On S-convexity and our IJPAM paper

I.M.R. Pinheiro∗

June 23, 2009

Abstract
In this paper, we review our results which have been published in the
prestigious academic vehicle IJPAM. The communications are a bit
sad, and we do apologize for them. Substantial part of our work will
have to be nullified regarding that paper. Most of the time the mistakes
were inherited from other people’s work, other authors, so that we are
also providing argumentation for the nullification of their results in an
indirect way.

AMS (2000): 26A42

Key-words: Analysis, Convexity, Definition, S-convexity.

I. Introduction
• Definitions & Symbols;

• Foundational theorems we make use of without mentioning ([6]);

• List of results from our IJPAM paper and due remarks;

• Actual results originating in that paper;

• Conclusion;

• 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 .

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, ∀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).

Remark 1. If1 the inequality is obeyed in the reverse situation by f then f

is said to be s2 −concave.
1
This remark applies to both definitions preceding it.

2
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. Foundational theorems we make use of without

mentioning ([6])
Theorem 4.1. If f is a bounded function defined in a closed, bounded in-
terval [a, b], and f is continuous except at countably many points, then f is
Riemann integrable.
Theorem 4.2. If f is a bounded function defined in a closed, bounded inter-
val [a, b], and f is Riemann integrable, then f is continuous in [a, b], except
possibly at countably many points.
Theorem 4.3. If g is a function defined in [a, b] such that g(x) < f (x) in
Rb Rb
[a, b], then a g(x)dx < a f (x)dx.

IV. List of results and due remarks

1. ([2], p.568) For any f : [a, b]− > <, f being convex and continuous in
[a, b], it is always true that
Z b
1 f (a) + f (b)
f (λa + (1 − λ)b) ≤ f (t)dt ≤ ,
b−a a 2
for each pre-determined value λ, λ ∈ [0, 1].
Pointed problems: We have only deduced this result because of the
proof presented in the paper of McAndrew and Dragomir, proof which
appears by the page 567 of [2]. However, that proof of theirs is ALSO
wrong and contains the same basic mistakes the other proofs, which
have been pointed by us throughout time, proofs usually associated
with the name of Dragomir, contain. Notice that, in the proof located
at the page 567 of [2], a basic problem is found: To start their proof,
Dragomir et al. make use of a severely well known consequence of the
definition of derivative of a function. However, y, there, is obviously a

3
 variable, and the theorem can only be stated with that condition, that
y is a variable. Unfortunately, by the seventh line of our reproduction
of that proof, Dragomir et al. replace a variable with a constant value,
0.5(a + b), with no sound mathematical justification for that, what
is obviously absolutely unacceptable. Because the proof in which we
have based ourselves is wrong, our result is also wrong, in principle.
In reading our result more carefully, it is indeed the case that simple
geometric reasoning tells us it should be an unacceptable result (left
side). Oh, well, if we would like to ‘pervert’ the rules of Mathematics
a bit, and assume we could write generically about derivatives over a
point, rather than deriving the function generically first then applying
the derivative function to that point, we would also have to adapt it
all for the x in the formula, trivially, what would have to mean that
x would not belong to (a, b) anymore, but to (a, 0.5(a + b)) instead,
at most, once there is an assumption that x 6= y (otherwise the ratio
would not be defined) and y > x in the derivative formula. This way,
the resulting inequality would not be what we would have hoped to
achieve.
As a plus, notice that the HH inequality, so that this is another crit-
icism to the trial of proof presented in [3], may be proved with not
much difficulty, just based on the geometric idea of the integral...there
is no need to think of anything else, then, for the basic rule of Science
is ‘making it simple if it can be’, not the opposite. This result will be
entirely nullified.

2. ([2], p.570) Let f be an almost everywhere continuous s2 −convex func-

tion obeying the following conditions: Imf ⊆ <+ , Df ⊆ I ⊆ <, {a, b} ⊂
I, a < b. Then:
Z b
s−1 1 f (a) + f (b)
2 f (0.5(a + b)) ≤ f (x)dx ≤ .
b−a a s+1

Pointed problems: In the proof provided by us for the right side of

this inequality, justified in this step because f ∈ Ks2 , we have

f ((1−λ)x+λ(x+δ)) ≤ (1−λ)s f (x)+λs f (x+δ), ∀λ ∈ [0, 1], x ∈ Df , δ > 0.

As a special instance of this inequality, once a ∈ Df , and b ∈ Df , we

may have:

f ((1 − λ)a + λb) ≤ (1 − λ)s f (a) + λs f (b), ∀λ ∈ [0, 1].

4
 There is some huge hesitation here, by time of this step, step which is
present in many deductions of Dragomir et al.: We now would like to
make of λ, previously treated by us as a constant, a variable, once we
would like to use it as an integration variable. However, if it is just
swapping its name from constant to variable, why would we not have
called it a variable at the first statement of the definition? Basically,
in the definition of Convexity, we are supposed to have x, at most
y (already proved by us to be equivocated designation), as variables,
when it comes to Cartesian graphs. Thus, as a side remark, we also
could not have changed the definition to constants a and b instead of
variables x and y or, as we have made it be, x and2 (x + δ). If λ were
a variable, we would then be obliged to have our domain at least in
some part of <2 (rigorously, before our correction of y to (x + λ), of
<3 then). Notice that f is not a function of λ, λ is used in order for
us to get a domain member for f , which will be a function of another
variable which has disappeared by means of the insertion of a,b, λ, as
well as the insertion of the results of a few operations between them.
Because λ is not the variable of f , to integrate f we need to do it
over an unknown variable, in the situation we find f in the above
inequality. Therefore, unfortunately, we cannot proceed from here as
we have suggested before, and the way it has been done in [4]. As
any other reasoning would lead to a different upper bound, and the
left side of the inequality also deserves nullification, as we will explain,
later on, in this very piece of work, we’d better nullify this result
entirely. There has also been a fundamental mistake, a typo, in our
published theorem: While our deduction involves 0 and 1 as limits for
our integral, we have used a and b in our original statement of the
definition instead.

3. ([2], p.571) For any choice of couple of elements of the domain of any
convex function, which is continuous almost everywhere, we have:
Z x2
1 f (x1 ) + f (x2 )
f (x)dx ≤ .
x 2 − x 1 x1 s+1

Pointed problems: The same problems as the previous nullified the-

orem appear here. Therefore, we should also nullify this result.

4. ([2], p.572) Via geometric deduction, for f ∈ Ks2 , f : [a, b]− > <, f
2
Well, with a lot of extra analytical work, these things might be possible, but the results
would be severely distinct from these.

5
continuous almost everywhere, we have:
Z b
1 f (0.5(a + b))
f (x)dx ≤ .
b−a a 2s−1
Pointed problems: For this result to be accurate, f needs to be non-
negative, first of all, therefore located in <+ . Now, we may remind the
reader that the maximum of the λ function will appear when λ = 0.5.
The proposal of ours in [2] is not so wrong in what is found there
described with English words, but it is atrociously wrong in what is
there described in Mathematics. Basically, we wished to refer to the
integral reasoning, therefore the instantaneous area reasoning, there-
fore we should obviously simply have restricted ourselves to picking the
supreme of the function in the interval, multiplying such a value by
two and then multiplying that result by 0.5s to get our upper bound
after due integration, once every value of the function will be less than,
or equal to, the limiting geometric line, as for the definition inequal-
ity for Ks2 . In mathematical terms, given any {w, w + δ} ⊂ (a, b),
f ∈ Ks2 , f ≥ 0, ∃ sup f (x) ∧ supf (x) = k we have:

f (aw + b(w + δ)) ≤ as f (w) + bs f (w + δ) ≤

≤ as supf (x) + bs supf (x) ≤ (as + bs )k ≤

2 × 0.5s × k.
Rb
The result just stated will imply a f (x)dx ≤ 0.5s ×2×k ×(b−a). Tak-
ing all the just exposed here into account, the right geometric deduc-
tion would have been f ∈ Ks2 , f : [a, b]− > <+ , ∃supf (x) ∧ supf (x) =
1
Rb s
k, f continuous almost everywhere =⇒ b−a a f (x)dx ≤ 0.5 × 2 × k.
Notice that all our confusional thoughts have originated by the original
confusion present in the works of Dragomir et al., which ends up affect-
ing us badly after we decide to respect the approval of all those pieces
of theory blindly only because they were found ‘published’ in academic
journals, therefore approved by several mathematicians in the World,
by entire editorial boards. It all resumes, basically, to accepting the
‘constant which may become variable’ and the ‘variable which may be-
come constant’ as a possibly sound mathematical reasoning...of course,
such is absurd, as already proven by us in public papers. We can only
apologize for this ‘temporary’ equivocated ‘acceptance’ of the unac-
ceptable. To supplement the explanation on the odd boundary, we
must also add that there is a typo in our work: Instead of f (0.5(a + b))

6
 it should have been f (a) + f (b) in our numerator, as for the previ-
ously published paper with IJPAM. Other suitable observation is that
f (x) ≤ supf (x) ≤ 0.5s × 2 × supf (x), therefore supf (x) is a better up-
per bound than the one we found, unfortunately, and f (x) ≤ supf (x)
will be true for any function, not only s2 −convex ones. Thus, this
result will simply be forgotten by now, or entirely nullified.

5. ([2], p.573) Let f be an almost everywhere continuous s2 −convex func-

tion obeying the following conditions: Imf ⊆ <+ , Df ⊆ I ⊆ <, {a, b} ⊂
I, a < b. Then:
Z b
s−1 1
2 f (0.5(a + b)) ≤ f (x)dx ≤ 21−s f (0.5(a + b)).
b−a a
Pointed problems: The right side of this inequality has already been
put down in our last item. The left side is not correct either. By vis-
iting the third page of [4], we find a series of basic mistakes. From
the first and second lines, already, departing from inequality (2.2): x
and y are variables appearing together in a simple sum, involving only
them as parcels. With Dragomir et al.’s suggestion, we actually get
x = ta + (1 − t)b and y = tb + (1 − t)a, which leads to x + y = a + b,
therefore to a single possible evaluation of the inequality (2.2), rather
than a set of them (variables), being it all unacceptable mathematically
from this point onwards (see that if we want to use the couple (1; 2),
for instance, as our (x; y) couple, in [a, b] = [1, 4], it is all hopeless).
Notice that there is the ‘insertion’ of one ‘variable’ when Dragomir
et al. supposedly simply ‘apply’ the definition of s2 −convexity to the
now ‘constant’ argument. Such has been observed before by us in their
work, and is fully unacceptable mathematically, leading to the nullifi-
cation of our previously ‘claimed-to-be’ left bound for this inequality.
Therefore, this result must be entirely nullified as well.

6. ([2], p.574) f : [a, a + λ]− > <, λ ≤ 1, f being convex, it is always true
that
1 a+λ
  Z
2a + λ
f ≤ f (t)dt ≤ (a + 0.5λ)λf (x) + (1 − 0.5λ − a)λf (y).
2 λ a

Pointed problems: This theorem has been obtained using the same
equivocated reasoning of making of λ, from the definition of S−convexity,
a ‘variable’. Another atrocious step in our proof has been re-writing,
following [4], our constants a and b as a function of λ, own a, and own

7
 b, obviously absurdity of no dimension in Analysis the way it has been
done. And the other atrocious step here present is what we have called
PROOF Z in [2], which is replacing only one of the variables in the
derivative definition with a constant value. Therefore, this result must
be entirely nullified.

7. ([2], p.576) Let f be an almost everywhere continuous s2 −convex func-

tion obeying the following conditions: Imf ⊆ <+ , Df ⊆ I ⊆ <, {a, b} ⊂
I, a < b. Then:
  Z b
a+b 1 f (a) + f (b)
f ≤ f (x)dx ≤ .
2 b−a a s+1

Pointed problems: The right side has already been opposed to by

us in this very paper. The left side, however, may be saved, for
s2 −convexity is an extension to the concept of Convexity, implying
that whatever is less than the integral involving the area under a con-
vex function curve should be less than the integral involving a larger
area, that is, the area under a s2 −convex function curve (only valid
given our update on the definition for the classes Ks2 of course, which
will make f be either non-negative or non-positive, exclusively, in the
chosen domain interval). We will then drop only the right bound
for this inequality.

8. ([2], p.578) Let f be an almost everywhere continuous s1 −convex func-

tion in an interval I ⊆ <+ and let {a, b} ⊂ I with a < b. Then:
Z b
1
f (a) ≤ f (x)dx ≤ f (b).
b−a a

Pointed problems: None. This theorem is a direct derivation of a

basic theorem from Real Analysis, found mentioned, for instance, in
[5], p. 28, Lemma 1.24.

V. Actual results originating in that paper

1. Let f be an almost everywhere continuous s2 −convex function obeying
the following conditions: Df ⊆ I ⊆ <, Imf ⊆ <+ , {a, b} ⊂ I, a < b.
Then:   Z b
a+b 1
f ≤ f (x)dx;
2 b−a a

8
 2. Let f be an almost everywhere continuous s1 −convex function in an
interval I ⊆ <+ and let {a, b} ⊂ I with a < b. Then:
Z b
1
f (a) ≤ f (x)dx ≤ f (b).
b−a a

VI. Conclusion
In this paper, we have managed to revise our published results, as for IJPAM,
in S-convexity. Our revision does imply the nullification of a large amount of
previous results, also of other researchers, given the horrible mathematical
mistakes which have remained ‘unnoticed’ by the editors of the journals
publishing them, as well as by the vast majority of the scientific community
this far.

9
VII. References
[1] M. R. Pinheiro. On S-convexity and our paper with Aequationes Mathe-
maticae. Submitted. 2009.

[2] M.R. Pinheiro. H-H Inequality for S-Convex Functions. International

Journal of Pure and Applied Mathematics, V. 44, no. 4, (2008), pp. 563-579.

[3] A. McAndrew, S. S. Dragomir. Refinements of the Hermite-Hadamard

for convex functions. Journal of Inequalities in Pure and Applied Mathe-
matics, V. 6, no. 5, Art. 140, (2005).

[4] S. S. Dragomir, S. Fitzpatrick. The Hadamard inequalities for s-convex

functions in the second sense. Demonstratio Mathematica, XXXII, no. 4,
(1999).

[5] Knapp, A. W. Basic real analysis. Birkhäuser, 2005, ISBN 0817632506,

9780817632502.

[6] Wachsmuth, B. G. Riemann Integral. Interactive Real Analysis. Found

online at http://www.mathcs.org/analysis/reals/integ/riemann.html, accessed
on the 23rd of June of 2009.

10