Separable neighbourhood of identity in C ∗ -algebras
Mizanur Rahaman, Mateusz Wasilewski · 2026
Read Separable neighbourhood of identity in C ∗ -algebras by Mizanur Rahaman, Mateusz Wasilewski (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-2603-29556.txt.
Introduction
Rahaman and Wasilewski ask how big a separable neighborhood around the identity can be in a C-star algebra. The finite case is well understood, but general algebras pose new challenges. They turn the size question into a norm estimate. The norm belongs to contractive positive maps between algebras. That link ties the size to traits of the algebra, such as its rank. As a side result, they resolve a recent conjecture by Musat and Rørdam.
Catalog record
| CREATOR | Mizanur Rahaman, Mateusz Wasilewski |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51834 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.29556
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Separable neighbourhood of identity in C ∗ -algebras Mizanur Rahaman, Mateusz Wasilewski Abstract We study the structure of separable elements in bipartite C ∗ -algebras, focusing on the existence and size of a separable neighbourhood around the identity element. While this phenomenon is well understood in the finite-dimensional setting, its extension to general C ∗ -algebras presents additional challenges. We show that the problem of determining such a neighbourhood can be reduced to estimating the completely bounded norm of contractive positive maps. This approach allows us to characterize the size of such neighbourhoods in terms of structural properties of the algebra, notably its rank. As a consequence, we also resolve a recent conjecture of Musat and Rørdam. 1. Introduction Understanding various aspects of separability and entanglement is a central theme in quantum information theory. In finite dimensional systems investigating the structure of separable states is still an active area of research ( [ LMW24 ] , [ CÐ15 ] ). In particular, the geometry of separable states around the identity (maximally mixed state) has been well studied and plays an important role in entanglement detection and robustness questions. However, exploring this phenomenon beyond matrix algebras to general C ∗ -algebras is both natural and technically subtle. More concretely, it is known ( [ GB02 ] ) that in the finite dimensional setting, there is a nontrivial neighbourhood of the bipartite identity element consisting entirely of separable elements. In other words, sufficiently small perturbations of the identity cannot generate entanglement. The main question is then, what happens in an arbitrary C ∗ -algebra setting? In this paper we undertake this task, namely, given two C ∗ -algebras A and B , we explore the problem of finding an optimal neighbourhood of separable elements of the bipartite identity 1_{A}\otimes 1_{B} . By “separable elements” we mean the closure of the set of positive elements x\in A\otimes_{\min}B that admit the decomposition x=\sum_{i}a_{i}\otimes b_{i} , where a_{i}\in A_{+},b_{i}\in B_{+} . The identity element is the natural reference point of the cone of positive elements, and in matrix algberas it is the dual of the trace functional, so it makes sense to investigate this phenomenon based on the identity element. In matrix algebras, separability questions can often be reformulated using positive maps via the well known Horodecki criterion ( [ HHH96 ] ). And using this connection, Gurvits and Burnum ( [ GB02 ] ) obtained a separable ball of radius 1/n of the element 1_{M_{n}(\mathbb{C})}\otimes 1_{M_{n}(\mathbb{C})} (when the operator norm is considered). Their result also captures various other values of the radius when one chooses various Schatten p -norms. We will mainly focus on the operator norm as for us this is the right generalization to the arbitrary C ∗ -algebra setting. The guiding observation of our present work is that the problem of finding the size of the separable neighbourhood can be encoded in computing the completely bounded (cb for short) norm problem of positive maps. This fact was implicit in the work of Ando ( [ And04 ] ) who although did not consider cb norms of positive maps, his estimates were primarily about cb norms. We spell this connection very clearly in our first result on the matrix algebra case (see Theorem 3.6 ). There has been a long effort in understanding the theory of entanglement and separability in operator algebras pioneered by Størmer ( [ Stø08 ] ), Arveson ( [ Arv09 ] ), Paulsen ( [ Pau ] ), and other researchers ( [ PTT11 , JKPP11 ] ). In particular the Horodecki criterion has been generalized by Størmer ( [ Stø09 ] ) to the setting where one of the algebras is nuclear and the other is an UHF algebra. Later Miller and Olkiewicz ( [ MO14 ] ) extended this to nuclear C ∗ algebras. Note that these generalizations were done to characterize "separable states" as opposed to “separable elements" in a C ∗ -algebra and there is a subtle difference in these two concepts if one goes beyond the matrix algebra case. Since we are dealing with separable elements, in Section 4 we establish the Horodecki criterion for for separable elements in nuclear C ∗ -algebras. Having this at our disposal, we identify the problem of finding the separable neighbourhood of the bipartite identity element 1_{A}\otimes 1_{B} is equivalent to the problem of computing how large the cb norm of a positive map can be from the algebra A to B . Now computing cb norms of completely bounded maps is a fundamental problem in operator algebras and underpins many interesting concepts of the underlying spaces ( [ Pau , Smi83 , Tom83 ] ). In particular, the rank of a C ∗ -algebra, i.e. the largest dimension of an irreducible representation, plays a crucial role in determining how large the cb norm of a linear map acting on it can be. We have our main result as follows Theorem . For two unital C ∗ -algebras A,B , let \gamma(A,B):=\sup\{r\in\mathbb{R}_{+}:\mathds{1}_{A}\otimes\mathds{1}_{B}-x\text{ is separable for all }x=x^{\ast}\text{ with }\|x\|\leqslant r\}. If both have infinite rank, then \gamma(A,B)=0 . In the case where at least one has a finite rank we have \frac{1}{\gamma(A,B)}=\eta(A,B)=\min(rank(A),rank(B)), where \eta(A,B):=\sup\{\|\Phi\|_{cb}:\Phi:A\rightarrow B,\ \text{contractive positive map}\}. Note that the statement and the proof of the above theorem (see Theorem 5.13 ) not only encompasses the known finite dimensional result, it also shows that the separable neighbourhood problem has no non-trivial solution when both the algebras in a bipartite system is B(H) , where H is a separable infinite dimensional Hilbert space, as the rank is infinite. This fact should be compared with the result of Clifton and Halvorson ( [ CH99 ] ) where they showed that entangled states are dense in the trace norm topology when one subsystem of the bipartite system is infinite dimensional. Although our primary focus is on the separable neighbourhood problem, the methods we develop have further consequences. In particular, they enable us to resolve a conjecture raised by Magdalena Musat and Mikael Rørdam in their recent work on entanglement theory in C ∗ -algebras [ MR25 ] . 2. Preliminaries We will study (completely) positive maps between C^{\ast} -algebras. Definition 2.1 . Let \Phi:A\to B be a linear map between C^{\ast} -algebras. It is called positive if \Phi(a^{\ast}a)\geqslant 0 for all a\in A . It is called completely positive if for all n\in\mathbb{N} the amplification \Phi_{n}:=Id_{n}\otimes\Phi:M_{n}\otimes A\to M_{n}\otimes B is positive. A good overview of completely positive maps can be found in [ Pau ] and [ BO08 ] . An important thing to note is that either the domain or the codomain is commutative then positive maps are automatically completely positive. There are also useful criteria for complete positivity if either the domain or the codomain is the matrix algebra. Lemma 2.2 ( [ BO08 , Chapter 1: Proposition 5.11, Proposition 5.13] ) . Let A be a C^{\ast} -algebra. The linear map \Phi:M_{n}\to A is completely positive if and only if its Choi matrix Choi(A):=\sum_{i,j=1}^{n}e_{ij}\otimes\Phi(e_{ij})\in M_{n}\otimes A is positive. The linear map \Psi:A\to M_{n} is completely positive if and only if the functional \widehat{\Psi}:M_{n}\otimes A\to\mathbb{C} is positive, where \widehat{\Psi}(\sum_{i,j}e_{ij}\otimes a_{ij})=\sum_{i,j=1}^{n}\Psi(a_{ij})_{ij} . Another important concept will be that