Abstract
In this paper we present parallel theories on constructing Wiener algebras in the bicomplex setting. With the appropriate symmetry condition, the bicomplex matrix valued case can be seen as a complex valued case and, in this matrix valued case, we make the necessary connection between bicomplex analysis and complex analysis with symmetry. We also write an application to superoscillations in this case.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
1.1 Prologue
In the present work we define and study the properties of the Wiener algebra in the framework of bicomplex numbers, i.e. \(\mathbb{B}\mathbb{C}\). Here we address \(\mathbb{B}\mathbb{C}\) questions in two ways and each, with the added the expense of doubling the dimension, reduces the problem to the complex setting. One way is through the introduction of symmetry and the other goes along idempotent decomposition. In the end, we also write an application to superoscillations in this case.
To set the framework we first recall the definition of the matrix-valued Wiener algebra.
Definition 1.1
The complex Wiener algebra \({\mathcal {W}}^{p\times p}\) consists of the matrix-valued functions of the form
with \(f_n\in {\mathbb {C}}^{p\times p}\) and \(\sum _{n\in \mathbb Z}\Vert f_n\Vert <\infty \) (where \(\Vert f_n\Vert \) denotes the operator norm of the matrix \(f_n\)) endowed with pointwise multiplication and norm
We note that in analysis quite often the Wiener algebra setting provides a convenient intermediate step, between the general \({\textbf{L}}_\infty \) case and the rational case (see Theorem 2.22 below for the latter). The algebra \({\mathcal {W}}^{p\times p}\) is an important example of Banach algebra. When \(p=1\) this algebra is commutative, and will be denoted by \({\mathcal {W}}\). Counterparts and extensions of the Wiener algebra have been studied in various other settings; see e.g. [4, 8, 27, 22, §II.1].
In the scalar case (that is, \(p=1\)), the celebrated Wiener-Lévy theorem asserts that point invertibility of an element of \(\mathcal W\) implies invertibility in the algebra itself; see [29, Lemma IIe p. 14]. We also note that in this case (\(p=1\)), the same theorem can be proved using Gelfand’s theory of commutative Banach algebras and we write the statement as follows:
Theorem 1.2
Suppose that: \(f(e^{i\theta })=\sum \nolimits _{-\infty }^{\infty }c_n\, e^{in\theta }\), with \(\sum _{-\infty }^{\infty }|c_n|<\infty \), and \(f(e^{i\theta })\ne 0\) for any \(\theta \in {\mathbb {R}}\), then there exists a sequence \(\{\gamma _n\}_{n\in {\mathbb {Z}}}\) such that:
and \(\sum _{-\infty }^{\infty }|\gamma _n|<\infty \).
We already note at this point that the above theorem yields a direct proof in the bicomplex case, which we present in Sect. 4.1. The result still holds in the matrix-valued case by considering the determinant of the function.
There exist two important subalgebras of \({\mathcal {W}}^{p\times p}\) used in spectral factorizations, namely \({\mathcal {W}}_+^{p\times p}\) and \({\mathcal {W}}_-^{p\times p}\), defined as follows:
Definition 1.3
The subalgebra \({\mathcal {W}}_+^{p\times p}\) consists of the elements \(f\in {\mathcal {W}}^{p\times p}\) for which \(f_n=0\) for \(n<0\) in the representation (1.1). Similarly, the subalgebra \({\mathcal {W}}_-^{p\times p}\) consists of the elements \(f\in \mathcal W^{p\times p}\) for which \(f_n=0\) for \(n>0\) in the representation (1.1).
A right (resp. left) spectral factorization of \(f\in \mathcal W^{p\times p}\) is a representation of the form \(f=f_-f_+\) (resp. \(f_+f_-\)) where \(f_+^{\pm 1}\) (resp. \(f_-^{\pm 1}\)) belongs to \({\mathcal {W}}_+^{p\times p}\) (resp. \({\mathcal {W}}_-^{p\times p}\)). We recall:
Theorem 1.4
Let \(f\in {\mathcal {W}}^{p\times q}\) be a function with a non-identically vanishing determinant. If \(f(e^{it})>0\) (pointwise, in the sense of positive matrices), then it admits both right and left spectral factorizations in \({\mathcal {W}}^{p\times q}\).
We note that, while \({\mathcal {W}}^{p\times p}\) is not commutative, factorization and inversion results are readily deduced from the scalar case using determinant; see e.g. [18, §2]. For completeness we also mention the sources [15, 20, 21] for more information on matrix-valued and operator-valued Wiener algebras. See in particular [21, XXIX].
We define and study the Wiener algebra in the bicomplex (\({\mathbb{B}\mathbb{C}}\)) setting in two different ways, both reducing the problem to the complex setting. One way, as in our previous paper [9], is through the introduction of a specific symmetry to complex matrices of twice the size, which allows us to construct a bridge with the bicomplex setting. The other way goes along the idempotent decomposition of bicomplex numbers. In both cases, one doubles the dimension of the underlying framework, but in a different way. The first approach, restricting a realization to be symmetric, comes to the expense of minimality, however we can write significant results in this direction as well, as this second approach is classical in \({\mathbb{B}\mathbb{C}}\) analysis.
In this paper, our research focuses on three types of results in the theory of matrix Wiener algebra on the bicomplex space:
-
(i)
Inversion Theorems,
-
(ii)
Factorization Theorems,
-
(iii)
Applications to Superoscillations.
We present our results in Sects. 1.2 and 1.3, with complete proofs written in the subsequent parts of the paper. In order to do so, we first recall some properties of \({\mathbb{B}\mathbb{C}}\), the algebra of bicomplex numbers.
1.2 \(\mathbb{B}\mathbb{C}\) and the scalar bicomplex case
We set the stage through the introduction of \({\mathbb{B}\mathbb{C}}\), in the same fashion as [17, 24, 25], the key definitions and results for the case of holomorphic functions of a bicomplex variable, with a description of the bicomplex matrix setting following in Sect. 3.
The algebra of bicomplex numbers, \({\mathbb{B}\mathbb{C}}\), is generated by i and j, two commuting imaginary units. The product of the two commuting units i and j is denoted by \( k:= i\, j\) and we note that k is a hyperbolic unit, i.e. it is a unit which squares to 1. Because of these various units in \({\mathbb{B}\mathbb{C}}\), there are several different conjugations that can be defined naturally. We will make use of these appropriate conjugations in this paper, and we refer the reader to [24, 28] for more information on bicomplex and multicomplex analysis.
\({\mathbb{B}\mathbb{C}}\) is not a division algebra: it has two distinguished zero divisors, \(\textbf{e}_1\) and \(\textbf{e}_2\),
which are idempotent, linearly independent over the reals, and mutually annihilating with respect to the bicomplex multiplication:
Just like \(\{1, j \},\) they form a basis of the complex algebra \({\mathbb{B}\mathbb{C}}\), which is called the idempotent basis. If we define the following complex variables in \({{\mathbb {C}}}(\i )\):
the \({{\mathbb {C}}}(\i )\)–idempotent representation for \(Z=z_1+j z_2\) is given by
A straightforward computation yields:
The \({{\mathbb {C}}}(\i )\)–idempotent representation is the only one for which multiplication is component-wise, as shown in the next lemma.
Lemma 1.5
The addition and multiplication of bicomplex numbers can be realized component-wise in the idempotent representation above. Specifically, if \(Z= \lambda _1\,\textbf{e}_2 + \lambda _2\,\textbf{e}_2\) and \(W= \mu _1\,\textbf{e}_1 + \mu _2\,\textbf{e}_2 \) are two bicomplex numbers, where \(a_1,a_2,b_1,b_2\in {{\mathbb {C}}}(\i )\), then
Moreover, the inverse of an invertible bicomplex number \(Z= \lambda _1\textbf{e}_1 + \lambda _2\textbf{e}_2 \) (in this case \( \lambda _1 \cdot \lambda _2 \ne 0\)) is given by
where \( \lambda _1^{-1}\) and \( \lambda _2^{-1}\) are the complex multiplicative inverses of \( \lambda _1\) and \( \lambda _2\), respectively.
One can see this also by computing directly which product on the bicomplex numbers of the form
is component wise, and one finds that the only one with this property is given by the mapping:
which corresponds to the idempotent decomposition
where \(z_1 = x_1+i x_2\) and \(z_2 = x_3+i x_4\).
A special subset of the bicomplex space is defined here:
Definition 1.6
The space of hyperbolic numbers \({\mathbb {D}}\) is defined as the subset of the bicomplex space \({\mathbb{B}\mathbb{C}}\) of elements the form \(Z=a+bk\), with a and b in \({\mathbb {R}}\).
Definition 1.7
The bicomplex number Z will be said to be positive if both \(\lambda _1\) and \(\lambda _2\) in (1.3) are positive real numbers in the idempotent representation, and we write \(Z\in {\mathbb {D}}_+\). The space \({\mathbb {D}}_+\) is called the cone of hyperbolic positive numbers.
This generalizes to a partial order on the bicomplex space that is fundamental in defining notions of positivity, as well as a hyperbolic valued norm which we use throughout our paper. Using the idempotent representation one can define three norms associated with it, see [26]. We will make use of all of these norms in our work, according to the types of properties that we wish to analyze. The first two definitions represent the Lie norm and the dual Lie norm as follows:
Definition 1.8
The dual Lie norm is, up to a factor of 2:
and
Definition 1.9
The Lie norm is:
A slightly non-standard norm is:
Definition 1.10
The hyperbolic valued norm: \(\Vert Z\Vert _{{\mathbb {D}}_+}=|\lambda _1|\textbf{e}_1+|\lambda _2|\textbf{e}_2\).
It is easy to see that \(\Vert Z\Vert _{{\mathbb {D}}_+}\in {\mathbb {D}}_+\).
Remark 1.11
We will note that if \(||Z||\le M\) then \({\mathcal {L}}(Z)\le M\) and \(\Vert Z\Vert _{{\mathbb {D}}_+}\le M\) and we will use this throughout our paper.
We will need the following results regarding the various norms:
Proposition 1.12
The norm in Definition 1.8 is sub-multiplicative, i.e.
Proof
Then, \(ZW=\lambda _1\mu _1e_1+\lambda _2\mu _2e_2\) and
\(\square \)
We will now revise notions of the unit bicomplex bi-disk and throughout this discussion, we will be associating to the bicomplex number \(Z=z_1+jz_2\) the complex pair \((z_1,z_2)\) and denote them interchangeably. We will not do so for the idempotent representation.
For \(\mathbb{B}\mathbb{C}\)-valued functions, the natural counterpart of the bi-disk in \({\mathbb {C}}^2\) is:
We note that \({\mathbb {K}}\) contains \(B(0,1)\times \left\{ 0\right\} \), and so the single variable theory is obtained as a special case. Here we discuss the boundaries and properties of this bicomplex domain \({\mathbb {K}}\).
Lemma 1.13
The distinguished boundary \(|\lambda _1|=|\lambda _2|=1\) of the bidisk in the decomposition along the idempotents corresponds to
i.e.
Proof
Since
the conditions \(|z_1\pm iz_2|=1\) are equivalent to the two conditions
Setting \(z_1=uz_2\) when \(z_2\not =0\), \(u\in {\mathbb {R}}\), we get
which is rewritten as (1.8). To conclude, note that \(\partial {\mathbb {K}}\)
\(\square \)
We will call (1.8) the distinguished boundary of \({\mathbb {K}}\) and denote it by \(\partial {\mathbb {K}}\). The definition is well posed, since this constitutes the distinguished boundary of \({\mathbb {K}}\) in \({{\mathbb {C}}}^2\) as well.
The counterpart of the unit circle can now be defined:
Definition 1.14
The bicomplex unit circle \(\partial {\mathbb {K}}\) is:
the following two lemmas show us the structure of \(\partial \mathbb K\):
Lemma 1.15
In the idempotent representation, the bicomplex unit circle becomes:
Proof
Since:
we have that \(\lambda _1\) and \(\lambda _2\), defined by (1.3), are equal to
and
\(\square \)
This bicomplex unit circle has also been defined [10, 24] in terms of hyperbolic curves, here we provide an alternate definition.
Lemma 1.16
Elements of \(\partial {\mathbb {K}}\) are invertible in \(\mathbb{B}\mathbb{C}\).
Proof
This follows from (1.10). \(\square \)
Definition 1.17
Let the bicomplex Wiener algebra, denoted \({\mathfrak {W}}\) be the set of functions \(f:\partial {\mathbb {K}} \rightarrow \mathbb{B}\mathbb{C}\), such that:
such that \(\sum _{n\in {\mathbb {Z}}} ||f_n||<\infty \), where \(||f_n||\) is the dual Lie norm in Definition 1.8.
We can also define the counterparts of \({\mathfrak {W}}_{+}\) and \({\mathfrak {W}}_{ -}\) in the usual way:
Definition 1.18
Let \({\mathfrak {W}}_+\) be the set of functions \(f:\partial {\mathbb {K}} \rightarrow \mathbb{B}\mathbb{C}\), such that:
such that \(\sum _{n\ge 0} ||f_n||<\infty \).
Then, let \({\mathfrak {W}}_-\) be the set of functions \(f:\partial {\mathbb {K}} \rightarrow \mathbb{B}\mathbb{C}\), such that:
such that \(\sum _{n\ge 0} ||f_{-n}||<\infty \).
At the end of Sect. 4 we write two corollaries that extend the Wiener algebra results in the scalar case. These theorems (Corollaries 4.6, 4.7) are a direct result of the main theorems proven in the bicomplex matrix case.
Theorem 1.19
Assume that f(Z) invertible on \(\partial {\mathbb {K}}\). Then f is invertible in the bicomplex Wiener algebra.
Theorem 1.20
Assume that \(f(Z)>0\) invertible on \(\partial {\mathbb {K}}\). Then f admits left and right spectral factorizations.
1.3 Outline of the matrix valued case
In Sect. 4 we define the Wiener algebra in the bicomplex matrix valued case as:
Definition 1.21
The bicomplex Wiener algebra \({\mathfrak {W}}^{p\times p}\) consists of the matrix-valued functions of the form
with \(Z\in \partial {\mathbb {K}}\), \(f_n\in {\mathbb{B}\mathbb{C}}^{p\times p}\) and \(\sum _{n\in {\mathbb {Z}}}||f_n||<\infty \).
Remark 1.22
Here \(||f_n||\) denotes the following extension of the dual Lie norm in Definition 1.8 to the space of bicomplex matrices \(f_n=f_{n,1}\textbf{e}_1+f_{n,2}\textbf{e}_2\):
where \(\Vert f_{n,1}\Vert \) and \(\Vert f_{n,2}\Vert \) are the operator norms as complex matrices.
This Wiener algebra is endowed with point-wise multiplication and norm
The elements of \({\mathfrak {W}}^{p\times p}_+\) are of the form
while elements of \({\mathfrak {W}}^{p\times p}_-\) are of the form
In Sect. 4 we prove an invertibility theorem in the bicomplex case (Theorem 4.2), namely if \(f\in {\mathfrak {W}}^{p\times p}\) be such that f(Z) is invertible for \(Z\in \partial {\mathbb {K}}\). Then f is invertible in \(\mathfrak W^{p\times p}\). We also prove a spectral factorization in this case (Theorem 4.4), i.e. for \(f\in {\mathfrak {W}}^{p\times p}\) be such that \(f(Z)>0\) for \(Z\in \partial {\mathbb {K}}\) we have that f admits a spectral factorization \(f(Z)=f_+(Z)f_+(Z)^*\) where \(f_+^{\pm 1}\in {\mathfrak {W}}^{p\times p}_+\).
In Sect. 4.2 we also prove an equivalence between this bicomplex Wiener algebra and one induced by a special type of symmetry, called \(\sharp \) symmetry in the complex matrix domain. We will have the same type of factorization result in this case, see Theorem 4.10. For an equivalence theorem between the two algebras see Theorem 4.11. In the same Section, in Sect. 4.3, we address the rational bicomplex case with \(\sharp \) symmetry as well.
Last but not least, as applications of these results, in Sect. 5 we introduce a theory of superoscillations in the bicomplex case.
2 Matrix-valued complex Wiener algebras
2.1 Wiener algebra and matrix symmetries
We first describe relations in the Wiener algebras induced by symmetries with respect to invertible matrices. The Wiener Algebra in the complex matrix case see [20] and [21], here we only extract the definitions and results that will be useful in the bicomplex case.
Definition 2.1
Let \(Y\in GL_n({\mathbb {C}})\) and we define the following map on the space of square complex matrices:
Remark 2.2
It is immediate that, for \(M,N\in {\mathbb {C}}^{n\times n}\) and \(\lambda \in {\mathbb {C}}\),
We now have, using the map in Definition 2.1:
Lemma 2.3
Let now \(f\in {\mathcal {W}}^{p\times p}\) be of the form (1.1). If we define \(f_Y(e^{it})=\sum \nolimits _{n\in {\mathbb {Z}}}(f_n)_Ye^{int}\) for \(Y\in GL_n({\mathbb {C}})\), it follows that
From the uniqueness of the Fourier coefficients and Theorem 1.4 one obtains the following:
Proposition 2.4
Let \(f,g\in {\mathcal {W}}^{p\times p}\) and \(Y\in GL_n({\mathbb {C}})\). We have:
Furthermore, \(f\in {\mathcal {W}}_+^{p\times p}\) if and only if \(f_Y\in {\mathcal {W}}_+^{p\times p}\).
Definition 2.5
The function \(f\in {\mathcal {W}}^{p\times p}\) is called Y-symmetric if \(f(e^{it})=f_Y(e^{it})\), \(t\in [0,2\pi ]\).
From uniqueness of the Fourier coefficients we obtain:
Proposition 2.6
The function \(f\in {\mathcal {W}}^{p\times p}\) is Y-symmetric if and only if its Fourier coefficients are Y-symmetric:
Theorem 2.7
Let \(f\in {\mathcal {W}}^{p\times p}\) be Y-symmetric for some \(Y\in {\mathbb {C}}^{p \times p}\) and such that \(f(e^{it})>0\) for \(t\in [0,2\pi ]\). Then, there is a uniquely determined spectral factorization \(f=f_+f_+^*\) where \(f_+\in {\mathcal {W}}_+^{p\times p}\) is Y-symmetric and satisfies \(f_+(1)=I_{p}\) and invertible and any other spectral factorization differs by an Y-symmetric unitary multiplicative constant on the right from \(f_+\).
Proof
We leave the proof to the reader, since it is very similar to the proof of theorem 2.16 and uses parallel techniques. \(\square \)
This above Y symmetry does not apply for the more general, non-square case. We introduce another symmetry that will overcome this limitation.
2.2 Wiener algebra with special symmetry
We now turn towards a type of symmetry that will be useful in the bicomplex case and could be applied to the non-square case as well.
Let \(J\in {\mathbb {C}}^{2\times 2}\) be a matrix such that \(J J^*=I_2\).
Definition 2.8
Let \(M\in {\mathbb {C}}^{2\times 2}\). We define the \(\sharp \)-conjugate with respect to J to be:
More generally, given \(M\in {\mathbb {C}}^{2a\times 2b}\) a block matrix, we define
with
and where \(\otimes \) denotes the Kronecker (tensor) product of matrices.
Proposition 2.9
Let \(M\in {\mathbb {C}}^{2p\times 2q}\) and \(N\in {\mathbb {C}}^{2q\times 2r}\) be matrices of possibly different sizes Then
Let \(M,N\in {\mathbb {C}}^{2p\times 2q}\). Then,
Definition 2.10
For \(f\in {\mathcal {W}}^{2p\times 2p}\) of the form (1.1), we define its \(\sharp -\) symmetric function to be:
Remark 2.11
It is easy to see that:
therefore \(f^{\sharp }\in {\mathcal {W}}^{2p\times 2p}\) of the form (1.1).
The counterpart of Proposition 2.4 is:
Proposition 2.12
Let \(p,q,s\in {\mathbb {N}}\) and let \(f\in {\mathcal {W}}^{2p\times 2q}\) and \(g\in {\mathcal {W}}^{2q\times 2s}\). We have:
Furthermore, \(f\in {\mathcal {W}}_+^{2p\times 2q}\) if and only if \(f^\sharp \in {\mathcal {W}}_+^{2p\times 2q}\).
Definition 2.13
The matrix \(M\in {\mathbb {C}}^{2\times 2}\) is called \(\sharp \)-symmetric with respect to J if and only if it satisfies \(M=M^\sharp \). More generally the block matrix \(M\in {\mathbb {C}}^{2a\times 2b}\) is called \(\sharp \)-symmetric with respect to J if \(M=M^\sharp \). This will hold if and only if all its \({\mathbb {C}}^{2\times 2}\) block entries are \(\sharp \)-symmetric with respect to J.
Definition 2.14
The function \(f\in {\mathcal {W}}^{2p\times 2p}\) is called \(\sharp \)-symmetric if \(f(e^{it})=f^\sharp (e^{it})\), \(t\in [0,2\pi ]\).
From the uniqueness of the Fourier coefficients we obtain:
Proposition 2.15
The function \(f\in {\mathcal {W}}^{2p\times 2p}\) is \(\sharp \)-symmetric if and only if its Fourier coefficients are \(\sharp \)-symmetric:
Theorem 2.16
Let \(f\in {\mathcal {W}}^{2p\times 2p}\) be \(\sharp \)-symmetric and such that \(f(e^{it})>0\) for \(t\in [0,2\pi ]\). Then, there is a uniquely determined spectral factorization \(f=f_+f_+^*\) where \(f_+\in \mathcal W_+^{2p\times 2p}\) is \(\sharp \)-symmetric and satisfies \(f_+(1)=I_{p}\) and invertible and any other spectral factorization differs by an \(\sharp \)-symmetric unitary multiplicative constant on the right from \(f_+\).
Proof
Using Proposition 2.4 we can write:
and
By uniqueness up to left multiplicative factor of the spectral factor we have
and so
for some \(U\in GL_p({\mathbb {C}})\) which is moreover unitary. To show that \(U=I_n\) remark that \(f_+(0)\) is invertible and so, with \(f_+(z)=\sum _{n=0}^\infty a_nz^n\) we have \(a_0\in GL_p({\mathbb {C}})\). Setting \(z=0\) in (2.9) leads to \(a_0=a_0U\) and so \(U=I_p\) since \(a_0\) is invertible. \(\square \)
In the case of the bicomplex algebra, as seen in [9] we have an important structural matrix:
which has the obvious property that \(J\, J^*=I_2\).
Definition 2.17
The matrix \(M\in {\mathbb {C}}^{2\times 2}\) is called bicomplex \(\sharp \)-symmetric if it satisfies \(M=M^\sharp \), with respect to J defined in (2.10) above.
As in [9] we have:
Remark 2.18
For J as in (2.10), we have that M is J-symmetric if and only if it is of the form
which, in Sect. 4.2 will be discussed at length in relation with bicomplex analysis. In this decomposition, the matrix J becomes the matricial counterpart of the bicomplex unit j and this symmetry is fundamental to the definition of bicomplex numbers.
Remark 2.19
As we will see in Sect. 3.3, a bicomplex number can be viewed as a matrix of the form (2.11) (this gives in fact an explicit construction of the bicomplex numbers in terms of complex matrices). Bicomplex matrices will be identified with \(\sharp \)-symmetric complex ones of the appropriate dimension. This allows us to follow the strategy introduced in our previous work [9], where various results are reduced to ones in a complex Wiener algebra with symmetry. Note that we will also use the second approach, splitting the problem into two classical complex ones via the idempotent representation.
2.3 Rational functions and the Wiener algebra
In the general theory of rational matrix-valued functions, we can speak about several types of realizations. We recall results that pertains to our work in the bicomplex case.
Theorem 2.20
Every matrix-valued (say \({\mathbb {C}}^{a\times b}\)-valued) rational function which is analytic at infinity admits a realization centered at infinity, i.e. can be written in the form:
Remark 2.21
If f is not analytic at infinity, then it has an additional polynomial part P(z) which will not be of this form, i.e.:
The realization is called minimal if N is minimal. We write a proof of Theorem 2.20 which gives the existence of a (possibly non minimal) realization; this sequence of arguments will allow us to sketch similar proofs in the subsequent parts of the paper, where we need specific symmetries arising in the bicomplex case. Theorems 2.23 and 2.24 make use of these arguments, which in turn prove the bicomplex rational case in Theorems 4.12 and 4.13.
Proof of Theorem 2.20
Let \(p_1,\ldots , p_M\) be the poles of f (in the most elementary sense, meaning points where any entry of f has a non-removable singularity). At each of these points f has a finite Laurent expansion and we can write (since f is assumed analytic at infinity)
where D and the coefficients \(H_{k,m}\) belong to \(\mathbb C^{a\times b}\). We now write
where
and
where \(A_m\in {{\mathbb {C}}}^{k_m a\times k_m a}\). Thus neither the realization of \(f_m\) nor the realization of f (given by \(f(z)=D+C(I_N-A)^{-1}B\)) are necessarily minimal, where
\(\square \)
We recall the following result (see [19] for more details and information). Note that we do not require the realization to be minimal as the proof in [19] still holds since A is assumed to have its spectrum off the unit circle. The realization is then called regular on the unit circle.
Theorem 2.22
Let f be a \({\mathbb {C}}^{p\times p}\)-valued rational function. If f has no singularities on the unit circle, then \(f\in \mathcal W^{p\times p}\). Let \(f(z)=D+C(zI_N-A)^{-1}B\) be a realization of f, possibly not minimal but with no spectrum on the unit circle, and let P denote the Riesz projection corresponding to the spectrum of A outside the closed unit disk:
Then \(f(z)=\sum _{n\in {\mathbb {Z}}}f_nz^n\) with
and \(\sum _{{\mathbb {Z}}} \Vert f_n\Vert <\infty \).
We now turn to the \(\sharp \) symmetric case. Let us consider \(J\in {\mathbb {C}}^{2\times 2}\) as above, with \(J\, J^*=I_2\).
Theorem 2.23
Let f(z) be a \({\mathbb {C}}^{2p\times 2p}\)-valued rational function, analytic at infinity, \(\sharp \)-symmetric with respect to J, and regular on the unit circle. Then f admits a \(\sharp \)-symmetric realization, analytic at infinity, and regular on the unit circle.
Proof
Going back to the construction in the proof of Theorem 2.20 we see that the matrix coefficients \(H_{k,m}\) are \(\sharp \)-symmetric, and so is the corresponding realization as written in Theorem 2.20. We note that the Riesz projection P obtained in Theorem 2.22 is also \(\sharp \)-symmetric. \(\square \)
Theorem 2.24
Let \(f\in {\mathcal {W}}^{2p\times 2p}\) rational and taking strictly positive values and \(\sharp \) symmetric. Then it admits rational \(\sharp \)-symmetric left and right factorizations.
Proof
This follows from the fact that the a \(\sharp \)-symmetric rational function admits a sharp-symmetric realization. \(\square \)
Assuming \(f(e^{it})>0\), one can write \(f(z)=w(z)(w(1/\overline{z}))^*\) where the spectral factor w(z) is rational. For the following, see also [5, Lemma 1.2, p. 145] where the realization is centered at the origin. The computations appear also in [6].
Proposition 2.25
Let
be a realization of the left spectral factor of f(z) with both \(\sigma (a)\) and \(\sigma (a-bd^{-1}c)\) inside the open unit disk. Then,
where
The Riesz projection P from (2.17) becomes:
where X is the unique solution to the Stein equation:
We now give formulas for the Fourier coefficients in terms of a, b, c and d.
Proposition 2.26
The Fourier coefficients of the spectral function are given by
in terms of a realization (2.25) of the spectral factor.
Using Definition 2.8, one can re-write Propositions 2.25 and 2.26 in terms of the \(\sharp \) conjugate as well.
Remark 2.27
Let us assume that \(f(e^{it})>0\) is also \(\sharp \)-symmetric and the dimensions are even. It follows that the spectral factor w(z), which yields the decomposition of f i.e. \(f(z)=w(z)(w(1/\overline{z}))^*\), is rational and \(\sharp \) symmetric. In this context, in Propositions 2.25 and 2.26 all factors a, b, c, d as well as A, B, C, D will be \(\sharp \) symmetric. The Riesz Projector P will be \(\sharp \)-symmetric as well.
The results in this section will be interpreted in Sect. 4 in the setting of the \(\mathbb{B}\mathbb{C}\) Wiener algebra.
2.4 Wiener algebras and classical Superoscillations
The notion of superoscillations originate with the works of Aharonov and Berry and the notion of weak measurements. We refer to the papers [1,2,3, 13,14,15]. We refer to [11, 12, 16] for recent developments and to the paper [7] for recent connections with Schur analysis.
Definition 2.28
A superoscillatory sequence is a sequence of complex-valued functions \(\{ F_m(t,a)\}_{m\ge 0}\) defined on \({\mathbb {R}}\) as follows:
where:
with \(a>1\).
One can then see that for fixed \(t\in {\mathbb {R}}\) we have:
and convergence is uniform on compact subsets of the real line. We note that:
Lemma 2.29
Let \(z=\rho e^{i\theta }\) with \(\theta \in (-\pi ,\pi )\), i.e. \(\displaystyle z\in \Omega = {\mathbb {C}}\setminus (-\infty ,0]\), and define for \(c\in {\mathbb {C}}\)
The function
coincides with \(F_m(t,a)\) for \(t\in (-\pi ,\pi )\), and \(z^mG_m(z^m,a)\in {\mathcal {W}}_+\).
Proof
The first claim is clear and the second follows from the equality
\(\square \)
The sequence \(F_m(t,a)\) plays an important role in the respective Wiener algebra setting. Let us now turn to the matrix-valued Wiener algebra \({\mathcal {W}}^{p\times p}\) case. We will show that the superoscillatory sequence \(F_m(t,a)\) provides a good approximation in this case as well. We have the following result which will expand to the \(\mathbb{B}\mathbb{C}\) as well.
Theorem 2.30
Let \(f\in {\mathcal {W}}^{p\times p}\), then f can be uniformly approximated on compact subsets of \({\mathbb {R}}\) by sums of the form
Proof
Let \(\epsilon >0\) and \(f\in {\mathcal {W}}^{p\times p}\). If we write
there exists \(N\in {\mathbb {N}}\) such that
and by (2.23) there exists M such that
Thus for \(m\ge M\)
and so
\(\square \)
In preparation of the following two results, recall that a (say \({\mathbb {C}}^{p\times q}\)-function analytic in the open unit disk and with a real positive part there is called Carathéodory function, and that a \({\mathbb {C}}^{p\times q}\)-valued function analytic and contractive in the open unit disk is called a Schur function. When \(p=q\), and given a Schur function S, the function \(I_p-S\) is a Carathéodory function.
Theorem 2.31
Let \(\varphi \) be a \({\mathbb {C}}^{p\times p}\)-valued Carathéodory function belonging to the Wiener algebra and such that \(\textrm{Re}\, \varphi (e^{it})\ge \epsilon _0 I_p\) for some \(\epsilon _0>0\). Then, \(\varphi \) can be approximated uniformly on compact subsets of the real line by functions \(\varphi _N\) which are finite linear combinations of functions of the form \(F_m(t,n)\) with the following property: these functions are not in the Wiener algebra, but for some integer \(m_0\), the functions \(z^{m_0}\varphi _N(z^{m_0})\) are in \({\mathcal {W}}_+^{p\times p}\).
Proof
The proof is a slight adaptation of the proof of Theorem 2.30. We set
where the \(f_n\in {\mathbb {C}}^{p\times p}\) satisfy \(\sum _{n=2}^\infty \Vert f_n\Vert <\infty \). We first choose N such that
and fix a compact subset, say K, of the real line. For such a N we choose \(m_0\) such that for \(m\ge m_0\),
With \(K=[-\pi ,\pi ]\) we define:
With \(G_m\) defined by (2.24) and for every \(t\in (-\pi ,\pi )\) we have
We have for \(t\in (-\pi ,\pi )\)
It follows that, still for \(t\in (-\pi ,\pi )\),
When \(p=1\) the triangle inequality implies that
Thus, with \(m_0\) defined by (2.25) (for \(K=[-\pi ,\pi ]\)), we have that \(z^{m_0}\varphi _N(z^{m_0})\) is in \({\mathcal {W}}_+\). The case of matrix-valued function is adapted as follows. For every \(c\in {\mathbb {C}}^p\) we have
Hence
The proof follows since \(\frac{cc^*\epsilon _0}{4}\le c^*\textrm{Re}\, \varphi (e^{it})c\). \(\square \)
Remark 2.32
A similar result holds for Schur functions belonging to the Wiener algebra and strictly contractive on the unit circle. The proof reduces to the case of Carathéodory functions in two steps. The function is then \({\mathbb {C}}^{r\times r}\)-valued with \(r=\max \left\{ p,q\right\} \), and first we can assume that the Schur function S is square (by adding a number of columns or rows with entries equal to 0). One can then consider the function \(\varphi (e^{it})=I_r-S(e^{it})\). These ideas will be discussed in future works.
In what follows, we extend these results to the bicomplex case using the intrinsic structure of its algebra and analysis.
3 Bicomplex analysis and symmetry domains
In this section we revisit concepts of bicomplex analysis and write the appropriate symmetry domains that will allow us to introduce a bicomplex Wiener algebra in this case.
3.1 Bicomplex algebra as \(2\times 2\) complex matrices
As described in [9], in a way similar to the space of complex numbers and the Pauli model, one can re-write the space of bicomplex numbers as a subspace of complex matrices \({\mathbb {C}}^{2\times 2}\) as:
where the complex unit in \({\mathbb {C}}\) is denoted by i. We write
with J as in (2.10). We will use the shorter notation \(Z=z_1+jz_2\) to denote a bicomplex number. For example, in this writing, we have that the original complex unit i becomes:
Remark 3.1
In the same notation as in Sect. 1.2 one obtains the corresponding matrix to the hyperbolic unit \(k=i\,j,\) (i.e. \(k^2=1\)) as \(K=i J\).
In the matrix notation k is represented by:
and one can easily check that the square of this matrix is the identity matrix.
In the matricial writing of a bicomplex number \(Z\in {\mathbb{B}\mathbb{C}}\), we make use of the following primary decomposition and the fact that a normal matrix is unitary diagonalizable to obtain:
The unitary and hermitian diagonalization matrix is:
and the eigenvalues are:
Remark 3.2
In this commutative setting we have the same zero divisors \(\textbf{e}_1,\,\textbf{e}_2\) as in Sect. 1.2, given by the diagonalization. As expected, these split the bicomplex space in \({\mathbb{B}\mathbb{C}}={\mathbb {C}} {\textbf{e}}_1\bigoplus {\mathbb {C}} {\textbf{e}}_2\), as:
Using normality again, the corresponding matrix form of a bicomplex number can be written as the associated sum of two weighted orthogonal projections
where the eigenvalues \(\lambda _{1,2}\) are given in (3.3) and the orthogonal projections are
It is easily seen that \(E_1\) and \(E_2\) are the corresponding matrix forms of \( {\textbf{e}}_1\) and \( {\textbf{e}}_2\).
We will now define the conjugates in the bicomplex setting, as in [17, 24].
Definition 3.3
For any \(Z\in {\mathbb{B}\mathbb{C}}\) we have the following three conjugates:
Remark 3.4
In the matrix form the space of hyperbolic numbers \({\mathbb {D}}\) is realized as the subset of matrices (3.1) with:
As in [10], one can define the corresponding notions of positivity:
Definition 3.5
A matrix \(M\in {\mathbb{B}\mathbb{C}}^{n\times n}\) is called Hermitian if \(M=M^*\) and positive if \(c^*Mc\ge 0\) for all \(c\in \mathbb{B}\mathbb{C}^n\). Here \(M^*\) is the \(*-\)conjugate of the matrix \(M^t\), namely, for \(M=(Z_{ab})_{1\le \,a,b\le \, n}\), we have \(M=(Z^*_{ba})_{1\le \, a,b\le \, n}\), in the sense of Definition 3.3.
In [9] we proved the following equivalent definition, in terms of our matricial form:
Proposition 3.6
Let \(M=M_1+jM_2\in \mathbb{B}\mathbb{C}^{n\times n}\) with \(M_1,M_2\in {\mathbb {C}}^{n\times n}\). Then M is Hermitian, (resp. positive), in the sense of Definition 3.5 if and only if
is Hermitian (resp. a positive matrix, in the sense of positive matrices with complex entries). If we write
with \(P_1=M_1-iM_2\) and \(P_2=M_1+iM_2\). Then, M is positive in the sense of Definition 3.5 if and only if both \(P_1\) and \(P_2\) are positive elements of \({\mathbb {C}}^{n\times n}\).
3.2 \(\mathbb{B}\mathbb{C}\) analyticity
As seen in [17, 24] and [28], the function \(F=F_1+jF_2\) is \(\mathbb{B}\mathbb{C}-\)analytic if and only if the functions \(F_1\) and \(F_2\) are complex holomorphic in \(z_1\) and \(z_2\) i.e.
and if the following Cauchy-Riemann like equations hold:
Remark 3.7
The identity function \(Z=z_1+jz_2\) is \(\mathbb{B}\mathbb{C}-\)analytic, but the \({\mathbb {C}}\)-valued functions \(Z\mapsto z_1\) and \(Z\mapsto z_2\) are not.
3.3 Bicomplex matrices as complex matrices with symmetry
We will now view bicomplex matrices in \({\mathbb{B}\mathbb{C}}^{p\times p}\) as matrices in \({{\mathbb {C}}}^{2p\times 2p}\) with the bicomplex \(\sharp -\)symmetry defined below. We then can reduce the various results to ones in a complex Wiener algebra with symmetry. Note that we will also use the second approach, splitting the problem into two classical complex ones via the idempotent representation.
Remark 3.8
This type of symmetry can be viewed as a complexification of the geometry of complex numbers as well. While in the complex framework a rotation matrix takes the Pauli form, in the bicomplex case it becomes a “double rotation”.
We recall that in the case the bicomplex algebra, as seen is [9] and in (2.10), we have an important structural matrix:
which has the obvious property that \(J\, J^*=I_2\). In this context, we recall Definition 2.17 and we have that the matrix \(M\in {\mathbb {C}}^{2\times 2}\) is called bicomplex \(\sharp \)-symmetric if it satisfies \(M=M^\sharp \), with respect to this J. This matrix gives us the structure of bicomplex numbers as M is J-symmetric if and only if it is of the form (see 2.11).
We now extend this definition to the general matrix case, by first defining
In the same way we define \(J_a^*=J^* \otimes I_a\), where \(\otimes \) denotes the Kronecker (tensor) product of matrices. Then we have:
Definition 3.9
The block matrix \(M\in {\mathbb {C}}^{2a\times 2b}\) is called bicomplex \(\sharp \)-symmetric if and only if it satisfies \(M=M^\sharp \), where
We will make use of all of these notions in the following Sect. 4, specifically in Sect. 4.2.
The following lemma is an immediate result:
Lemma 3.10
Let \(M\in {\mathbb{B}\mathbb{C}}^{p\times p}\), with \(M=(Z_{ab})_{1\le a,b\le p}\) and \(Z_{ab}=z^1_{ab}+j z^2_{ab}\in {\mathbb{B}\mathbb{C}}\). We can decompose M as a matrix in \({{\mathbb {C}}}^{2p\times 2p}\) first as \({\widetilde{M}}\) in Proposition 3.6 and second as a matrix where each \(Z_{ab}=z^1_{ab}+j z^2_{ab}\) is written as the \(2\times 2\) block
namely as \(Z_1\otimes I_{2} + Z_2\otimes J\), where \(Z_1=(z^1_{ab})_{1\le a,b\le p}\) and \(Z_2=(z^2_{ab})_{1\le a,b\le p}\). The matrix from the second decomposition is bicomplex \(\sharp \) symmetric and the definitions of positivity coincide.
We now turn our attention to building the various bicomplex Wiener algebras.
4 The scalar and matrix-valued bicomplex Wiener algebra
We extend the Wiener algebra setting to the set of bicomplex numbers in two ways, as follows. The bicomplex Wiener algebra can be defined in a direct way using the idempotent split, or as a symmetry based Wiener algebra in the complex domain, using a block matrix that fixes the bicomplex structure.
4.1 Wiener bicomplex algebra using bicomplex analyticity
In this section we describe the general matrix-valued bicomplex Wiener algebra, induced by the analytic and algebraic structures of the set of bicomplex numbers, via the idempotent split.
We first recall the definition of matrix-valued bicomplex Wiener algebras, as in Definition 1.16.
Definition 4.1
The bicomplex Wiener algebra \({\mathfrak {W}}^{p\times p}\) consists of the matrix-valued functions of the form
with \(Z\in \partial {\mathbb {K}}\), \(f_n\in {\mathbb{B}\mathbb{C}}^{p\times p}\) and \(\sum _{n\in {\mathbb {Z}}}||f_n||<\infty \).
Here \(||f_n||\) denotes the dual Lie norm extended to set of bicomplex matrices \(f_n=f_{n,1}\textbf{e}_1+f_{n,2}\textbf{e}_2\): \(||f_n||=\Vert f_{n,1}\Vert +\Vert f_{n,2}\Vert \), where \(\Vert f_{n,1}\Vert \) and \(\Vert f_{n,2}\Vert \) are the operator norms as complex matrices.
This Wiener algebra is endowed with point-wise multiplication and norm
The elements of \({\mathfrak {W}}^{p\times p}_+\) are of the form
while elements of \({\mathfrak {W}}^{p\times p}_-\) are of the form
We have the following theorems that describe the structure of the Wiener algebras in this setting.
Theorem 4.2
Let \(f\in {\mathfrak {W}}^{p\times p}\) be such that f(Z) is invertible for \(Z\in \partial {\mathbb {K}}\). Then f is invertible in \(\mathfrak W^{p\times p}\).
Proof
We write \(f(Z)=\sum \nolimits _{n\in {\mathbb {Z}}} f_nZ^n\) with \(f_n\in \mathbb{B}\mathbb{C}^{p\times p}\), and \(f_n=f_{1n}e_1+f_{2n}e_2\) where \(f_{1n}\) and \(f_{2n}\) belong to \({\mathbb {C}}^{p\times p}\).
As \(Z\in \partial {\mathbb {K}}\) is given by corresponding \(\lambda _1=e^{i(t+s)}\) and \(\lambda _2=e^{i(t-s)}\) given by Lemma 1.15
Setting \(s=0\), the classical inversion result leads to
we set
then \(g\in ({\mathcal {W}}(\mathbb{B}\mathbb{C}))^{p\times p}\) and \(f(Z)g(Z)=I_p\) for \(Z\in \partial {\mathbb {K}}\) and this completes the proof. \(\square \)
Corollary 4.3
In the previous theorem it is enough to require invertibility on \(e^{it}\), \(t\in [0,2\pi ]\) to ensure invertibility in \(\mathfrak W^{p\times p}\).
Theorem 4.4
Let \(f\in {\mathfrak {W}}^{p\times p}\) be such that \(f(Z)>0\) for \(Z\in \partial {\mathbb {K}}\). Then f admits a spectral factorization \(f(Z)=f_+(Z)f_+(Z)^*\) where \(f_+^{\pm 1}\in {\mathfrak {W}}^{p\times p}_+\).
Proof
In the same notation as in the proof of the previous theorem, and with \(Z\in \partial {\mathbb {K}}\) given by (1.10) and corresponding \(\lambda _1\) and \(\lambda _2\) given by (5.2)
Setting \(a=0\), the classical factorization result leads to the spectral factorization
with \(f_{\ell +}\in {\mathcal {W}}_+^{p\times p}\) and invertible in \({\mathcal {W}}_+^{p\times p}\). Let
with \(\sum _{n=0}^\infty \Vert g_{\ell ,n}\Vert <\infty \) for \(\ell =1,2\), and set
Then, \(f(Z)=f_+(Z)f_+(Z)^*\). To conclude we need to verify that \(f_+\)is invertible in \({\mathfrak {W}}^{p\times p}_+\). Let
with \(\sum _{n=0}^\infty |h_{\ell ,n}|<\infty \) for \(\ell =1,2\). We have
so that \(f_+(Z)\) is invertible in \({\mathfrak {W}}^{p\times p}_+\). \(\square \)
Corollary 4.5
In the previous theorem it is enough to assume that \(f(e^{ib})>0\) for \(b\in [0,2\pi ]\).
When \(p=1\) the above results reduce to the bicomplex scalar case as shown in the following two corollaries:
Corollary 4.6
Assume that f(Z) invertible on \(\partial {\mathbb {K}}\). Then f is invertible in the bicomplex Wiener algebra \({\mathfrak {W}}\).
Corollary 4.7
Assume that \(f(Z)>0\) invertible on \(\partial {\mathbb {K}}\). Then f admits left and right spectral factorizations in \({\mathfrak {W}}_+\) and \({\mathfrak {W}}_-\).
The proofs of these theorems follow directly from Theorems 4.2 and 2.7, in the case when \(p=1\).
4.2 Bicomplex Wiener algebras using \(\sharp \) symmetries
Let us return to bicomplex matrices with \(\sharp \) symmetry as in Definition 3.9.
Proposition 4.8
Using our specific bicomplex \(\sharp \) definition (with J defined in (2.10)) and setting \(p,q,s\in {\mathbb {N}}\) and \(f\in \mathcal W^{2p\times 2q}\) and \(g\in {\mathcal {W}}^{2q\times 2s}\), we have:
Furthermore, \(f\in {\mathcal {W}}_+^{2p\times 2q}\) if and only if \(f^\sharp \in {\mathcal {W}}_+^{2p\times 2q}\).
Proposition 4.9
The function \(f\in {\mathcal {W}}^{2p\times 2p}\) is bicomplex \(\sharp \)-symmetric if and only if its Fourier coefficients are bicomplex \(\sharp \)-symmetric:
Theorem 4.10
Let \(f\in {\mathcal {W}}^{2p\times 2p}\) be bicomplex \(\sharp \)-symmetric and such that \(f(e^{it})>0\) for \(t\in [0,2\pi ]\). Then, there is a uniquely determined spectral factorization \(f=f_+f_+^*\) where \(f_+\in {\mathcal {W}}_+^{2p\times 2p}\) is bicomplex \(\sharp \)-symmetric and satisfies \(f_+(1)=I_{2p}\) and invertible and any other spectral factorization differs by a bicomplex \(\sharp \)-symmetric unitary multiplicative constant on the right from \(f_+\).
Using Lemma 3.10, we can now establish the equivalence between the two types of Wiener algebras in the bicomplex case.
Theorem 4.11
\({\mathfrak {W}}^{p \times p}\) is isomorphic to the space of \(\mathcal W^{2p \times 2p}\) with bicomplex \(\sharp \) symmetry.
4.3 Rational bicomplex functions and the bicomplex Wiener algebra
We now re-write Theorems 2.23 and 2.24 in the bicomplex case. The proofs are left to the reader.
Theorem 4.12
Let f(z) be a \({\mathbb {C}}^{2p\times 2p}\)-valued rational function, analytic at infinity, bicomplex \(\sharp \)-symmetric, and regular on the unit circle. Then f admits a bicomplex \(\sharp \)-symmetric realization, analytic at infinity, and regular on the unit circle.
Proof
Follows from proof of Theorem 2.23. \(\square \)
Theorem 4.13
Let \(f\in {\mathcal {W}}^{2p\times 2p}\) rational and taking strictly positive values and bicomplex \(\sharp \) symmetric. Then it admits rational bicomplex \(\sharp \)-symmetric left and right factorizations.
Proof
This follows from the fact that the a bicpomplex \(\sharp \)-symmetric rational function admits a bicomplex \(\sharp \)-symmetric realization. \(\square \)
Remark 4.14
Using this isomorphism between the bicomplex Wiener algebra and the complex Wiener algebra of double dimension with specific symmetries, our proof can avoid the issue of singularities which are not isolated points, just as in the case of realization of rational functions of several complex variables. We point the reader to [23] for a thorough discussion of Laurent expansions of bicomplex functions where sigularities are not isolated.
5 Bicomplex superoscillations
In this Section we apply the previous matrix decomposition of the bicomplex Wiener algebra to write a result on bicomplex superoscillations. We start with the distinguished boundary \(|\lambda _1|=|\lambda _2|=1\) of the bicomplex bidisk. We use Lemma 1.13, and formulae (1.8) and (1.10), to show that the decomposition along the idempotents corresponds to
i.e.
We have for \(z_1=uz_2\) when \(z_2\not =0\):
which is rewritten as (1.8). To conclude, note that
and that from Lemma 1.15 we have
and
Following Definition 2.28 in the classical case, we now write a superoscillatory sequence in the bicomplex case. For given \(x,y,a,b\in {{\mathbb {R}}}\) and \(m\ge 0\) we take specific bicomplex numbers \(Z=\lambda _1\textbf{e}_1+\lambda _2\textbf{e}_2\), where
and
Definition 5.1
With the notation above, we have the following bicomplex superoscillatory sequence \(F_m(x,y,a,b)=(\lambda _1 \textbf{e}_1+\lambda _2 \textbf{e}_2)^m=\lambda _1^m\textbf{e}_1+\lambda _2^m\textbf{e}_2\), namely
Using Lemma 2.29 in the complex case we have the following structure of bicomplex superoscillations:
Lemma 5.2
For the bicomplex superoscillatory sequence in Definition 5.1 we have that:
where
We also have the same type of limiting behavior:
Lemma 5.3
We have that:
and \(F_m(x,y,a,b)=F_m(x,a)\textbf{e}_1+F_m(y,b)\textbf{e}_2\).
Theorem 5.4
Let \(f\in {\mathfrak {W}}^{p\times p}\), then \(f(Z)=\sum _{n\in {\mathbb {Z}}} f_nZ^n,\) where \(Z=e^{i t}e^{j s}\) can be uniformly approximated on compact subsets of \({\mathbb {R}}\) by a superoscillatory sequence defined by:
Proof
We have that
Therefore
Using Theorem 2.30 we have that \(f_1(e^{i(t-s)})\) is approximated by:
Similarly, \(f_2(e^{i(t+s)})\) is approximated by:
Using the Lie norm induced on the space of bicomplex matrices we can write an approximation for f(Z) to be \({\mathcal {E}}_1\textbf{e}_1+ \mathcal E_2\textbf{e}_2\):
which yields:
namely:
This completes the proof. \(\square \)
This application concludes the paper and we expect to be able to apply this framework to other settings as well. Wiener Algebras have many applications in the theory of Linear Systems in the classical case, so we expect results in this direction as well.
We are also investigating whether the setting of bicomplex Wiener algebras is useful in other applications of Quantum Mechanics.
Data availability
Please note that no datasets were generated or analyzed during the current study.
References
Aharonov, Y., Albert, D., Vaidman, L.: How the result of a measurement of a component of the spin of a spin-1/2 particle can turn out to be 100. Phys. Rev. Lett. 60, 1351–1354 (1988)
Aharonov, Y., Vaidman, L.: Properties of a quantum system during the time interval between two measurements. Phys. Rev. A 41, 11–20 (1990)
Aharonov, Y., Rohrlich, D.: Quantum Paradoxes: Quantum Theory for the Perplexed. Wiley, Weinheim (2005)
Alpay, D., Attia, H., Ben-Porat, S., Volok, D.: Spectral factorization in the non-stationary Wiener algebra. arXiv (2003)
Alpay, D., Gohberg, I.: Inverse spectral problems for difference operators with rational scattering matrix function. Integral Equ. Oper. Theory 20(2), 125–170 (1994)
Alpay, D., Colombo, F., Diki, K., Sabadini, I., Volok, D.: Discrete analytic functions, structured matrices and a new family of moment problems. Bull. Sci. Math. 179, 103175 (2022)
Alpay, D., Colombo, F., Sabadini, I.: Superoscillations and analytic extensions in Schur analysis’. J. Fourier Anal. Appl. 27(2), 1–19 (2021)
Alpay, D., Colombo, F., Kimsey, D., Sabadini, I.: Wiener algebra for the quaternions. Mediterr. J. Math. 13(5), 2463–2482 (2016)
Alpay, D., Lewkowicz, I., Vajiac, M.: Interpolation with symmetry and a Herglotz Theorem in the bicomplex setting. J. Math. Anal. Appl. 524(2), 127201 (2023)
Alpay, D., Luna-Elizarraras, E., Shapiro, M., Struppa, D.: Basics of Functional Analysis with Bicomplex Scalars, and Bicomplex Schur Analysis SpringerBriefs in Mathematics. Springer, Cham (2014)
Aoki, T., Colombo, F., Sabadini, I., Struppa, D.C.: Continuity theorems for a class of convolution operators and applications to superoscillations. Ann. Mat. Pura Appl. (4) 197(5), 1533–1545 (2018)
Aoki, T., Colombo, F., Sabadini, I., Struppa, D.C.: Continuity of some operators arising in the theory of superoscillations. Quantum Stud. Math. Found. 5, 463–476 (2018)
Berry, M., Dennis, M.R.: Natural superoscillations in monochromatic waves in D dimension. J. Phys. A 42, 022003 (2009)
Berry, M.V., Popescu, S.: Evolution of quantum superoscillations, and optical superresolution without evanescent waves. J. Phys. A 39, 6965–6977 (2006)
Bochner, S., Phillips, R.S.: Absolutely convergent Fourier expansions for non-commutative normed rings. Ann. Math. 2(43), 409–418 (1942)
Colombo, F., Struppa, D.C., Yger, A.: Superoscillating sequences towards approximation in \({\cal{S} }\) or \(\cal{S^{\prime }} \)-type spaces and extrapolation. J. Fourier Anal. Appl. 25(1), 242–266 (2019)
Colombo, F., Sabadini, I., Struppa, D.C., Vajiac, A., Vajiac, M.B.: Singularities of functions of one and several bicomplex variables. Arkiv for matematik, Institut Mittag-Leffler (2011)
Dym, H., Gohberg, I.: Extensions of matrix valued functions with rational polynomial inverses. Integral Equ. Oper. Theory 2, 503–528 (1979)
Gohberg, I., Kaashoek, M.A.: Block Toeplitz operators with rational symbols, Oper. Theory Adv. Appl., Contributions to operator theory and its applications (Mesa, AZ, 1987), Birkhäuser, pp. 385–440 (1988)
Gohberg, I.C., Feldman, I.A.: Convolution equations and projection methods for their solution. Translations of Mathematical Monographs, Vol. 41. American Mathematical Society, Providence. Translated from the Russian by F. M. Goldware (1974)
Gohberg, I., Goldberg, S., Kaashoek. , M.A.: Classes of linear operators. Vol. II, volume 63 of Operator Theory: Advances and Applications. Birkhauser Verlag, Basel (1993)
Gohberg, I., Kaashoek, M.A., Woerdeman, H.J.: The band method for positive and strictly contractive extension problems: an alternative version and new applications. Integral Equ. Oper. Theory 12(3), 343–382 (1989)
Luna-Elizarraras, E., Perez-Regalado, C.O., Shapiro, M.: Singularities of bicomplex holomorphic functions. MMAS (2021)
Luna-Elizarraras, E., Shapiro, M., Struppa, D., Vajiac, A.: Bicomplex Holomorphic Functions. The Algebra, Geometry and Analysis of Bicomplex Numbers. Birkhäuser, Cham (2015)
Price, G.B.: An Introduction to Multicomplex Spaces and Functions, Monographs and Textbooks in Pure and Applied Mathematics. Marcel Dekker, Inc., New York (1991)
Sabadini, I., Vajiac, A., Vajiac M.B.: Bernstein-type Inequalities for Bicomplex Polynomials, Advances in Complex Analysis and Operator Theory, pp. 281–299 (2017)
Shelah, Y.: Quaternionic Wiener algebras, factorization and applications. Adv. Appl. Clifford Algebra 27(3), 2805–2840 (2017)
Vajiac, A., Vajiac, M.: Multicomplex hyperfunctions. Complex Var. Elliptic Equ. 56(12), 1–13 (2011)
Wiener, N.: Tauberian theorems. Ann. Math. 33, 1–100 (1932)
Funding
Open access funding provided by SCELC, Statewide California Electronic Library Consortium. D. Alpay’s research is supported by the Foster G. and Mary McGaw Professorship in Mathematical Sciences.
Author information
Authors and Affiliations
Contributions
All authors were equally involved in all aspects of this work, from conception to description of all details involved.
Corresponding author
Ethics declarations
Conflict of interest
The authors (Alpay, Lewkowicz, and Vajiac) declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Ethical approval
Not applicable.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
D. Alpay thanks the Foster G. and Mary McGaw Professorship in Mathematical Sciences, whsich supported his research.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Alpay, D., Lewkowicz, I. & Vajiac, M. Discrete Wiener algebra in the bicomplex setting, spectral factorization with symmetry, and superoscillations. Anal.Math.Phys. 13, 54 (2023). https://doi.org/10.1007/s13324-023-00812-y
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s13324-023-00812-y