Every bounded operator on an infinite-dimensional Banach space has norm attainment

Does there exist an infinite-dimensional Banach space XX such that every bounded linear operator T∈L(X,X)T\in\mathcal{L}(X,X) attains its norm; that is, for every T∈L(X,X)T\in\mathcal{L}(X,X) there exists x∈Xx\in X with ∥x∥=1\lVert x\rVert=1 and ∥Tx∥=∥T∥\lVert Tx\rVert=\lVert T\rVert?

References

Primary source

arXiv

Progress summary

Refreshed
Open

The existence of an infinite-dimensional Banach space where every bounded operator reaches its norm remains unresolved; a new paper settles only related second-adjoint cases.

The problem asks whether there exists an infinite-dimensional Banach space XX such that every operator in L(X)L(X) attains its norm. M. I. Ostrovskii posed this question; no proof or counterexample for the original statement is reported.

Known results

  • James’s characterization implies that a hypothetical XX and L(X)L(X) would both be reflexive.
  • Holub proved that such an XX cannot have the approximation property; more broadly, it cannot have the bounded A\mathcal{A}-approximation property for any nontrivial operator ideal A\mathcal{A}.
  • Kalton proved that L(X)L(X) cannot be reflexive when XX is nonseparable.
  • Thus possible examples must be separable reflexive spaces without suitable approximation properties.

September 2026 related preprint

A preprint claims that every bounded operator on c0c_0 and ℓ1\ell_1 has a norm-attaining second adjoint, and distinguishes second- from higher-adjoint behavior on cc. It does not resolve the original universal norm-attainment question, and the claim is unverified.

Current status (as of September 2026): The original existence problem remains open; only necessary obstructions and related second- and higher-adjoint results are recorded.

Sources

Solutions 0

No solutions have been posted yet.