Log-Concavity of Multinomial Likelihood Functions Under Interval Censoring Constraints on Frequencies or Their Partial Sums

This provides a theoretical foundation for statistical inference under censoring, but it is incremental as it extends known log-concavity results to more general constraints.

The paper tackled the problem of analyzing the likelihood function for multinomial vectors under interval censoring constraints, proving that it is completely log-concave by showing the constrained sample spaces are M-convex subsets of the discrete simplex.