← All problems

Invariant subspace problem

91 years old · traced to

Let HH be a separable complex Hilbert space and let T ⁣:H→HT\colon H\to H be a bounded linear operator. A closed linear subspace W⊆HW\subseteq H is invariant under TT if T(W)⊆WT(W)\subseteq W, and nontrivial if W≠{0}W\ne\{0\} and W≠HW\ne H.

Conjecture. If dim⁡H≥2\dim H\ge2, then every bounded linear operator TT on HH has a nontrivial closed invariant subspace.

If 2≤dim⁡H<∞2\le\dim H<\infty, an eigenvector of TT spans a nontrivial invariant subspace, so the content of the problem is the case where HH is separable and infinite-dimensional. Nonseparable spaces are excluded because every orbit closure span⁡‾{Tnx:n≥0}\overline{\operatorname{span}}\{T^nx:n\ge0\} with x≠0x\ne0 is then a nontrivial invariant subspace. Equivalently, the conjecture asks whether every bounded operator on ℓ2(N)\ell^2(\mathbb N) has a nonzero vector xx that is not cyclic, that is, with

span⁡‾{x,Tx,T2x,… }≠H.\overline{\operatorname{span}}\{x,Tx,T^2x,\dots\}\ne H .

The problem is to prove this statement, or to construct a bounded operator on a separable infinite-dimensional Hilbert space with no nontrivial closed invariant subspace. [1, 9, 10]

Background

The problem arose in the 1930s and 1940s from work of von Neumann and others on the structure of operators [9]. In finite dimensions the Jordan form decomposes an operator into invariant pieces. In infinite dimensions there may be no eigenvalues at all, and invariant subspaces are the natural substitute. A positive answer would give every operator a triangular form relative to a chain of invariant subspaces.

The question makes sense on any Banach space. There the answer is negative. Enflo announced a counterexample in the 1970s and published it in 1987 [3]. Read constructed counterexamples independently, including one on ℓ1\ell^1 [4, 5]. The Hilbert space case remains open, as does the case of reflexive Banach spaces. On some Banach spaces every operator has an invariant subspace. Argyros and Haydon built a space on which every operator is a scalar multiple of the identity plus a compact operator, so Lomonosov's theorem applies to all its operators [8].

References
  1. N. Aronszajn and K. T. Smith, "Invariant subspaces of completely continuous operators", Ann. of Math. 60 (1954), 345–350. doi:10.2307/1969637
  2. V. I. Lomonosov, "Invariant subspaces for the family of operators which commute with a completely continuous operator", Funct. Anal. Appl. 7 (1973), 213–214. doi:10.1007/BF01080698
  3. P. Enflo, "On the invariant subspace problem for Banach spaces", Acta Math. 158 (1987), 213–313. doi:10.1007/BF02392260
  4. C. J. Read, "A solution to the invariant subspace problem", Bull. London Math. Soc. 16 (1984), 337–401. doi:10.1112/blms/16.4.337
  5. C. J. Read, "A solution to the invariant subspace problem on the space ℓ1\ell_1", Bull. London Math. Soc. 17 (1985), 305–317. doi:10.1112/blms/17.4.305
  6. S. W. Brown, "Some invariant subspaces for subnormal operators", Integral Equations Operator Theory 1 (1978), 310–333. doi:10.1007/BF01682842
  7. S. W. Brown, B. Chevreau and C. Pearcy, "On the structure of contraction operators. II", J. Funct. Anal. 76 (1988), 30–55. doi:10.1016/0022-1236(88)90047-X
  8. S. A. Argyros and R. G. Haydon, "A hereditarily indecomposable L∞\mathcal L_\infty-space that solves the scalar-plus-compact problem", Acta Math. 206 (2011), 1–54. doi:10.1007/s11511-011-0058-y
  9. H. Radjavi and P. Rosenthal, Invariant Subspaces, Springer (1973). doi:10.1007/978-3-642-65574-6
  10. I. Chalendar and J. R. Partington, Modern Approaches to the Invariant-Subspace Problem, Cambridge Univ. Press (2011). doi:10.1017/CBO9780511862434
  11. A. R. Bernstein and A. Robinson, "Solution of an invariant subspace problem of K. T. Smith and P. R. Halmos", Pacific J. Math. 16 (1966), 421–431. doi:10.2140/pjm.1966.16.421
  12. P. H. Enflo, "On the invariant subspace problem in Hilbert spaces", preprint (2023). arXiv:2305.15442
  13. Wikipedia, Invariant subspace problem.

Source: Wikipedia

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.