Ball–Nayar–Tkocz entropy concavity conjecture

Let XX and YY be independent identically distributed real-valued random variables with a common log-concave density and finite differential entropy. Define F:[0,1]→RF:[0,1]\to\mathbb{R} by F(t)=h(1−t X+t Y)F(t)=h\bigl(\sqrt{1-t}\,X+\sqrt{t}\,Y\bigr). The Ball–Nayar–Tkocz entropy concavity conjecture asserts that FF is concave on [0,1][0,1]; equivalently, for all s,t∈[0,1]s,t\in[0,1] and λ∈[0,1]\lambda\in[0,1], F((1−λ)s+λt)≥(1−λ)F(s)+λF(t)F\bigl((1-\lambda)s+\lambda t\bigr)\ge (1-\lambda)F(s)+\lambda F(t).

References

Primary source

arXiv

Progress summary

Refreshed
Claimed progress

A 2026 preprint gives a counterexample when the variables are not symmetric, while the symmetric case remains open.

Ball, Nayar, and Tkocz conjectured in 2015 that entropy along normalized sums of independent identically distributed log-concave variables is concave. The new result addresses the original nonsymmetric formulation, not its symmetry-restricted variant.

Known results

  • The 2015 source records the conjecture and proves only a weakened related projection-entropy inequality, with constant κ=1/5\kappa=1/5 rather than the conjectured κ=1\kappa=1.

September 2026 asymmetric counterexample

Congyi Luo’s preprint Entropy concavity for log-concave random variables: an asymmetric counterexample reports a counterexample to the conjecture without symmetry. It therefore refutes the unrestricted formulation if confirmed, while leaving the symmetric log-concave case open; the claim has not been independently verified here.

Current status (as of September 2026): the unrestricted conjecture is claimed refuted by an unverified counterexample, while the symmetry-restricted version remains open.

Sources

Solutions 0

No solutions have been posted yet.