arXiv Science⌕ Search

arXiv · 0707.3801

$N_{\p}$-type quotient modules on the torus

Abstract

Structure of the quotient modules in $\hh$ is very complicated. A good understanding of some special examples will shed light on the general picture. This paper studies the so-call $N_{\p}$-type quotient modules, namely, quotient modules of the form $\hh\ominus [z-\p]$, where $\p (w)$ is a function in the classical Hardy space $H^2(\G)$ and $[z-\p]$ is the submodule generated by $z-\p (w)$. This type of quotient modules serve as good examples in many studies. A notable feature of the $N_{\p}$-type quotient module is its close connections with some classical single variable operator theories.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Keiji Izuchi, Rongwei Yang. 2007-07-25. $N_{\p}$-type quotient modules on the torus. https://arxiv.org/abs/0707.3801

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

Matrix-valued regular functions of a quaternionic variable - a decomposition and some approximation results

A quaternionic matrix-valued regular function is a map $F: Ω\rightarrow M_n(\mathbb{H})$ whose entries are (left) regular functions of a quaternion variable, where $Ω$ is a domain in $\mathbb{H}$. Given any $n \times n$ matrix-valued regular function $F(q)$ over $\mathbb{H}$ having norm at most $1$, we derive a decomposition of $F(q)$ in the form $U \star \begin{bmatrix} ||F(q_0)|| & 0\\0 & G(q)\end{bmatrix} V$ for some constant unitary matrices $U, V$, some vector $q_0$ and an $(n-1) \times (n-1)$ matrix-valued regular function $G(q)$, where $\star$ denotes the star product over $\mathbb{H}$. We use this to deduce that any matrix-valued regular function $F(q)$ having norm at most $1$ can be uniformly approximated by a convex combination of rational unitary functions.

math.FA↗

Classification of bipartite observables I: stable and uniformly stable observables

When two quantum systems are prepared independently and each is driven to an idealized limit, the expectation of a bipartite observable may depend on which system is idealized first. We call an observable stable when it does not. We show that stability is equivalent to two operational properties. First, every idealized preparation of one system can be simulated to arbitrary accuracy, uniformly over all probes, by finite mixtures of actual preparations, provided that the probing system is idealized first. Second, any two idealized preparations have a unique, symmetric independent product. Under a uniform form of stability, we prove, the number of preparations that $n$ adaptive probes can tell apart at a fixed margin is bounded by a polynomial in $n$, and responses can be learned with a bounded number of mistakes; otherwise, at some margin, protocols with $n$ probes tell apart $2^n$ preparations. The swap test, Hong--Ou--Mandel coincidences and discrete equality tests turn out to be uniformly stable, with bounds independent of the dimension, while order comparisons of discrete spectra unbounded above are unstable. The proofs rest on the classical fact that a single double limit condition appears, under different names, in topology (Grothendieck), functional analysis (Krivine--Maurey), algebra (Arens), probability (Simons) and combinatorics (Pták), and in model theory as Shelah's stability, where the independent product corresponds to Harrington's lemma. Model theory thus serves as a Rosetta stone, pairing each physical question with its mathematical formulation.

math.FA↗