On the eigenvalues of operators with gaps. Application to Dirac operators (Q1579199)

Let \(H\) be a Hilbert space and \(A: D(A)\subset H\to H\) a self-adjoint operator. We denote by \({\mathfrak I}(A)\) the form domain of \(A\). Let \(H_+\), \(H_-\) be two orthogonal Hilbert spaces such that \(H= H_+\oplus H_-\). We denote \(\Lambda_+\), \(\Lambda_-\)the projections on \(H_+\), \(H_-\). We assume the existence of a core \(F\), which is dense for the norm \(\|.\|_{D(A)}\), such that: (i) \(F_+= \Lambda_+F\), \(F_-= \Lambda_-F\) are two subspaces of \({\mathfrak I}(A)\). (ii) \(a= \sup_{x_-\in F_--\{0\}}(x_-, Ax_-)/\|x_-\|^2_H< +\infty\). We consider the sequence of min-max levels \(\lambda_k= \inf_{V\subset F_+,\dim V=k}\sup_{x\in (V\oplus F_-)- \{0\}} {(x, Ax)\over\|x\|^2_H}\), \(k\geq 1\). (iii) \(\lambda_1> a\). Now let \(b= \inf(\sigma_{\text{ess}}(A)\cap (a,+\infty))\in [a,+\infty]\). For \(k\geq 1\), we denote by \(\mu_k\) the \(k\)th eigenvalue of \(A\) in the interval \((a,b)\), counted with multiplicity, if this eigenvalue exists. If there is no \(k\)th eigenvalue, we take \(\mu_k= b\). The main result of this paper is: Theorem: With the above notation, and under assumptions (i), (ii), (iii), \(\lambda_k= \mu_k\), \(\forall k\geq 1\). As a consequence, \(b=\lim_{k\to \infty}\lambda_k= \sup_k \lambda_k> a\). The authors apply this theorem to the case of Dirac operator with a Coulomb-like potential. The result is optimal for the Coulomb potential.