Strong log-concavity holds for probit regression likelihoods on fixed designs and Gaussian designs when d/n is sufficiently small, with the condition number finite and asymptotically independent of the ratio.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.ST 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Strong log-concavity in probit regression
Strong log-concavity holds for probit regression likelihoods on fixed designs and Gaussian designs when d/n is sufficiently small, with the condition number finite and asymptotically independent of the ratio.