)(x) + bs (f + g)(x + δ),

that is, the sum of two S−convex functions is an S−convex function, as

wished for.
F
Lemma 2. f (u) = kus , {u, k} ⊂ <+ , 0 < s ≤ 1 is an s−convex function
(both senses).

Proof. Choose x ∈ Df , 0 < δ, (x + δ) ∈ Df , 0 ≤ a ≤ 1, and either b = 1 − a

1
or b = (1 − as ) s , depending on whether we have f ∈ Ks2 or f ∈ Ks1 , then:
f (ax + b(x + δ)) = k(ax + b(x + δ))s ≤ kas xs + kbs (x + δ)s , a.w.s.

Lemma 3. f (u) = kus + d, {u, k, d} ⊂ <+ , 0 < s ≤ 1 is an s−convex

function (both senses).

Proof. f (u) = d is an s-convex (both senses) function. Therefore, we have a

sum of s−convex functions and the first lemma trivially applies.

Theorem 4.1. f (u) = kus + d, {u, k} ⊂ <+ , d ∈ <− , 0 < s ≤ 1 is an

s1 −convex function.

Proof. f (u) = d is an s1 −convex function, so that the first lemma applies

once more.

4
Lemma 4. For every function in Ks2 , it has to be the case that the whole
set of images is either located entirely in the positive share of the counter-
domain or entirely in the negative share of it. In the case of the reals,
(f (x) ∈ <+ ) Y (f (x) ∈ <− ), ∀x ∈ Df .

Proof. The derivative of the limiting curve for Ks2 demands the whole func-
tion to belong to either the positive side of the real axis entirely or to its
negative side entirely, with no mix allowed. This because if f (x) is found in
the positive side, but f (y) is found in the negative side of the axis, one is left
with no possible value for λ, in terms of a maximum existing there, and such
is not acceptable, or compatible, with our deductions and definitions.

• On 1: It seems that all the conditions are unnecessary, once we now

hold a better theory to hang on to, that of the lemmas above, so that
this item should simply be forgotten in full. Also, notice that we have
already decided on the mandatory continuity of this sort of function,
so that it must be the case that c ≡ a at least, as stated before;

• On 2.1: It is true that we cannot guarantee that the function will

be S−convex in the second sense in case the constant c is negative
in the model M. However, if c is irrelevant enough, and our rounding
system allows (number of decimal places considered), then we may
have c < 0 and still claim we hold a function which is S−convex in the
second sense, provided both the right conditions of domain and the
right function are found;

• On 2.2: We need to have c ≡ a in order for the function to enjoy

continuity, which we have already decided is a necessary condition for
the function to be classified as s2 -convex. The rest of the item does
match our lemmas, which are stronger than it, so that we should simply
abandon it in the name of something larger in scope.

This way, one of the models for the S-convexity phenomenon may easily be
described by the theorem below:
Theorem 4.2.
0 00 0
f (u) = ko us + k1 us + k2 us + ... + d, {u, kn , d} ⊂ <+ , n ∈ N, 0 < sm ≤ 1,

where m stands for as many ’ as wished for. f (u), formed this way, will be
0
( min (s, s0 , .., sm ))-convex.

Proof. Trivial application of all the previous lemmas.

5
IV. On the name: What is special about the S?
A very well published researcher has asked us the above question. In his
mind, there should be a special reason for the S to be there. Nonetheless,
a bit of research on the subject, easily found condensed in [3], proves the
name has originated from a very logical mind: Both r− and R− convexity
had already been defined by the time of the invention of S-convexity (one of
the citations there refers to 1972. Hudzik’s article is from 1994, Breckner’s
is from 1978, therefore it is simply logical that the next sort of convexity
would be called S−convexity)!
As a general conclusion, there is nothing special connected to the choice
of the letter S for the name S−convexity, even though we could come up
with several ideas on how to make of this S something special, such as: S
has been chosen because the concept encompasses both what is under the
straight line and what is a bit above it.
S-convexity is also a specific topic in Statistics, which has got little, or noth-
ing, in common with the equally named topic in Mathematics (as possible
sources of information on the matter, see, for instance, [4]-[6]).

V. Conclusions
In this paper, we review the basic model to generate S-convex functions and
we present better, as well as more complete, alternatives to it, made out of
our new ‘generating’ theorems presented here. We also go through the issue
‘name’, that is, the why of our functions being named ‘S-convex’.

6
VI. References
[1] M.R. Pinheiro. On S-convexity and a few of our results. www.scribd.com/
illmrpinheiro2. Accessed on the 21st of July of 2009. Submitted, (2009).

[2] S.S. Dragomir, C.E.M. Pierce. Selected Topics On Hermite-Hadamard In-

equalities and Applications. RGMIA, Monographs, located at http : //rgmia.
vu.edu.au/monographs/hermitehadamard.html, and last seen on the 20th
of August of 2007; written in 2000.

[3] M. R. Pinheiro. Convexity Secrets. Trafford Publishing. ISBN: 1425138217,

(2008).

[4] C. Curtois, M. Denuit. S-convex extremal distributions with arbitrary

discrete support. Journal of Mathematical Inequalities, V. 2, nr. 2, pp. 197-
214, (2008).

[5] M. Denuit, C. Lefvre, M. Shaked. On s-Convex Approximations. Ad-

vances in Applied Probability, V. 32, nr. 4, pp. 994-1010, (Dec. 2000).

[6] M. Denuit. List of publications. http : //209.85.173.104/search?q =

cache : rKKbf F 4−mJgJ : www.actu.ucl.ac.be/staf f /denuit/P ubListDen
uit.pdf + m. + denuit + publicationshl = enct = clnkcd = 3gl = au, as seen
on the 6th of June of 2008, (2007).