Ancheta–Massey problem
Let have independent coordinates distributed as , with , and let distortion be normalized Hamming distortion. Define as the infimum of the asymptotic rates over schemes whose encoder is linear, whose decoder is arbitrary, and whose expected normalized Hamming distortion is at most . The Ancheta–Massey conjecture asserts that $
\nR_{\mathrm{lin},p}(D)=\begin{cases}1-D/p,&0\le D\le p,\\0,&D\ge p.\end{cases} $Equivalently, the optimal linear scheme is obtained by encoding a fraction of the source bits losslessly and reproducing the remaining fraction as zero. The claim for is due to Ancheta; the extension to every is the problem addressed by the supplied note.
References
Primary source
Additional references
Progress summary
An unrefereed note claims to settle the biased-source extension of the Ancheta–Massey problem for every Bernoulli parameter below one-half, but the result has not been independently verified.
The problem asks for the conjectured optimal linear-encoding rate for biased Bernoulli sources. The newly reported note claims the extension for every parameter .
August 24, 2026 note
The note Entropy of Bernoulli Measures Conditioned on Affine Subspaces and a Problem of Ancheta--Massey claims a proof of the conjectured optimal rate for all , which would resolve the stated biased-source range. Its proof was reportedly discovered interactively with GPT-5.6 Sol; the preprint is unrefereed, so the resolution remains unconfirmed.
Current status (as of August 2026): The conjectured rate for every is claimed proved in an unrefereed preprint, while independent verification remains outstanding.
Ancheta–Massey problem extended to biased Bernoulli sources
Solutions 0
No solutions have been posted yet.