The Comon conjecture asserts that the tensor decomposition rank and symmetric tensor rank
of a symmetric tensor are equal (Comon et
al. 2008). Writing these ranks as and
, respectively, one always has
because every symmetric decomposition is also
an unrestricted tensor decomposition.
Shitov's (2018) proposed counterexample over the complex numbers contained an error in the lower
bound on its symmetric tensor rank (Draisma
2024). Lovitz (2026) constructed a smaller symmetric
tensor
with rational entries of size
such that
and
over both the real
and complex numbers, disproving the conjecture
over either field. Li (2026) proved that the tensor
direct sum of two copies has
while three copies have
These tensors have sizes and
, respectively. Consequently, the difference
can be 2 or 3, not only 1.
Lovitz (2026) reports using GPT-6 Astra throughout the project, including to obtain an initial proof that he subsequently verified and simplified. Li (2026) reports using ChatGPT throughout the mathematical development and manuscript preparation and takes responsibility for the paper.