High energy resolvent estimates for generalized many-body Schrödinger operators (Q2638440): Difference between revisions

Let \(X={\mathbb{R}}^ m\), \(H_ 0=-\Delta\) on \(D(H_ 0)=H^ 2(X)\), \(\pi_ j\) \((j=1,2,...,N)\) be orthogonal projections in X, \(X_ j=\pi_ j(X)\), and \(\nabla_ j\) the gradient in \(X_ j\). Let \(V_ j\in C^ n(X_ j)\), \(n\geq 1\), such that \[ (1)\quad \sup_{x_ j\in X_ j}\sum^{n}_{\ell =0}| ((x_ j\cdot \nabla_ j)^{\ell}V_ j)(x_ j)| <\infty \] holds. Let \(V(g)=\sum^{N}_{j=1}g_ jV_ j(\pi_ j(x))\) for \((g_ 1,...,g_ N)\in {\mathbb{R}}^ N_+\), \(H=H_ 0+V(g)\), and \(R(z;V(g))=(H-z)^{-1}\) denote the resolvent. If \(V_ j(x_ j)\to 0\) as \(| x_ j| \to \infty\), then H is called a generalized many-body Schrödinger operator. Let \(L^{2,s}(X)=\{f\in L^ 2_{loc}(X)|\| (1+x^ 2)^{s/2} f\|_{L^ 2}<\infty \}\), \({\mathcal B}(s,s')\) the space of bounded operators from \(L^ s(X)\) to \(L^{s'}(X)\), and \(\mu (V_ j)=\max \{0,\sup_{x_ j\in X_ j}(V_ j(x_ j)+x_ j\nabla_ jV_ j(x_ j)/2)\}.\) Theorem. Suppose that \(V_ j\in C^ n(X_ j)\) satisfy the condition (1) for some \(n\geq 2\). Then there exists \(\lambda_ 0>0\) such that the following (i) and (ii) hold for all \(\lambda \geq \lambda_ 0\) and for all \(g\in {\mathbb{R}}^ N_+\) satisfying \(``\lambda \geq C_ 0\max \{g_ j\}\) and \(\lambda\geq \delta \sum^{N}_{j=1}g_ j(V_ j)\) for some \(C_ 0>0\) and \(\delta >1'':\) (i) \(R(\lambda \pm i0;V(g))=\lim_{\epsilon \downarrow 0}R(\lambda +i\epsilon;V(g))\) exist in B(s,-s) for \(s>1/2\) in operator norm. (ii) The map \(\lambda \to R(\lambda \pm i0;V(g))\) is k times continuously differentiable from \((\lambda_ 0,\infty)\) to \({\mathcal B}(s,-s)\) for \(s>k+1/2\), and the high energy estimate \[ (2)\quad \| (d/d\lambda)^ kR(\lambda \pm i0;V(g))\|_{{\mathcal B}(s,-s)}\leq C\lambda^{-(k+1)/2} \] holds, where \(k=0,1,...,n-2.\) An extension of the Mourre's commutator method is used for the proof, since the parameter dependence can be followed precisely. Finally the author treats \(H(\hslash)=\hslash^ 2H_ 0+V(g)\) and \(H=-\Delta +c| x|^{-\nu}\), \(c\in {\mathbb{R}}\), \(0<\nu <\min \{m/2,2\}\), and gives similar considerations.