Lifting Normal Elements in Nonseparable Calkin Algebras

Ye Zhang, Don Hadwin, Yanni ChenarXiv 2014 · DOI · arXiv

TL;DR

This paper solves a long-standing problem in operator theory by proving that "normal" elements in nonseparable Calkin algebras can always be "lifted" to normal operators on the original Hilbert space. By leveraging a powerful new distance estimate from 2014, the authors bridge a gap that has remained open since 1981, showing that for most large-scale operator algebras, there are no structural obstructions to finding these normal representatives.

What Problem Does This Paper Address?

In the world of functional analysis, we often study the algebra of all bounded linear operators on a Hilbert space, denoted as $B(H)$. When we "divide" this algebra by a closed ideal (a specific subset of operators), we get a quotient called a Calkin algebra.

A central question in this field is lifting: if you have an element in the Calkin algebra that has a nice property—like being "normal" (meaning it commutes with its own adjoint, $tt^* = t^*t$)—can you find an actual operator in the original space $B(H)$ that is also normal and "maps down" to that element?

For "small" (separable) Hilbert spaces, this was famously solved in the 1970s by Brown, Douglas, and Fillmore (BDF theory), who found that the only thing stopping a lift is a numerical value called the Fredholm index. However, for "large" (nonseparable) Hilbert spaces, the situation is much more complex. While some cases were solved in 1981, a specific scenario involving "countably cofinal" cardinals (infinite sizes that can be reached by a sequence of smaller sizes) remained a mystery for over 30 years.

Methods

The authors achieve their breakthrough by utilizing a very recent (at the time of writing) tool: the Kachkovskiy-Safarov inequality. This is a "remarkable distance estimate" that allows mathematicians to calculate how far any given operator is from the set of normal operators.

The proof strategy involves:

  1. Decomposition: Breaking the Hilbert space down into a direct sum of smaller subspaces.
  2. Von Neumann Algebra Theory: Applying the distance estimate to different "types" of von Neumann algebras (Type I, II, and III factors).
  3. Approximation: Showing that if an element is "almost normal" in the Calkin algebra, it is actually the limit of a sequence of truly normal operators.
  4. Unitary Equivalence: Using sequences of unitary operators to "glue" these approximations together into a single, global normal operator that serves as the lift.

Datasets

As this is a theoretical paper in pure mathematics, no empirical datasets were used. The "data" consists of abstract mathematical structures like Hilbert spaces and Calkin algebras.

Key Insights and Ideas

  • The Power of Distance Estimates: The core "aha moment" is realizing that the Kachkovskiy-Safarov inequality provides exactly the quantitative grip needed to handle the "countably cofinal" case. It turns a structural problem into a distance-minimization problem.
  • Cardinality Matters: The paper highlights how the behavior of operator algebras is deeply tied to the set-theoretic properties of the underlying Hilbert space's dimension.
  • Completing the Map: By solving the countably cofinal case, the authors effectively complete the characterization of lifting for all closed ideals in $B(H)$, except for the classical ideal of compact operators (which is already governed by BDF theory).

Key Results

  • Theorem 1: The main result states that if $H$ is a nonseparable Hilbert space and $m$ is a countably cofinal cardinal, every normal element in the $m$-Calkin algebra has a normal lift in $B(H)$.
  • Closing the Gap: This result proves that for these large-scale algebras, there are no obstructions (like the Fredholm index in the separable case) to lifting normality. If it looks normal in the quotient, it can be made normal in the original space.

Limitations and Future Work

  • The Compact Operator Exception: The paper specifically excludes the ideal of compact operators ($\mathcal{K}(H)$). This is intentional, as the "obstruction-free" nature of lifting only applies when the ideal is "larger" than the compact operators.
  • Other Properties: While this paper focuses on normal elements, the authors briefly mention that other properties (like being subnormal or hyponormal) are preserved under certain conditions. Future work might explore whether the Kachkovskiy-Safarov approach can be extended to provide a unified lifting theory for a broader class of operator properties in nonseparable spaces.

Summary updated

Abstract

We use the remarkable distance estimate of Ilya Kachkovskiy and Yuri Safarov, to show that if $H$ is a nonseparable Hilbert space and $K$ is any closed ideal in $B(H)$ that is not the ideal of compact operators, then any normal element of $B(H)/K$ can be lifted to a normal element of $B(H)$.
BibTeX
@misc{zhang2014lifting,
  author = {Ye Zhang and Don Hadwin and Yanni Chen},
  title = {{Lifting Normal Elements in Nonseparable Calkin Algebras}},
  year = {2014},
  eprint = {1403.6228},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1403.6228},
}

Loading related content...

Figure