Abstract
We construct orthogonal trigonometric polynomials satisfying a new spectral condition and such that their -norms are bounded below and the uniform norm of their partial sums has extremely small order of growth. We obtain new results that relate the uniform norm and -norm on subspaces of the vector space of trigonometric polynomials.
Export citation and abstract BibTeX RIS
§ 1. Introduction
We begin with some necessary definitions. The sets of positive integers, integers and real numbers are denoted by , and , respectively. Given , put
Given , we consider the Fourier coefficients
Given a real number , we denote by its integer part and by the least integer such that . We write for the cardinality of a finite set . We denote the exact order of a non-zero trigonometric polynomial by . The Lebesgue measure of is denoted by .
If , then stands for the distance around the circle, that is,
Let be a real number and a subset of . We define as follows. Put
when and when . We also define for : put when is a non-empty subset of having components with respect to the topology of the circle and when .
Let . For a positive integer put
The function is called the Fejér kernel of order . We mention some properties of the Fejér kernel. If is a positive integer, then is a non-zero real positive even trigonometric polynomial, , , , , and the following inequality holds:
Let be a positive integer. Given , put
We denote by , , the th coordinate of . We write for the dimension of a finite-dimensional vector space over .
Put
Let be a positive integer. We denote by the vector space of real trigonometric polynomials of the form
and by the vector space of real trigonometric polynomials of the form
It easy to see that and are finite dimensional and that and . It is also evident that
§ 2. Some trigonometric polynomials with extremely small uniform norm
We first note the following fact. Let be a real number and a sequence of orthogonal trigonometric polynomials such that , . Then
Thus, under the assumptions above, the uniform norm of cannot have order of growth smaller than . We have the following result.
Theorem 1. Let be a real number, a positive non-increasing function, and a sequence of positive integers such that
Also, let a sequence of positive integers satisfy the condition , . Then there is a sequence of complex trigonometric polynomials such that
where is an absolute constant.
Remark 1. Theorem 1 is a generalization of a result in [1], where the corresponding assertion is obtained in the case when . The proof of Theorem 1 is based on ideas in [1]–[3].
Proof of Theorem 1. A celebrated result of Carleson and Hunt (see [4]) implies that if , and
then
where is an absolute constant. It is easy to see that . Note that if are positive integers, then
Put . We will construct trigonometric polynomials , , such that
It is obvious that this will prove the theorem. There is no loss of generality in assuming that , . Put
Note that is a positive integer and
We proceed to construct by induction trigonometric polynomials , , satisfying (2.3) and then show that (2.4) also holds. Put . It is clear that it satisfies (2.3). Assume that is a positive integer and have been constructed. We introduce the notation
Put
When , the union in (2.8) is defined as the empty set.
We claim that
Indeed, if , then and so (2.11) holds. Suppose that . Then
is not empty and
We will first find an upper bound for . Define
It is evident that
Then we have (see (2.6), (2.7))
Applying Chebyshev's inequality and (2.1), and then using the orthogonality of the polynomials and the bound , , we obtain
Let . Then
The trigonometric polynomial is non-zero since the spectra of , , are pairwise disjoint and . Thus there is an such that . Therefore and so . Hence is a non-zero real trigonometric polynomial. It is easy to see that . Put . It is obvious that
and since . Therefore there is an such that . Hence is a non-zero real trigonometric polynomial and
It follows from well-known facts about trigonometric polynomial (see, for example, [5], V. 2, Ch. X, Theorem 1.7) that the number of zeros of in (counted with multiplicities) is equal to . Note that
Therefore we can find an such that . Since there is an such that , there is an such that . Hence . Applying (2.13), we obtain that is the union of finitely many pairwise-disjoint open intervals on the circle and
Since
we have
Combining (2.7), (2.8), (2.12), (2.14) and (2.2), we obtain
To continue estimating we need the inequality
(with and ). This bound follows easily from the equality
for all and , as is easily checked. Also, using the fact that
we have
We now claim that
Two cases may occur.
1) Let . Then . Hence,
Therefore,
2) Let . Then . Therefore,
Thus (2.17) is proved.
It is not hard to see that is decreasing when , where is a real number. It follows from the inequality
that
Because
and also , , it follows that
Since when , we have
Thus,
We claim further that
Two cases may occur.
1) Let . Then . It follows from that
2) Let . Then . Therefore,
Thus (2.18) is proved.
Consequently,
This bound implies, in particular, that . Thus is the union of finitely many pairwise-disjoint open intervals on the circle .
We claim, in addition, that
We have (see (2.14), (2.2), (2.16)) that
It follows from
that
Two cases may occur.
1) Let . Then . It follows from that
2) Let . Then and so
Thus (2.20) is proved.
Put
It is obvious that the length of is and the intervals are pairwise disjoint. Consider
the complement of the subset (see (2.9)). We will find an upper bound for . It is not hard to see that is at most the number of intervals , , intersecting non-trivially. Such intervals are of two types: those contained in and those containing a boundary point of . Denote by and the numbers of intervals of the first and second types, respectively. Then (see (2.19))
Therefore,
It is clear (see (2.20)) that
Hence,
Thus,
and (2.11) is proved.
We have (see (2.10))
Let . Then (see (2.10), (1.1), (2.5))
Hence . It is clear that has the form (see (2.10))
Thus we have constructed trigonometric polynomials , , satisfying the conditions in (2.3).
We now claim that these trigonometric polynomials satisfy (2.4). Let be a positive integer. If , then (2.4) holds. Suppose that . Put
Since
the function is well defined. Let . Two cases may occur.
1) Let . Then
2) Let . Then
We will estimate the last summand. Suppose that . It follows from that . We claim that
Indeed, since
and , the inclusion (2.21) holds (see (2.6), (2.8)). Therefore (see (2.8), (2.9))
for any . We have (see (2.10) and also (1.1), (2.5))
Hence,
Therefore,
and inequality (2.4) is proved. This completes the proof of Theorem 1.
Corollary 1. Let , , be real numbers and
Then there are complex trigonometric polynomials , , such that
where depends only on , and .
§ 3. Some properties of the space of quasi-continuous functions
Let have the Fourier series , where , and
We put
where is the Rademacher system (see, for example, [6]). The norm in (3.1) is called the -norm. The closure of the set of trigonometric polynomials with respect to the -norm is called the space of quasi-continuous functions. This norm and space were introduced by Kashin and Temlyakov in [7], [8], where they obtained many interesting results on the properties of this norm. In particular, the following theorem was proved in [8] (see [8], Theorem 2.1). If is a positive integer and , where , , then
In connection with the problems of approximation theory (see the details in [8]), it is interesting to relate the - and -norms. It follows from the inequality (3.2) and a result of Grigor'ev [2] that
where is an absolute constant. Oskolkov proved (see [8]) that
where is an absolute constant. Oskolkov's example was generalized in [9]. The following question was open for a long time: do there exist subspaces , , such that , , and
for any and , where , , are absolute constants? It was shown in [10] that the answer is negative. In the present paper, we strengthen Theorem 2 in [10] (see Corollary 2 below). Our result is obtained by specifying functions of the variable considered in Theorems 1 and 2 in [10]. This is accomplished by a significant improvement of the argument in several places; in particular, Lemmas 1, 2 and 4 of [10] are refined. Note also that our Corollary 2 is stronger than results in [11]. The following theorem is central in this section.
Theorem 2. Let be a real number, a sequence of positive integers, and a subspace of such that , . Then there are trigonometric polynomials , , such that
where , , are positive absolute constants.
Proof. We need the following lemmas, of which Lemma 2 is well known. Lemma 1 can also be considered as well known, but we include a proof for completeness.
Lemma 1. Let be positive integers and a subspace of , . Then there is a non-zero such that
Proof. The case when is trivial. Suppose that and put
Since is compact, we can replace by in (3.3). Let be an element of such that . Denote by the number of coordinates of having absolute value .
We claim that . Assume the opposite, that is, . Let and be tuples of positive integers such that , , ,
If , then is empty. Put
It is clear that is a subspace of with . We have
Therefore,
Put . If is a real number and , define
Put
We claim further that
Indeed, let . Then , where . Hence and so
We have
Note that
Therefore,
Thus,
and (3.4) is proved.
Since and , we have . Hence . Note that
Consider the ball with centre and radius . Then lies on the boundary of the ball. Also, is either tangent to the boundary of at or intersects it transversally at . In both cases, there is a such that . Hence and . But this is not the case because of (3.3). This contradiction implies that , as is claimed above.
Thus , whence . Since , we have and . Therefore,
So we can take as the desired vector . The proof of Lemma 1 is complete.
Lemma 2 (see [12], p. 40, Theorem 2.5). Let be a positive integer, , , and
Then
where , , are positive absolute constants.
The proof of Lemma 3 given below is based on Lemmas 1 and 2. It was communicated to the author by Y. V. Malykhin. We notice that, when one does not specify , the conclusion of the lemma follows from results in [13].
Lemma 3. Let be a real number, a positive integer, and a subspace of such that . Then there is a trigonometric polynomial such that
where is an absolute constant.
Proof. Put . Since , the subspace is not trivial, , and . Consider the map
It is not hard to show that is linear and injective. Therefore is a subspace of and . By Lemma 1, there is a non-zero such that . Then , and (because otherwise ). Applying Lemma 2, we have
Since
we obtain
Therefore,
where is an absolute constant. Since , we have . Put . Then , and . We can take as the desired trigonometric polynomial . Lemma 3 is proved.
Lemma 4. Let a real number, a positive integer, and a subspace of such that . Then there is a trigonometric polynomial such that
where is an absolute constant.
Proof. Put . Then is a positive integer, is a subspace of and
Since , it follows from Lemma 3 that there is a trigonometric polynomial such that and
where is an absolute constant. Lemma 4 is proved.
Lemma 5. Let be a real number, , a positive integer, and . Then
Then . Note that
because . We introduce the sets ,
It is clear that .
Note that
We claim that
Since , we have
and it suffices to show that
But this inequality holds because . Thus (3.7) is proved.
Consider the sum
We have
We claim that if is a real number, then
Indeed, it suffices to apply Parseval's identity to the function
We claim further that if , then
Indeed, if , then (3.9) is trivial. Suppose that . In view of the well-known inequality when , it suffices to prove that . The last is equivalent to
Since
it is enough to show that . But this inequality holds because . Thus (3.9) is proved.
We now claim that
It suffices to prove that . Since , it is enough to show that . It is enough to show, in turn, that
Applying (3.6), we obtain
Thus the inequality (3.10) is proved.
Therefore,
when and hence (see (3.9))
Then
Applying (3.8) with (it is clear that ) and using the fact that , we have
Next, we will find a bound for . Applying (1.1), (3.5) and , we deduce (see also the definition of ) that
We finally have
Since , we obtain
Therefore,
Lemma 5 is proved.
Lemma 6. Let , , be a real number, , a positive integer, , , and
Then
where is an absolute constant.
Proof. We first claim that if the real trigonometric polynomial
satisfies , then
where is as in Lemma 5. Let the absolute value of be the largest for the , . If , then (3.11) is trivial. Suppose that . Put
Then . Hence,
We have
Applying Lemma 5, we obtain
Therefore,
Thus (see (3.12))
and (3.11) is proved.
We continue with the proof of the lemma. Put
By hypothesis, . Consider the points
Then , , , and so
Since , it follows from the celebrated theorem of Marcinkiewicz (see [5], V. II, Russian p. 54, Theorem 7.28) that
where is an absolute constant. Therefore,
It follows from the argument above that , , , that is,
Put
Then , , , and so
Again applying Marcinkiewicz' theorem, we obtain
Therefore, and so , , that is,
By Lemma 5, we have
where is an absolute constant. Lemma 6 is proved.
Lemma 7. Let be a real number, , a positive integer, , , and
Then , , .
Proof. If , , , then (3.13) holds. Suppose that is a non-zero tuple of real numbers such that (3.13) holds. Suppose for definiteness that . Put
Let be a real number. Then . By Lemma 6, we have . But this is not the case when is sufficiently large (for example, when ). We have a contradiction. Thus (3.13) holds only when , , . Lemma 7 is proved.
We continue with the proof of the theorem. Suppose that . Put
where is the absolute constant in Lemma 6 and is the absolute constant in (2.1). Then
We have
Hence is a positive integer.
We claim that
It is not hard to see that when . Hence,
Therefore,
The last inequality is equivalent to , which obviously holds. Thus (3.16) is proved.
Note that if are positive integers, then
We define
It is clear that , .
We write
where is the absolute constant in Lemma 4. Thus , are are positive absolute constants. We put
Hence , are are positive absolute constants.
We will construct trigonometric polynomials , , such that
We use induction to construct trigonometric polynomials , , satisfying the conditions in (3.20) and then show that (3.21) also holds. Suppose that . By Lemma 4, there is a trigonometric polynomial such that , . We define . Thus , have been constructed. We now proceed by induction. Suppose that is a positive integer and are given. We shall now construct . Write
Put
We claim that
Indeed, if , then
and (3.26) is proved.
Let . Then the set
is not empty and
We will first find an upper bound for . Put
It is obvious that
Applying Chebyshev's inequality and (2.1), and then using the orthogonality of the polynomials and the bound , , we obtain
Let . As in the proof of Theorem 1, one can show that is the union of finitely many pairwise-disjoint open intervals on the circle and
Taking (3.23), (3.24), (3.27) and (3.28) into account, we obtain
Applying (3.17) and (2.15) with and , we have
This bound implies, in particular, that and so . Indeed (see (3.15), (3.16)),
Thus is the union of finitely many pairwise-disjoint open intervals on the circle . We have (see (3.28), (3.17))
Put
It is obvious that the length of is and the intervals are pairwise disjoint. Consider
the complement of the subset (see (3.25)). We will find an upper bound for . It is not hard to see that is at most the number of , , intersecting non-trivially. Such intervals are of two types: those contained in and those containing a boundary point of . Denote by and the numbers of intervals of the first and second types, respectively. Then (see (3.29))
Since , we obtain
It is clear (see (3.30)) that
Therefore,
and (3.26) is proved.
We define as the linear hull (over ) of
It is obvious that is a subspace of . Since , we have
It follows from Lemma 7 that . Hence,
By Lemma 4, there is a such that , . Put . Since , we have
where , , , are real numbers. Since and , it follows from Lemma 6 that
Thus we have constructed trigonometric polynomials , , satisfying the conditions in (3.20).
We now claim that these trigonometric polynomials satisfy (3.21). Let be a positive integer. If , then
Suppose that . Put
Since
the function is well defined. Let . Two cases may occur.
1) . Then
2) . Then
We will estimate the last summand. Suppose that . We claim that
Indeed, since
and , the inclusion (3.33) is proved (see (3.22), (3.24)). Therefore (see (3.24), (3.25)), we obtain
for any . We have (see (3.31), (3.32), (1.1), (3.34)) that
Hence,
We finally have that in the second case
and (3.21) is proved. Thus we have constructed trigonometric polynomials , , satisfying (3.20) and (3.21).
It is easy to see that when . Since , we have . Hence (see (3.14)),
Let . We obtain (see (3.18))
Thus we have constructed a sequence of trigonometric polynomials such that
It is evident that the constants , and can be replaced by , and , respectively (see (3.19)).
Now suppose that . Then
It follows from the argument above that there is a sequence of trigonometric polynomials such that
for any positive integer . Theorem 2 is proved.
Theorem 3. Let be a real number, a sequence of positive integers, and a subspace of such that , . Then
where , , and are positive absolute constants and, moreover, and are the constants in Theorem 2.
Proof. By Theorem 2, there is a sequence of trigonometric polynomials such that
where , and are positive absolute constants. Let be a positive integer. Put . Then and
Since the spectra of the trigonometric polynomials , , are pairwise disjoint and , we obtain that is a non-zero real trigonometric polynomial. By the inequality (3.2),
where is a positive absolute constant. Thus
and the first inequality in (3.35) is proved. Further,
where is a positive absolute constant. Thus the second inequality in (3.35) is also proved. The proof of Theorem 3 is complete.
Corollary 2. Suppose that sequences and of real numbers sasisfy the conditions
i) , () and when ;
ii) , () and, starting at some ,
Let be a subspace of such that
Then
where is a positive absolute constant. In particular,
Proof. Let be a positive integer. Write . Then is a real number and, by condition ii), the inequality holds for any positive integer , . Put when and when .
We claim that is a subspace of and
Indeed, if , then is a subspace of and
If , then is a subspace of and
and so (3.36) is proved.
Putting , , in Theorem 3 we have
In particular,
when . Since ,
Thus,
where is a positive absolute constant. Corollary 2 is proved.
The author grateful to the referee for his careful reading of the paper.