On the spectrum of the two-dimensional periodic Dirac operator (Q1568015): Difference between revisions

The spectrum of the Dirac operator \(\widehat D\) in \(\mathbb R^2\) is investigated, \[ \widehat D=\widehat D_0+V- \sum_{j=1} ^2 A_j \widehat \alpha_j=\sum_{j=1} ^2 \widehat \alpha \Big(-i\frac{\partial}{\partial x_j}-A_j \Big) +\widehat \beta m +V, \] where \(m \in \mathbb R\) and \(\widehat \alpha\) and \(\widehat \beta\) are Hermitian \(M \times M\) matrices satisfying the anticommutation relations \[ \widehat \alpha_i \widehat \alpha_j+\widehat \alpha_j\widehat \alpha_i=2\delta_{ij}\widehat I,\quad \widehat \alpha_i \widehat \beta +\widehat \beta\widehat \alpha_i=\widehat 0, \quad \widehat \beta^2=\widehat I. \] Here \(\widehat \alpha_i\) and \(\beta_i\) are for example Dirac and Pauli matrices or their generalizations. The scalar electric potential \(V\) and vector magnetic potential components \(A_j\) are periodic functions with the periodic lattice \(\Lambda \subset \mathbb R^2\). Let \(E_j\) is a basis of \(\Lambda\). We set \(K=\mathbb R^2/\Lambda\). The following theorems are proved. Let \(\gamma\) be a vector in \(\Lambda \setminus{0}\). We define the unit vector \(e\in \mathbb R^2\) as \(e=| \gamma| ^{-1}\gamma\). Theorem 1. Let \(V, A_j: \mathbb R^2 \to \mathbb R\) be periodic functions with period lattice \(\Lambda \subset \mathbb R^2\) such that \(V, A_j \in L_{loc} ^q (\mathbb R^2)\), \(j=1,2, \quad q>2\). Then the spectrum of the Dirac operator \(\widehat D\) is absolutely continuous. To prove this theorem the author uses the \textit{L. E. Thomas} method [Commun. Math. Phys. 33, 335--343 (1973)] in which the invertibility of the operator \(\widehat D(k)=\widehat D_0(k)+V-A_1 \widehat\alpha_1 -A_2\widehat\alpha_2\) and the estimates of the norm of the inverse operators \(\widehat D(k+i\kappa e)-\lambda\), \(\lambda \in \mathbb R, \kappa \in \mathbb R\) for complex valuse of quasi momentum \(k+i\kappa e \in \mathbb C^2\) are used. Theorem 2. Let the functions \(V, A_j: K \to \mathbb C\), \(j=1,2\), belong to the space \(L^q(K),\quad q>2\), and \(\lambda \in \mathbb C\). Then for any \(\theta \in [0,1)\), there exists a sufficiently large number \(\kappa>0\) such that for all \(k\in \mathbb R^2\) satisfying the condition \((k-a,e)=\pi| \gamma| ^{-1}\) and for all vector functions \(\phi \in H^1(K;\mathbb C^M)\), there is the inequality \[ | | (\widehat D_0(k+i\kappa e)+V-A_1 \widehat \alpha_1-A_2 \widehat \alpha_2- \lambda) \phi| | \geq \theta \frac{\pi}{| \gamma| }exp \Big( -c_1 \sum_{j=1} ^2 | | A_j-a_j| | _{L^q(K)} \Big) | | \phi| | , \] where the constant \(c_1(q,\Lambda)>0\) depends only on \(q\) and on the lattice \(\Lambda\).