We present a flexible framework for predicting error measures in multiple testing situations under dependence. Our approach is based on modeling the distribution of the probit transform of the p-values by mixtures of multivariate skew-normal distributions. The model can incorporate dependence among p-values and it allows for shape restrictions on the p-value density. A nonparametric Bayesian sc...

Multiple testing methods to control the False Discovery Rate (FDR), the expected proportion of falsely rejected null hypotheses among all rejections) have received much attention. It can be valuable instead to control not the mean of this false discovery proportion (FDP) but the probability that the FDP exceeds a specified bound. In this paper, we construct a general class of methods for exceed...

At each failure of a system, with probability p it is perfectly repaired, and with probability 1 − p it is restored to its condition just prior to failure. It is shown that there exists a unique p minimizing the long-run repair cost, provided the failure time distribution of the system has strictly DMRL.

Very little is known about the asymptotic behaviour of classes of matroids. We make a number of conjectures about such behaviours. For example, we conjecture that asymptotically almost every matroid: has a trivial automorphism group; is arbitrarily highly connected; and, is not representable over any field. We prove one result: The proportion of labeled n-element matroids that are connected is ...

Abstract: Real options approaches can be employed to value the flexibility in the decision-making courses of phased investments. This paper focuses on the two-stage investment problems of venture firms. Firstly, based on the analysis of profit function and real options thinking, stochastic models are proposed to describe the uncertainties of such problems. Then the value function of decision-ma...