Solving the Sylvester equation in Banach modules
Bogdan D. Djordjević · 2026
Read Solving the Sylvester equation in Banach modules by Bogdan D. Djordjević (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-13419.txt.
Introduction
Bogdan Djordjevic fully characterises when the Sylvester equation ax minus xb equals c has a solution in Banach modules. The setting uses two unital complex Banach algebras and a Banach module acting between them, with given elements a, b and c whose spectra overlap. He states checkable rules that are both sufficient and necessary for a solution to exist. He also gives formulas for particular solutions when the equation can be solved. Finally, he characterises when solutions are unique.
Catalog record
| CREATOR | Bogdan D. Djordjević |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 67953 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.13419
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Solving the Sylvester equation in Banach modules Bogdan D. Djordjević Abstract For given unital complex Banach algebras 𝒜 1 {\mathcal{A}}_{1} and 𝒜 2 {\mathcal{A}}_{2} , let 𝔐 \mathfrak{M} be a Banach module acting between them. Let a ∈ 𝒜 1 a\in{\mathcal{A}}_{1} , b ∈ 𝒜 2 b\in{\mathcal{A}}_{2} , and c ∈ 𝔐 c\in\mathfrak{M} be provided such that σ 𝒜 1 ( a ) ∩ σ 𝒜 2 ( b ) ≠ ∅ \sigma_{{\mathcal{A}}_{1}}(a)\cap\sigma_{{\mathcal{A}}_{2}}(b)\neq\emptyset . In this paper we completely characterize the consistency of the Sylvester equation a x − x b = c . ax-xb=c. Precisely, we establish verifiable sufficient and necessary solvability conditions, and we provide some formulas for particular solutions x ∈ 𝔐 x\in\mathfrak{M} when the equation is solvable. Moreover, we characterize the uniqueness of the solutions. MSC 2020 Classification: 46H25; 46H10; 46H30; 47D03; 47A08. Keyphrases: Sylvester equation; Banach modules; Banach algebras; Operator matrices. 1 Introduction Let {\mathcal{A}}_{1} and {\mathcal{A}}_{2} be unital complex Banach algebras. A Banach space \mathfrak{M} defines a Banach ({\mathcal{A}}_{1},{\mathcal{A}}_{2})- module, if it is simultaneously a left Banach module for {\mathcal{A}}_{1} , and a right Banach module for {\mathcal{A}}_{2} , with respect to some (fixed) multiplications \cdot_{1}:{\mathcal{A}}_{1}\times\mathfrak{M}\to\mathfrak{M} , and \cdot_{2}:\mathfrak{M}\times{\mathcal{A}}_{2}\to\mathfrak{M} . It is inherently understood that the sub-multiplicativity is preserved within such structures, that is, \|a_{1}\cdot_{1}c\cdot_{2}a_{2}\|_{\mathfrak{M}}\leq\|a_{1}\|_{{\mathcal{A}}_{1}}\cdot\|c\|_{\mathfrak{M}}\cdot\|a_{2}\|_{{\mathcal{A}}_{2}}, for any a_{i}\in{\mathcal{A}}_{i} , c\in\mathfrak{M} , and that the associative law holds: \left(a_{1}\cdot_{1}c\right)\cdot_{2}a_{2}=a_{1}\cdot_{1}\left(c\cdot_{2}a_{2}\right)\in\mathfrak{M}. It is also assumed that \mathfrak{M} is not a trivial module, meaning that, if c\neq 0_{\mathfrak{M}} , then 1_{{\mathcal{A}}_{1}}\cdot c\neq 0_{\mathfrak{M}} and c\cdot 1_{{\mathcal{A}}_{2}}\neq 0_{\mathfrak{M}} . With respect to the previous notation, let a\in{\mathcal{A}}_{1} , b\in{\mathcal{A}}_{2} , and c\in\mathfrak{M} be given elements. In this paper we study and solve the algebraic Sylvester equation ax-xb=c (1) for the unknown solution(s) x\in\mathfrak{M} : we establish both sufficient and necessary solvability conditions, and provide several formulas for particular solutions. In addition, we obtain new necessary and sufficient conditions under which the solution is unique. To the best of our knowledge, these results are original and represent a significant progress compared to the existing findings in the available literature. 1.1 Sylvester equations: Known results The Sylvester equation ( 1 ) is a well-known equation which has been extensively studied in the matrix and operator setting (see [ 2 ] , [ 7 ] , [ 8 ] , [ 17 ] , [ 24 ] , [ 28 ] , [ 31 ] , [ 32 ] ), and even in rings with involution [ 21 ] , Hilbert C^{*}- modules [ 25 ] , and semisimple Banach algebras [ 29 ] . The most well-known criterion regarding the existence and uniqueness of a solution lies in spectral properties of a and b : when a and b are bounded linear operators given in the corresponding Banach spaces, provided such that their spectrums are disjoint, then for any given bounded linear operator c , there exists a unique bounded linear operator x (in the respective operator space), such that ( 1 ) holds. Moreover, when a , b , and c are matrices (or operators in finite-dimensional vector spaces), then there exists a unique solution x to ( 1 ), if and only if the spectra of a and b are disjoint. For this reason, in the bounded operator (and matrix) setting, the equation is said to be regular whenever the spectrums of a and b are disjoint, and it is called singular otherwise. Regular Sylvester equations and their generalizations in matrices and operators have been extensively studied in the available literature, see e.g. [ 2 ] , [ 7 ] , [ 8 ] , [ 17 ] , [ 28 ] , [ 31 ] , [ 32 ] , and numerous references therein. Conversely, singular Sylvester equations lack such results. There are some special cases in which the equation has been solved successfully: for instance, when {\mathcal{A}}_{1} , {\mathcal{A}}_{2} , and \mathfrak{M} are finite-dimensional, i.e., when the equation ( 1 ) is provided in matrices, solvability conditions have been established and the solution set has been described completely, see [ 10 ] , [ 11 ] , [ 15 ] , [ 19 ] , [ 20 ] , [ 22 ] , and [ 30 ] . Moreover, when {\mathcal{A}}_{1} , {\mathcal{A}}_{2} , and \mathfrak{M} are Banach spaces of bounded linear operators (or even closed operators in some instances), papers [ 6 ] , [ 13 ] , [ 14 ] , [ 25 ] , and [ 32 ] , provide some sufficient solvability conditions for ( 1 ), as well as some forms for particular solutions in those settings. These results usually rely on the existence of Riesz points in the respective spectra, or demand some stronger properties, like self-adjointness, Drazin-Koliha invertibility, von Neumann regularity, etc. The difficulty with the singular equation is that the nonempty spectral intersection for a and b does not guarantee (nor disprove) neither the existence nor the uniqueness of solutions, therefore this scenario always has some underlying issues. Moreover, the solutions obtained in the said papers ( [ 12 ] , [ 13 ] , [ 14 ] , [ 25 ] , and [ 32 ] ) rely heavily on the spectral properties of a and b and the corresponding eigenspaces, thus making them not easily obtainable. Apart from the previously mentioned results, papers [ 12 ] , [ 16 ] , and [ 23 ] offer some interesting necessary solvability conditions for ( 1 ), without an explicit form for a particular solution. Finally, the most well-known necessary sovlability criterion for ( 1 ) (sans uniqueness) is the Roth’s removal rule, which states that, if ( 1 ) is solvable, then the matrices \left[\begin{array}[]{cc}a&c\\ 0&b\\ \end{array}\right] and \left[\begin{array}[]{cc}a&0\\ 0&b\\ \end{array}\right] are similar, and the converse statement holds only in finite-dimensional spaces. Still, none of the necessary conditions mentioned here can be utilized for obtaining sufficient sovlability conditions for ( 1 ) in the infinite-dimensional case, nor do they offer a way for finding a particular solution. 1.2 Banach module setting Throughout this paper we are working with fixed unital complex Banach algebras {\mathcal{A}}_{1} and {\mathcal{A}}_{2} , and a fixed Banach ({\mathcal{A}}_{1},{\mathcal{A}}_{2})- module \mathfrak{M} . We denote this convention as ({\mathcal{A}}_{1},{\mathcal{A}}_{2},\mathfrak{M}) . For easier notation, we simplify the indexing and write a_{1}\cdot_{1}c\cdot_{2}a_{2}\equiv a_{1}ca_{2},\quad a_{i}\cdot_{{\mathcal{A}}_{i}}b_{i}\equiv a_{i}b_{i},\quad c\in\mathfrak{M},\ a_{i},b_{i}\in{\mathcal{A}}_{i},\quad i=1,2. Similarly, we write \|\cdot\| instead of \|\cdot\|_{\mathfrak{M}} , or \|\cdot\|_{{\mathcal{A}}_{i}} , etc. The spectrum of a_{i} within the respective algebra {\mathcal{A}}_{i} , i=1,2 , will be written as \sigma_{{\mathcal{A}}_{i}}(a_{i}) , or as \sigma(a_{i}) if there is no chance of confusion. The algebra of bounded linear operators over \mathfrak{M} is denoted as {\mathcal{B}}(\mathfrak{M}) . For a given L\in{\mathcal{B}}(\mathfrak{M}) , the image (range) of the operator L is denoted as {\mathcal{R}}(L) . Finally, for any a_{i}\in{\mathcal{A}}_{i} , the set (a_{i})^{\prime} denotes the set of elements in {\mathcal{A}}_{i} which commute with a_{i} . For more on Banach (and Hilbert C^{*}- ) modules consult [ 1 ] , [ 3 ] , [ 4 ] , [ 5 ] , [ 16 ] , [ 25 ] , and numerous references therein. Accordingly, we assume that a\in{\mathcal{A}}_{1} , b\in{\mathcal{A}}_{2} , and c\in\mathfrak{M} from ( 1 ) are also provided and