On the ratio of the expected maximum of a martingale and the \(L_ p\)- norm of its last term (Q1113189): Difference between revisions

A tight bound is derived for the expected supremum of a non-negative submartingale in terms of the \(L^ p\)-norm of the last term of the submartingale, improving the bound given in Theorem 3.4 of \textit{J. Doob}, Stochastic processes (1953; Zbl 0053.268). Further a tight bound is given in terms of the \(L_ p\)-norm of the last term for the expected maximum of a zero-mean martingale which starts at zero. For both of these bounds examples are give where the bounds are attained. En route to these results it is shown that if Y is a random variable with an exponential distribution then the value of t which minimizes \(\| Y- t\|_ p\) is the minimum of \(\| Y-t\|_ p\).