1. Introduction
The paper explores sets of functions that are holomorphic in the open unit disk , normalized by and satisfy the inequality
| | |
where and . In addition to intrinsic interest, these sets appeared in the investigation of extreme points of classes of univalent functions in[8], in a relation to certain integral transforms, see[11], as well as in the study of infinitesimal generators of semigroups in[4]. For more results on different families of holomorphic functions the reader can consult the book [7]. Here we are interested in the set-theoretic structure of the family .
It appears that to investigate certain set-theoretic properties, a prerequisite understanding of Gauร hypergeometric functions is necessary. In this connection, it should be noted that in recent decades many authors have studied geometric properties of hypergeometric functions (see, for example, [1, 13, 15]). New results regarding sums of products and ratio of hypergeometric functions were established in [3, 10]. In [12], the zero-balanced hypergeometric function was applied to establishing new conditions for univalence and starlikeness of certain transforms.
Section2 considers a zero-balanced hypergeometric function . We discover its subtle characteristics as a function of . In the subsequent sections, we elaborate on an approach that capitalizes on the dependence of the hypergeometric function on its parameter.
In Section3, we concentrate on the two-parameter family which is the main object of the study in this paper. Conditions that entail/exclude the inclusion of two elements of this family into one another are derived.
The results on the inclusion relation are applied in Section4 to answer our main questions. The first one is
How to characterize all filtrations included in this family? Recall that a one-parameter family of sets is a filtration (see, for example, [2, 4, 6]) if it is ordered, more precisely, whenever .
This problem is partially addressed in [4]. In Theorem4.2 we give the complete answer.
Another question is
Is the whole family a lattice? Recall that a partially ordered family endowed with the relation islattice if each pair of elements has the unique supremum and the unique infimum.
By definition, the supremum of the pair (if it exists) is the element of denoted by such that and if for some , then . Analogously, the infimum is the element such that and the inclusion implies .
Definition4.3 introduces refined concepts: sets of quasi-infima and quasi-suprema. We give the complete description of quasi-extrema for each pair of elements of in Theorem4.4.
Furthermore, the observation below shows that if a pair has a supremum, then the quasi-supremum coincides with the supremum and so is unique. Since, according to our results, it is not the case that for every pair of elements of there is a unque quasi-supremum, we conclude:
The family is not a lattice.
In the last Section5, we pose several questions for a forthcoming investigation.
2. Some new properties of the hypergeometric function
To prove the main result of this section we need two auxiliary lemmata.
Lemma 2.1.
Let and be continuous functions defined for by the formulas
| | |
and . Then the equation has a unique solution in .
The proof of this lemma is very technical and long. For this reason, we present it in Appendix at the end of the paper.
The next assertion is a simple consequence of the theorem on integral average.
Lemma 2.2.
Let and functions satisfy
- (i)
is bounded, positive and decreasing;
- (ii)
there is such that as and as ;
- (iii)
the improper integral equals zero.
Then .
Proof.
Conditions (ii) and (iii) imply that Therefore for any there is a unique such that
| | |
and as . By the integral average theorem, there are points and such that
| | | | |
| | | | |
Thus
| | | | |
| | | | |
| | | | |
because and thanks to condition (i).โ
Choosing in this lemma , we conclude the following:
Corollary 2.3.
Let function , as for some , as , and .Then the Laplace transform is negative in .
We now turn to the Gauร hypergeometric function . Here are parameters that satisfy . Recall that this function is defined for by
| | | (2.1) |
where is the Pochhammer symbol.For geometric properties of , we refer to the useful papers [1, 13, 15] and the references therein. If , the hypergeometric function is called zero-balanced.
We now consider the following functions:
| | | (2.2) |
and
| | | (2.3) |
Theorem 2.4.
The functions and are continuous on Moreover,
- (i)
function is decreasing and maps onto and such that the function is decreasing;
- (ii)
function is decreasing and maps onto ;
- (iii)
function is increasing and maps onto ;
- (iv)
function is increasing and maps onto .
Thus, since these functions are monotone, they can be extended to and even be defined by continuity at .
Proof.
Since , function is decreasing. In addition, , so, statement (i) follows.
Further, note that which implies statement (ii).
As for function , fix arbitrary . According to Cauchyโs mean value theorem applied to the functions , there is such that
| | |
Since the function is decreasing, . Letting , we conclude that . Because the point is arbitrary, one has , or, which is the same, . Thus statement (iii) is proved.
To prove statement (iv), one has show . This inequality is equivalent to
| | | (2.4) |
Return to the integral in (2.2) defining the function and substitute there :
| | |
where is the Laplace transform. Similarly,
| | |
and
| | |
Thus takes the form
| | | | |
| | | | |
In order to calculate the convolution, we first find the primitive function:
| | |
Thus
| | | | |
and
| | | | |
| | | | |
To understand the behavior of this expression, consider functions and defined in Lemma2.1. This leads us to the relation
| | |
Lemma2.1 states that the pre-image has a unique root for . Then by Corollary2.3. So, inequality (2.4) holds, which completes the proof.โ
It is worth mentioning that Theorem2.4, in fact, presents certain properties of the values of the Gauร hypergeometric function at because functions can be expressed by it.
Corollary 2.5.
Denote . The functions and are decreasing while and are increasing on . Moreover, the following sharp estimates hold:
| | |
3. A two-parameter family and inclusion property
Denote by the set of all holomorphic functions in the open unit disk normalized by . Let .From now on we are dealing with the two-parameter family consisting of the sets
| | | (3.1) |
and .
These classes were introduced in[11], where an integral transform between different sets was established. The sets were studied even earlier in[8].Subsequently, in [4] we considered these classes with a different parametrization and found certain functions for which the sets form filtrations.
The following facts are evident.
Lemma 3.1.
For each , the set is a convex body. Moreover,
- (a)
;
- (b)
;
- (c)
if , then ;
- (d)
if , then .
An additional useful property of the classes was established in [4]:
| | |
Since our primary focus of investigation is the family equipped with inclusion as the inherent partial order, this section is devoted to the subsequent relevant problem:
Given two sets and of the family (3.1), find conditions that entail or exclude the inclusion of one of them into the other.
Since the case is covered by assertion (c) of Lemma3.1, we advance, without loss of generality, assuming that .
Theorem 3.2.
Let . Then .
Proof.
By Lemma3.1 (c), . Hence, to prove our result, it suffices to find such that as .
Let us define the function as follows
| | | (3.2) |
Formula (3.2) yields
| | | (3.3) |
Since the function maps the open unit disk onto the half-plane , we conclude that . Thus the function defined by belongs to by Lemma3.1(d).
To show that , let us consider the expression
| | |
We already know that the boundary values of equals .Since less than is arbitrary, it is enough to verify that the following claim holds:
Claim: 111It seems that formula(B18) in the book [9] implies , which contradicts our claim. In this connection we notice that the last formula is correct in the non-tangential sense only.
Indeed, function defined by (3.2) can be represented by
| | |
see (2.1). Combining this with (3.3), one concludes
| | | | |
| | | | |
| | | | |
| | | | |
Because the hypergeometric function (and hence ) can be analytically extended at any boundary point excepting , we can put in the last formula . In this case we get
| | | | |
| | | | |
| | | | |
Denote . Using this notation, we have
| | |
Using the elementary calculus tools we get
| | |
where is a bounded function.Therefore this integral tends to infinity as . So, our Claim holds, which completes the proof.โ
Thus, due to Theorem3.2, the inclusion is impossible when . We present conditions ensuring the opposite inclusion that involve function defined by(2.2).
Theorem 3.3.
Let and .
- (i)
If then inclusion holds and is sharp in the sense that whenever.
- (ii)
If , then Consequently, .
- (iii)
In addition, if , and the inclusions hold, then the inclusion is not sharp.
Proof.
By (3.1), the identity mapping belongs to all classes .Let . (So, by Lemma3.1 (a).) This function can be represented in the form . It follows from Lemma3.1 (d) that function satisfies the inequality
| | | (3.4) |
Therefore, the function defined by satisfies for all and . Then
| | |
Function being the solution of this differential equation is
| | | (3.5) |
By Harnackโs inequality,
| | |
This inequality and (3.4) imply
| | | | |
| | | | |
| | | | |
see (2.2). Thus . To show that this estimate is sharp, let us choose function in (3.5) to be and,consequently,
| | |
Setting in this equality , we obtain statement (i).
Statement (ii) follows from (i) by direct calculations.
To prove (iii), we note that by statement (ii) the given inclusions imply
| | | (3.6) |
Assume by contradiction that the inclusion is sharp. Then is equal to by statement (i). Comparing this fact with the second inequality in (3.6) gives us
| | |
Note that the coefficient of in the right-hand side is positive. Therefore, one can replace by a larger expression. Taking in mind the first inequality in (3.6) and reducing , weget
| | |
This inequality is equivalent to
| | | | |
| | | | |
| | | | |
which coincides with This contradicts statement (iv) of Theorem2.4. The proof is complete.โ
4. Filtrations and quasi-extrema
In this section, we explore the set-theoretic structures within the family of sets defined by equation (3.1). To do so, we introduce certain geometric objects tied to the outcomes of the preceding section.
Initially, let us recognize that the first statement (i) in Theorem3.3 can be interpreted as follows. Given , consider the function defined by
| | | (4.1) |
We designate its graph as the forward extremal curve for the point . Every point lying on or below this graph corresponds to the set including , while all other points correspond to sets that do not include .In addition, if , then lies below by Theorem3.3(iii).
Similarly, one can defined , the backward extremal curve for the point . This is the curve such that every point lying on or above it corresponds to the set included in , while all other points correspond to sets not included in . is the graph of the implicit function defined by
| | | (4.2) |
which is obviously well-defined and non-negative for all where is the unique solution to the equation .
In this connection the following construction is natural and quite interesting.Start from a point and let . If calculate (otherwise, we are dealing with ). Continue by setting and . At the next step, let , calculate by (4.1), and so on. Letting , we obtain the differential equation with initial point . Its solution is
| | | (4.3) |
By construction, the graph of the last function has the peculiarity: if , then . We say that this graph is the curve of infinitesimally sharp inclusions.The following result describes the relationship between the extremal curves and the curve of infinitesimally sharp inclusions.
Theorem 4.1.
Let . Then the curve of infinitesimally sharp inclusions lies below the forward extremal curve and above the backward extremal curve .
Proof.
To prove the first statement, compare the formulas (4.1) and (4.3). We need to show that the inequality
| | |
holds for all . This is equivalent to , where
| | |
Assertion (iv) of Theorem2.4 implies
| | |
Since , this proves the desired.
Regarding the second assertion, we have by (4.2). So, the inequality for means thatwhich is equivalent to , where
| | |
Since after the permutation , this function coincides with the function applied above, the proof is complete.โ
We are at the point where we can address the main problems outlined in this paper.
Let be a differentiable function. The first inquiry is:
What conditions on provide that the one-parameter family forms a filtration?
We answer it as follows.
Theorem 4.2.
Let function be differentiable on . Then is a filtration if and only if
| | | (4.4) |
Proof.
Let and analyze the function with . It follows from (4.3) that inequality (4.4) means that . Consequently, no part of the graph of can lie above the curve of infinitesimally sharp inclusions .
Take any such that . First assume that inequality (4.4) holds. Then . Therefore, by Theorems3.3 and4.1. Thus, since are arbitrary, we conclude that is a filtration.
Otherwise, assume that for some . Hence there is such that for all the inequality holds. This implies
| | |
or, which is the same, . Hence by Theorem3.3, that is, is not a filtration.โ
Now, we shift our attention to the whole family .As this family equipped with the relation constitutes a partially ordered family, our second inquiry is:
Does indeed form a lattice?
As we strive to comprehend this question, we uncover that the answer is negative, showing that the sets of so-called quasi-suprema and quasi-infima are not singletons.
Definition 4.3.
Given a pair , we say that
- โข
is a quasi-supremum of this pair and write if and there is no such that .
- โข
is a quasi-infimum of this pair and write if and there is no such that .
We are now going to describe all quasi-suprema and quasi-infima of pairs of sets defined by (3.1).
Let and the point lies on or below the forward extremal curve . Then , and so is the infimum as well as is the supremum of this pair. Therefore we need to focus on the case and .
Theorem 4.4.
Let and lie above . Then the following assertions hold:
- (a)
the set consists of such that and ;
- (b)
the set consists of such that and .
Proof.
We prove each one of the assertions by examining all points of .
We commence with (a). If then according to Theorem3.2.If and , then by Theorem3.3 either or . So, .
If and , then by Lemma3.1 and Theorem3.3.On the other hand, it follows from the above explanation that there is no such that . Thus is a quasi-supremum.
If and , then by Lemma3.1 and Theorem3.3. Hence, . Assertion (a) is proven.
Similarly to the above, if then according to Theorem3.2. If and , then either or by Theorem3.3. So, .
If and , then by Lemma3.1 and Theorem3.3.In addition, there is no such that . Thus is a quasi-infimum.
If and , then by Lemma3.1 and Theorem3.3. Hence, โ
Observe that if a pair has the supremum, then by definition . On the other hand, Definition4.3 implies that the relation is impossible. So, the quasi-supremum coincides with the supremum, in particular, it is unique.Since not for all pairs the sets and are singletons, we have:
Corollary 4.5.
The family is not a lattice.
5. Upcoming questions
In the preceding sections, we introduced an approach for establishing set-theoretic properties of a family of sets consisting of holomorphic functions. We demonstrated the effectiveness of this method with a significant example involving sets defined by (3.1). Furthermore, it turns out that this approach relies on previously established characteristics of the hypergeometric function. For this reason, it appears imperative that prior to effectively disseminating this approach, one should address the following question:
Question 1.
Expand Theorem2.4 to the case of , , or a more general hypergeometric function instead of .
An additional family that can be explored using the presented approach consists of the sets
| | |
These sets were studied in [14] within the context of geometric function theory. A recent investigation delved into the specific case where , addressing problems in filtration theory in [4] and [5]. We now pose the following questions:
Question 2.
What conditions on a function provide that the one-parameter family forms a filtration?
Question 3.
Is the family a lattice?
In the case of affirmative answer, the method of finding of the unique supremum and infimum for each pair of sets should be established. Otherwise, one asks about the sets of quasi-suprema and quasi-infima.
As for a general situation, we have already shown at the end of the previous section that if each pairs of elements of a family has the unique supremum (infimum), then the set of all quasi-suprema (quasi-infima) is a singleton. We do not know whether the converse statement is valid in general. At the same time, known examples lead us to the following
Conjecture A.
A partially ordered family is a lattice if and only if each pair of its elements has a unique quasi-supremum and a unique quasi-minimum.