Variational inequalities for singular integrals in the Dunkl setting (Q6950466)

In this paper, one considers the variation operator of a family of truncated operators of a singular integral and a space in the Dunkl setting, where the integral kernel has estimations involving two different metrics, namely, the Euclidean and the Dunkl metrics. Assuming the boundedness of the variation operator on some space \(L^{(p_0)}\) with respect to the Dunkl measure one obtains its boundedness in all \(L^p\) spaces, the weak boundedness from \(L^1\) to \(L^{(1,\infty)}\) and from the space of bounded measurable functions with compact suport to the BMO space. On the Euclidean space \(R^N\), there is exactly one weight function \(\mu(x)\) associated with a normalized root system \(R\) and a multiplicity function \(k(\alpha)\geq 0\) for any \(\alpha\in R\) such that the Dunkl measure is defined by\N\[\Nd\mu(x)=\prod_{(\alpha\in R)}|<\alpha,\ x>|^(k(\alpha))dx,\N\]\Nwhere \(dx\) denotes the Lebesgue measure in \(R^N\). If \(\alpha\) is a nonzero vector in \(R^N\) then \(s_\alpha\) will denote the reflection with respect to the hyperplane orthogonal to \(\alpha\) and the set of reflections \(s_\alpha\) for \(\alpha\) in \(R\) generates a finite group \(G\), the Weyl group of the root system \(R\). Besides the Euclidean metric \(||x-y||\), one considers in \(R^N\) the Dunkl metric \(d(x,y)\) defined as \(d(x,y)=\min_{(s\in G)}||x-s(y)||\). The kernel \(K(x,y)\) of a singular integral operator is called a Dunkl-Calderón-Zygmund kernel if it satisfies the following estimates, for some \(\varepsilon>0\),\N\Ni) \(|K(x,y)|\leq 1/(\mu(B(x,d(xy)).(d(x,y)/||x-y||)^\varepsilon\),\N\Nii) \(|K(x,y)-K(x^\prime,y)|\leq(||x- x^\prime||/||x- y||)^\varepsilon.1/\mu(B(x, d(x,y))\), for all \(||x-x^\prime||\leq d(x,y)/2\)\N\Niii) \(|K(x,y)-K(x,y^\prime)|\leq (||y-y^\prime||/ ||x-y||)^\varepsilon.1/\mu(B(x,d(x,y))\) for all \(||y-y'||\leq d(x,y)/2\).\N\NFor \(p\in [1,\infty)\) denote \(L^p(R^N,\mu)\) by \(L^p(\mu)\) and for \(f\in L^p(\mu)\) and \(t>0\), let\N\[\N_t(f)(x)=\int_d(x,y)>t K(x,y)f(y)d\mu(y).\N\]\NFor any \(q>2\), the variation operator \(V_qK\) is\N\[\NV_q Kf(x)=\sup_{(\varepsilon_i \lozenge o)}(\sum_i|K_{(\varepsilon_i)}f(x)-K_(\varepsilon_{(i+1)})f(x)|^q)^{(1/q)},\N\]\Nwhere the supremum is taken over all sequences \(\{\varepsilon_i\}\) that are decreasing and converge to zero. The main result is the following one: Let \(q\in (2,\infty)\) and assume that the variation operator \(V_q K\) is bounded from \(L^{(p_0)}\) to \(L^{(p_0)}\) for some \(p_0\in (1,\infty)\), where \(K\) is a Dunkl-Calderón-Zygmund kernel. Then we have the following conclusions\N\Ni) \(V_qK\) is of weak type from \(L^{(1)}(\mu)\) to \(L^{(1,\infty)}(\mu)\).\N\Nii) For any \(p\in (1,\infty)\), \(V_qK\) is bounded from \(L^p(\mu)\) to \(L^p(\mu)\).\N\Niii) \(V_qK\) is bounded from the space of bounded \(\mu\)-measurable functions with compact support to the BMO space with respect to the measure \(\mu\). \N\NA similar result is obtained for so-called oscillation operators.