Bézout-domain elementary-divisor problem
Does every commutative Bézout domain have the property that every matrix admits a Smith normal form; that is, do there exist invertible matrices and such that is diagonal, with nonzero diagonal entries satisfying ? Equivalently, is every commutative Bézout domain an elementary divisor domain?
References
Primary source
Additional references
- A Bézout domain that is not an elementary divisor domain — arXiv — Christian Hägg, Anders Mörtberg
Progress summary
A September 2026 paper claims to disprove the statement that every ring of this type has Smith normal forms, but the result has not been independently verified.
The problem asks whether every commutative Bézout domain is an elementary divisor domain, meaning that matrices over it admit Smith normal form. Earlier literature treated this as open; a recent paper claims a counterexample arising from the Möbius line bundle.
Known results
- Bézout domains of Gelfand range are elementary divisor domains, including -domains and local Gelfand domains (2015).
- For semihereditary Bézout rings, the Shores criterion reduces the question to the factor rings (2012).
- Bézout domains with all maximal ideals principal are elementary divisor domains; further criteria use (2015).
- Feckly zero-adequate Bézout rings are elementary divisor rings (2015).
September 28, 2026 counterexample claim
Christian Hägg and Anders Mörtberg report that a topological obstruction from the Möbius line bundle produces a Bézout domain that is not an elementary divisor domain. If correct, this settles the unrestricted question negatively and explains a concrete failure of Smith normal form; the claim is presently unverified.
Current status (as of September 2026): A counterexample has been claimed, while the earlier positive results for restricted classes remain established and independent verification of the claimed counterexample is pending.
Solutions 0
No solutions have been posted yet.