Abstract:
In this note we announce our investigation on the Lp properties for periodic and discrete multilinear pseudo-differential operators. First, we review the periodic analysis of multilinear pseudo-differential operators by showing classical multilinear Fourier multipliers theorems (proved by Coifman and Meyer, Tomita, Miyachi, Fujita, Grafakos, Tao, etc.) in the context of periodic and discrete multilinear pseudo-differential operators. For this, we use the periodic analysis of pseudo-differential operators developed by Ruzhansky and Turunen. The s-nuclearity, 0 < s ≤ 1, for the discrete and periodic multilinear pseudo-differential operators will be investigated. To do so, we classify those s-nuclear, 0 < s ≤ 1, multilinear integral operators on arbitrary Lebesgue spaces defined on σ-finite measures spaces. Finally, we present some applications of our analysis to deduce the periodic Kato-Ponce inequality and to examine the s-nuclearity of multilinear Bessel potentials as well as the s-nuclearity of periodic Fourier integral operators admitting suitable types of singularities.
MSC2010: 58J40, 47B10, 47G30, 35S30.
Keywords: Pseudo-differential operatorPseudo-differential operator,discrete operatordiscrete operator,periodic operatorperiodic operator,nuclearitynuclearity,boundednessboundedness,Fourier integral operatorFourier integral operator,multilinear analysismultilinear analysis.
Resumen: En esta nota anunciamos los resultados de nuestra investigación sobre las propiedades Lp de operadores pseudodiferenciales multilineales periódicos y/o discretos. Primero, revisaremos el análisis multilineal de tales operadores mostrando versiones análogas de los teoremas clásicos disponibles en el análisis multilineal euclidiano (debidos a Coifman y Meyer, Tomita, Miyachi, Fujita, Grafakos, Tao, etc.), pero, en el contexto de operadores periódicos y/o discretos. Se caracterizará la s-nuclearidad, 0 < s ≤ 1, para operadores multilineales pseudodiferenciales periódicos y/o discretos. Para cumplir este objetivo se clasificarán aquellos operadores lineales s-nucleares, 0 < s ≤ 1, multilineales con núcleo, sobre espacios de Lebesgue arbitrarios definidos en espacios de medida σ-finitos. Finalmente, como aplicación de los resultados presentados se obtiene la versión periódica de la desigualdad de Kato-Ponce, y se examina la s-nuclearidad de potenciales de Bessel lineales y multilineales, como también la s-nuclearidad de operadores integrales de Fourier periódicos admitiendo símbolos con tipos adecuados de singularidad.
Palabras clave: Operador pseudo-diferencial, operador discreto, operador periódico, nuclearidad, continuidad, operador integral de Fourier, Análisis multilineal.
Artículos originales
Multilinear analysis for discrete and periodic pseudo-differential operators in Lp-spaces
Análisis multilineal para operadores pseudodiferenciales periódicos y discretos en espacios Lp
Received: 24 September 2018
Accepted: 29 November 2018
The goal of this note is to announce the main results about the Lp-multilinear analysis developed by the authors in [10] for periodic and discrete pseudo-differential operators. These operators can be defined by using the multilinear Fourier transform as follows. If
is a suitable function, then the periodic multilinear-pseudo-differential operator associated to m is the operator defined as

where
and

is the periodic multilinear Fourier transform of f. We have denoted by
the space of smooth functions on the torus
endowed with its usual Fréchet structure. On the other hand, if α:
is a measurable function, then the discrete multilinear-pseudo-differential operator associated to α is the multilinear operator defined by

where
is
the discrete Fourier transform of gi. For r ≥ 2, these operators have been studied by V. Catană in [12]. If r = 1, these quantization formulae can be reduced to the known expressions

and

Periodic pseudo-differential operators (see (3)) were defined by Volevich and Agranovich [1]. The subsequent works of McLean [29], Turunen and Vainikko [47], and Ruzhansky and Turunen [44] developed a periodic analysis from Hörmander classes to applications to periodic equations, although the symbolic calculus was consistently developed by Ruzhansky and Turunen [44]. Nevertheless, the references Ruzhansky and Turunen [44], [45], Cardona [6], [7], [8], Delgado [15] and Molahajloo and Wong [34], [35], [36] provide some complementary results for the subject. Mapping properties for more general operators as periodic Fourier integral operators appear in Ruzhansky and Turunen [45] and Cardona, Messiouene and Senoussaoui [11].
In a more recent approach, pseudo-differential operators on
(discrete pseudo-differential operators) were introduced by Molahajloo in [36], and some of its properties were developed in the last years in the references [9], [17], [28], [38], [39], [40], [41], [43]. However, only the fundamental work L. Botchway, G. Kibiti, and M. Ruzhansky [5] includes properties about a discrete pseudo-differential calculus and applications to difference equations. The reference [9] discusses those relations of the theory of discrete pseudo-differential operators with important problems in number theory as the Waring problem and the hypothesis K* by Hooley.
An overview to the mapping properties for pseudo-differential operators on
provides the expected results in the discrete and periodic setting. On
these operators have the form

with
the euclidean Fourier transform of f (see Hörmander [25]). The nuclearity of pseudo-differential operators on
has been treated in Aoki and Rempala [2], [42]. In a context closely related to our work, multilinear pseudo-differential operators have been treated in Bényi, Maldonado, Naibo, and Torres, [3], [4], Michalowski, Rule and Staubach, Miyachi and Tomita [30], [31], [32], [33] and references therein. The multilinear analysis for multilinear Fourier multipliers

born with the works of Coifman and Meyer [13], [14], where the condition

for sufficiently many multi-indices α = (α1,α2, · · · , ar), was proved to be sufficient for the boundedness of Ta from
into
provided that 1/p = 1/p1 + 1/p2 + · · · + 1/pr, and 1 ≤ pi, p < ∞. A generalization for this result was obtained by Tomita in [46], where it was proved that the multilinear Hörmander condition

implies the boundedness of Ta from
into
provided that 1/p = 1/p1 + 1/p2 + · · · + 1/pr, and 1 ≤ pi, p < ∞. The case r = 1 is known as the Hörmander-Mihlin theorem. These multilinear theorems have been generalized to Hardy spaces
for suitable values of 0 < pi, p < ∞, in the works of Grafakos, Torres, Miyachi, Fujita, Tomita, Kenig, Stein, Muscalo, Thiele and Tao [19], [21], [22], [23], [24], [27], [37]. The main novelty of this work is that we provide discrete and periodic analogues for these works in the multilinear setting.
This note is organized as follows. In section 2 we provide those results on the boundedness of pseudo-differential operators on
and the torus. Later, in Section 3 we classify those s-nuclear multilinear integral operators on arbitrary σ-finite measure spaces and we apply this classification to periodic and discrete multilinear pseudo-differential operators.
In this section we explain in detail the main results of our investigation on the bounded-ness of the multilinear operators considered. Our starting point is the following multilinear version of the Stein-Weiss multiplier theorem (see Theorem 3.8 of Stein and Weiss [48]). Sometimes we denote (x,ξ) := (x, ξ1, · · · , ξr) = x · (ξ1 + · · · + ξr).
Theorem 2.1. Let 1 <p < ∞ and let α:
be a continuous bounded function. If the multilinear Fourier multiplier operator

extends to a bounded multilinear operator from
into
then the periodic multilinear Fourier multiplier

also extends to a bounded multilinear operator from
into
provided that

Moreover, there exists a positive constant Cpsuch that the following inequality holds:

Remark 2.2. Theorem 2.1 can be proved in the following way. By the density of the trigonometric polynomials, we can prove that under the conditions of this theorem, we have the estimate

where the constant C does not dependent of every trigonometric polynomial Pi. For this, we will prove that

for some positive constant cn,r,p > 0. We will assume that

Observe that by linearity, we only need to prove (10) when
for k and mi in
The main step in our proof (see Cardona and Kumar [10]) is to show (10) and how it implies (9).
With the help of the previous result we prove the following fact. We use the notation

for all
Now, we provide the following discrete version of the known result of Coifman and Meyer mentioned in the introduction.
Theorem 2.3. Let Tmbe a periodic multilinear Fourier multiplier. Under the condition

the operatorTmextends to a bounded multilinear operator from
provided that

If we consider Fourier integral operators (FIOs) with periodic phases, we can recover the following multilinear version for FIOs of the multiplier theorem of Stein and Weiss.
Theorem 2.4. Let 1 < p < ∞ and let ф be a real valued continuous function defined on
is a continuous bounded function, and the multilinear Fourier integral operator

extends to a bounded multilinear operator from
into
then the periodic multilinear Fourier integral operator

also extends to a bounded multilinear operator from
into
provided that

Moreover, there exists a positive constant CPsuch that

Now, we present some results about the boundedness of periodic multilinear pseudodifferential operators where explicit conditions on the multilinear symbols are considered.
Theorem 2.5. Let us assume thatmsatisfies the Hörmander condition of order s > 0,

Then the multilinear periodic pseudo-differential operator Tmassociated with m extends to a bounded operator from
provided that
and

Remark 2.6. The proof of Theorem 2.5 is based on a suitable Littlewood-Paley decomposition of the symbol m. Indeed, we decompose m as

We prove that by assuming (11), we can decompose the operator Tm as

where every operator Tmj is associated to the symbol mj, and we prove that the operator norm of every Tmj is less than
multiplied by a factor proportional to
We conclude our proof in [10] by observing that

The following theorem is an extension of the Coifman-Meyer result presented above in the multilinear pseudo-differential framework.
Theorem 2.7. Let us assume that m satisfies the discrete symbol inequalities

sup |A|11 A|2 ··· A|; m(xfe, ··· ,ξr )| < Cafe-H (14)
for all
Then the periodic multilinear pseudo-differential operator Tmextends to a bounded operator from
provided that

Remark 2.8. We prove Theorem 2.7 by observing that (14) implies (11). We develop this delicate argument in [10] where we use, among other things, the periodic analysis developed by Ruzhansky and Turunen.
The condition on the number of discrete derivatives in the preceding result can be relaxed if we assume regularity in x. We show it in the following theorem.
Theorem 2.9. Let Tmbe aperiodic multilinear pseudo-differential operator. If m satisfies toroidal conditions of the type

where
extends to a bounded multilinear operator from
provided that

Example 2.10. Theorem 2.7 applied to the bilinear operator

where Jsis the periodic fractional derivative operator
or the periodic Bessel potential of order
implies the (well known) periodic Kato-Ponce inequality:

where
is the Laplacian on the torus.
Boundedness of discrete multilinear pseudo-differential operators. Our main results about the boundedness of discrete multilinear pseudo-differential operators are stated as follows.
Theorem 2.11. Let
If σ satisfies the discrete inequality

for all β with
then Tσextends to a bounded operator from
provided that 1 ≤ pj ≤ p ≤ ∞, and

The following result can be derived of the previous result with r = 1 and s = p.
Corollary 2.12. Let
If σ satisfies the discrete inequality

for all β with
then Tσextends to a bounded operator from
provided that
In this section we study the s-nuclearity, 0 < s ≤ 1 of multilinear discrete and periodic pseudo-differential operators. We prove Theorem 3.1 regarding the characterization of s-nuclear multilinear operators on abstract σ-finite measure spaces, and Theorem 3.2 and Theorem 3.3 regarding the characterization of s-nuclearity of periodic and discrete pseudo-differential operators. Although these theorems are multilinear extensions of the results due to Delgado [16], Delgado and Wong [17], JamalpourBirgani [26] and Ghaemi, JamalpourBirgani and Wong [20], we can recover their results from our results by considering r = 1. In order to study these multilinear operators admitting s-nuclear extensions, we prove the following multilinear version of a result by Delgado, on the nuclearity of integral operators on Lebesgue spaces (see [16], [18]). So, in the following multilinear theorem we characterize those s-nuclear (multilinear) integral operators on arbitrary (σ-finite) measure spaces (X,μ).
Theorem 3.1. Let (Χί,μί), 1 ≤ i ≤ r and (Y, ν) be σ-finite measure spaces. Let 1 ≤ Pi,ρ < ∞, 1 ≤ i ≤ r and let
be such that
for 1 ≤ i ≤ r. Let
be a multilinear operator. Then T is a s-nuclear, 0 < s ≤ 1, operator if, and only if, there exist sequences {gn}nwith gn = (gn1,gn2,...,gnr) and
and LP(v), respectively, such that
and for all f = (f1, f2,...,fr) ∈ LP1 (μ) χ LP2 (μ2) χ ··· χ LPr (μr ) we have

for almost every y ∈ Y.
Remark 3.2. The proof of Theorem 3.1 is based on an important lemma proved in [10, Lemma 4.1]. The proof of the if part of Theorem 3.1 follows using the definition of nuclear operators, Lemma 4.1 (iv) of [10] and the fact that Lp-convergence of a sequence implies the convergence of a sequence almost everywhere.
The only if part of Theorem 3.1 is a straightforward using the part (iv) of [10, Lemma 4.1] and applications of monotone convergence theorem of B. Levi and Lebesgue dominated convergence theorem.
This criterion applied to discrete and periodic operators gives the following characterizations (for the proof we refer the reader to [10]).
Theorem 3.3.Let a
be a measurable function defined on
. The multilinear pseudo-differential operator
for all 1 ≤ i ≤ r, is a s-nuclear, 0 < s ≤ 1, operator if, and only if, the following decomposition holds:

where
are two sequences in
respectively, such that
Similarly, we can classify the s-nuclearity of periodic multilinear operators.
Theorem 3.4.Let m
be a measurable function on
Then the mutlilinear pseudo-differential operator
is a s-nuclear, 0 < s ≤ 1, operator if, and only if, there exist two sequences {gk}kwith
for 1 ≤ i ≤ r
and
respectively, such that
and

where
Now, we present the following sharp result on the s-nuclearity of periodic Fourier integral operators.
Theorem 3.5.Let us consider the real-valued function
Let us consider the Fourier integral operator

with symbol satisfying the summability condition

Then A
extends to a s-nuclear, 0 < s ≤ 1, operator from
into
provided that 1 ≤ pj < ∞, and 1 ≤ p ≤ ∞.
Remark 3.6. The proof of Theorem 3.5 follows using Theorem 3.1 by considering the function

the functional

and their estimates

and

Example 3.7.In order to illustrate the previous conditions, we consider the multilinear Bessel potential. This can be introduced as follows. Consider the periodic multilinear Laplacian denoted by

acting on
by

For r = 1, we recover the usual periodic Laplacian

The multilinear Bessel potential of order

can be defined by the Fourier analysis associated to the torus as

From the estimate

Theorem 3.5 applied to
implies that the multilinear Bessel potential
extends to a s-nuclear operator from
into
for all 1 ≤ pj < ∞ and 1 ≤ p ≤ ∞, provided that

This conclusion is sharp, in the sense that if we restrict our analysis to r = 1 and p1 = ρ = 2, the operator
extends to a s-nuclear operator on
if, and only if,
Example 3.8.Now, we consider FIOs with symbols admitting some type ofsingularity at the origin. In this general context, let us choose a sequence
Let us consider the symbol

we consider the Fourier integral operator associate to α(·, ·),

the condition
0 < ρ < n/p
implies that the periodic Fourier integral operator A extends to a s-nuclear multilinear operator from
for all 1 ≤ pj < ∞ and 1 ≤ p ≤ ∞. In fact, by Theorem 3.5, we only need to verify that

But, for every ρ > 0, this happens if, and only if, 0 < ρ < n/p.
Vishvesh Kumar thanks the Council of Scientific and Industrial Research, India, for its senior research fellowship. Duván Cardona was partially supported by Pontificia Universidad Javeriana, Mathematics Department. We would like to thank the anonymous referee for his/her careful evaluation which helped to improve the presentation of this work.
* E-mail: duvanc306@gmail.com