A ``Tanaka formula'' in harmonic analysis and some applications (Q1272803)

The main theorem in this article is inspired by a ``Tanaka formula'' in the theory of martingales (it gives an expression for \(| f| \)) and the author gives some applications of this formula for functions in \(H^1\) and \(L\log L\). To describe this formula, let \(P(f)\) be the Poisson integral of a function in \(\mathbb R^n\), which is a harmonic function and \(| P(f)-r| \) is a subharmonic function for all real numbers \(r\). Define the positive function \(D_*^r(f)\) by \[ D_{*}^r(f)(\xi)= \int_{\mathbb R^{n+1}_+}yp_\xi(z)\Delta| P(f)-r| (dz) \] where \(p_\xi\) is the Poisson kernel, and \(\mathcal{M}\) is the set of measurable \(f\) for which \(p_\xi(0,1)f(\xi)\in L^1\). The main result here is that if \(f\in\mathcal{M}\) is such that \(D_*^0(f)\not\equiv \infty\) then \[ | f| =\widetilde{f}+D_*^0(f) \] where \(D_*^0(f)\in\mathcal{M}\) and \(\widetilde{f}\) has the same properties that \(f\) has. If \(f\) belongs to either \( L^1\), \(L^p\), \(H^1\) or \(\text{BMO}\cap\mathcal{M}\) then \(\widetilde{f}\) belongs to the same space, and if \(f\in L^1\) and \(B_1\), \(B_1\) are balls such that \(\overline{B_1}\subset B_2\) then \(N(f)\in L^1\) implies \(N(\widetilde{f})\in L^1\), where \(N(f)\) is the usual nontangential maximal operator.