Strong differential subordinates for noncommutative submartingales (Q2280551)

Let \(f=(f_n)_{n\geq0}\) and \(g=(g_n)_{n\geq0}\) be two adapted sequences of integrable random variables with the corresponding differences \(df =(df_n)_{n\geq0}, dg =(dg_n)_{n\geq0}\) given by \(df_0 =f_0\) and \(df_n = f_n-f_{n-1}\) for \(n\geq1\) (with an analogous formula for \(dg\)). Then \(g\) is said to be differentially subordinate to \(f\) if, for any \(n\geq0\), we have \(|dg_n|\leq|df_n|\) almost surely, and \(g\) is conditionally differentially subordinate to \(f\) if, for any \(n\geq1\), we have \(|\mathbb{E}_{n-1}(dg_n)|\leq|\mathbb{E}_{n-1}(df_n)|\) almost surely. Working under this assumption, \textit{D. L. Burkholder} [Ann. Probab. 12, 647--702 (1984; Zbl 0556.60021)] proved some sharp weak-type \((1, 1)\) and strong-type \((p,p)\) estimates. In the present paper, the authors introduce a notion of strong differential subordination of noncommutative semimartingales, extending \textit{D. L. Burkholder}'s definition [Ann. Probab. 22, No. 2, 995--1025 (1994; Zbl 0816.60046)] from the classical case, and establish the maximal weak-type \((1,1)\) inequality under the additional assumption that the dominating process is a submartingale. Furthermore, they also show that the corresponding strong-type \((p,p)\) estimate for \(1 < p <\infty\) holds under the assumption that the dominating process is a nonnegative submartingale.