Stability of near-extremizers for the high-entropy increment

Characterize the additive structure forced when the high-entropy increment H(X+Y) - (H(X)+H(Y))/2 approaches 1/2, including the effects of additive embeddings into the integers and near-equality cases of the continuous entropy power inequality.

Background

The paper proves that the asymptotically sharp additive constant in the entropy lower bound is 1/2 for independent discrete random variables in torsion-free abelian groups and, under an entropy-deficit condition, in prime cyclic groups. Binomial distributions attain this constant asymptotically, but the authors note that this example may not describe all near-extremizers.

The proof reduces finite-support distributions to integer-valued variables through additive embeddings that preserve only finitely many additive relations. Consequently, a stability theorem would need to identify which additive structures are necessarily present when the entropy increment is close to the optimal value, while also incorporating near-equality structure from the continuous entropy power inequality.

References

The proof also leaves a stability question: what additive structure is forced when the high-entropy increment approaches $1/2$? The binomial example is one possibility, but the reduction to integers preserves only finitely many additive relations and does not identify a canonical geometry for near-extremizers. A stability theorem would have to account for these additive embeddings as well as the near-equality cases of the continuous entropy power inequality.

— Sharp High-Entropy Bounds for Sums of Independent Discrete Random Variables  (2609.21459 - Wang, 18 Sep 2026) in Section 6, Discussion