Dice Question Streamline Icon: https://streamlinehq.com

Existence of a flat deformation whose all sufficiently small specializations are non-semisimple

Determine whether there exist finite-dimensional complex algebras N and A of the same dimension, with A semisimple, and a formal flat deformation \mathcal N of N such that \mathcal N specializes to an algebra isomorphic to A at t = s0 for some real s0 > 0, yet for every sufficiently small real s > 0 the specialized algebra \mathcal N at t = s is not semisimple (equivalently, has a nonzero nilpotent ideal).

Information Square Streamline Icon: https://streamlinehq.com

Background

The paper considers finite-dimensional complex algebras N and A of the same dimension and formal deformations \mathcal N of N. A deformation is called strongly flat if arbitrarily small positive specializations yield an algebra isomorphic to a fixed semisimple algebra A. Using the finiteness of isomorphism types of semisimple algebras of a fixed dimension, the authors observe that if a flat deformation \mathcal N deforms to A at some parameter value, then either \mathcal N is strongly flat (case (i)) or, alternatively, for all sufficiently small parameters the specializations are not semisimple (case (ii)).

While the paper proves that strongly flat deformations stabilize to A for all sufficiently small parameters, it explicitly notes that it is unknown whether the alternative case (ii) can occur at all. Establishing the existence or non-existence of such deformations would clarify the landscape between flat and strongly flat deformations in finite-dimensional settings.

References

It follows that if $\mathcal N$ is a flat deformation of $N$ which deforms to $A$ at $t=s$ then we have the following possibilities (i) $\mathcal N$ is a strongly flat deformation from $N$ into $A'$, for some semisimple $\mathbb C$-algebra $A'$, where $A'$ is uniquely determined by $\mathcal N$, (ii) for all sufficiently small real numbers $s>0$ the algebra $\mathcal N$ specialised at $t=s$ is not semisimple, and hence it has a non-zero nilpotent ideal. We don't know if there are $N,A$ and $\mathcal N$ such that (ii) holds.

On flat deformations and their applications (2509.10121 - Smoktunowicz, 12 Sep 2025) in Introduction (Section 1), after listing cases (i) and (ii)