Open-ended tests for Koopman-Darmois families

摘要: The generalized likelihood ratio is used to define a stopping rule for rejecting the null hypothesis θ = θ0 in favor of > θ0. Subject bound α on probability ever case θ0, expected sample sizes are minimized within multiple log α-1, depending θ. An heuristic error procedure derived and verified normal mean by consideration Wiener process. Useful lower bounds small-sample efficiency thereby obtained.