Comon's conjecture on tensor rank versus symmetric rank
Let be a vector space and let . Let be the symmetric rank of , and let be its tensor rank. Comon's conjecture.
The supplied text states this as Comon's conjecture and gives no resolution.
References
- Jarosław Buczyński, Adam Ginensky and J. M. Landsberg, “Determinantal equations for secant varieties and the Eisenbud-Koh-Stillman conjecture”, arXiv:1007.0192 (2012).
Progress summary
A published counterexample overturned the conjecture, although a gap in the original proof means this summary does not certify the result.
Comon's conjecture asserted that every symmetric tensor has equal tensor rank and symmetric rank. Shitov's construction disproved this over the complex numbers; analogous real counterexamples followed.
Known results
- Equality holds when tensor rank is at most the tensor order (Qi, 2016).
- Equality holds for symmetric rank at most (Comon and collaborators, 2015).
- Over , equality holds for symmetric rank at most (Comon and collaborators, 2015).
- Further positive cases include tangential Veronese tensors and several bounded-rank families (Alex et al., 2018).
2018–2026 counterexamples and repair history
Shitov's 2018 example has format , tensor rank , and symmetric rank at least . A 2024 erratum identified a gap in the published proof, while a repaired construction was circulated. Shitov later gave a real counterexample, and Seigal and Wang reported an order- real example in 2023. A 2026 preprint claims a smaller rational example with ranks and , but that claim is unverified.
Current status (as of October 2026): The original equality conjecture is refuted by claimed complex and real counterexamples, while the proof repair and the 2026 smaller example remain unverified; the border-rank analogue and minimality questions remain open.
Solutions 0
No solutions have been posted yet.