← All problems

Closedness of bounded strength and partition-rank loci

Let KK be an algebraically closed field. For integers d≥2d\ge 2, r≥0r\ge 0, and a finite-dimensional KK-vector space VV, define the strength locus Sd,r(V)⊆Sym⁡d(V∗)S_{d,r}(V)\subseteq \operatorname{Sym}^d(V^*) by Sd,r(V)={f:f=∑i=1rgihi, deg⁡gi,deg⁡hi≥1, deg⁡gi+deg⁡hi=d}S_{d,r}(V)=\{f:f=\sum_{i=1}^r g_i h_i,\ \deg g_i,\deg h_i\ge 1,\ \deg g_i+\deg h_i=d\}. For integers k≥2k\ge 2 and r≥0r\ge 0, and finite-dimensional vector spaces V1,…,VkV_1,\ldots,V_k, define the partition-rank locus Pk,r(V1,…,Vk)P_{k,r}(V_1,\ldots,V_k) to consist of tensors expressible as a sum of at most rr tensors, each of which factors across some nontrivial bipartition of {1,…,k}\{1,\ldots,k\}. Determine, for all d,r,Vd,r,V and all k,r,V1,…,Vkk,r,V_1,\ldots,V_k, whether Sd,r(V)S_{d,r}(V) and Pk,r(V1,…,Vk)P_{k,r}(V_1,\ldots,V_k) are Zariski closed in their respective ambient spaces. In particular, determine whether there exists a quartic form of strength exactly 33 in the Zariski closure of the locus of quartics of strength at most 22, and determine the corresponding closedness behavior for partition rank.

References

Source: arXiv

Progress summary

Refreshed
Open · progress reported

An unrefereed preprint claims a new counterexample and a broad classification, but no independent verification has appeared.

The problem asks when bounded-strength and partition-rank loci are Zariski closed. The central new claim concerns a quartic of strength 33 lying in the closure of the strength-22 locus, with suspension arguments extending the failure pattern.

Known results

  • Bounded strength is closed for cubics; arbitrary-degree closedness was previously open (Ballico, Bik, Oneto, and Ventura, 2018).
  • For quartics in sufficiently many variables, the strength-33 locus is not closed (2021); the minimal threshold and variable count were left open.
  • A survey records that strength is not Zariski closed and that partition-rank loci were already believed nonclosed for tensors of order 44.

October 2026 claimed classification

Benjamin Baily and Alessandro Danelon claim that a quartic of strength 33 lies in the closure of strength at most 22, and that suspension classifies closedness failures across all rank bounds. The result is reported only in an unrefereed preprint and remains unverified.

Current status (as of October 2026): Earlier nonclosedness results are established, while the claimed strength-22 closure example and all-threshold classification remain unverified.

Sources

Solutions 0

No solutions have been posted yet.