The spectral flow for Dirac operators on compact planar domains with local boundary conditions (Q358354)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | The spectral flow for Dirac operators on compact planar domains with local boundary conditions |
scientific article |
Statements
The spectral flow for Dirac operators on compact planar domains with local boundary conditions (English)
0 references
8 August 2013
0 references
Let \(X\) be a compact planar domain bounded by \(m\) smooth curves (thus \(X\) is homeomorphic to a disk with \(m-1\) holes). The author proves a formula for the spectral flow of paths connecting Dirac type operators on \(X\) conjugated by a scalar gauge transformation, and so that all operators of the path have the same symbol and the same local boundary condition. Three theorems are stated in the paper giving the same formula, but with increasing degree of generality. Let us state the theorem for the simplest case. Spinor-valued functions on \(X\) can be considered as functions \(u:X\to\mathbb{C}^2\), with components \(u^\pm\). A path of Dirac operators on \(X\) is of the form \(D_t=\mathbb{D}+Q_t\), \(t\in[0,1]\), where \(\mathbb{D}=(\mathbb{D}_{kl})\) with \(\mathbb{D}_{11}=\mathbb{D}_{22}=0\), \(\mathbb{D}_{12}=-i(\partial_1-i\partial_2)\) and \(\mathbb{D}_{21}=-i(\partial_1+i\partial_2)\), and \(Q_t\) is a smooth function on \(X\), depending continuously on \(t\), whose values are Hermitian \(2\times 2\) matrices. Assume that \(D_1=\mu D_0\mu^{-1}\) for some smooth scalar gauge transformation \(\mu:X\to U(1)\). All operators \(D_t\) are considered with the boundary condition \(i(n_1+in_2)u^+=Bu^-\), where \(n=(n_1,n_2)\) is the outward conormal to the boundary \(\partial X\), and \(B:\partial X\to\mathbb{R}\) is a smooth non-vanishing function. The operators \(D_t\) with this boundary condition define a path of unbounded self-adjoint operators in \(L^2(X,\mathbb{C}^2)\) with discrete spectrum, which is continuous with respect to the graph topology. If \(D_0\) and \(D_1\) are invertible, the spectral flow of \(D_t\), with this boundary condition, is defined as the number of eigenvalues that pass zero from the negative side to the positive side as \(t\) goes from \(0\) to \(1\), minus the number of eigenvalues that pass zero in the other direction, taking multiplicity into account. If zero is in the spectrum of \(D_0\) and \(D_1\), a simple trick still allows to define the spectral flow. The author shows that this spectral flow is equal to \(c_m\sum_{j=1}^mb_j\mu_j\), where \(c_m\in\mathbb{Z}\) depends only on \(m\), \(\mu_j\) is the degree of the restriction of \(\mu\) to each component \(\partial X_j\) of \(\partial X\), and \(b_j\) is equal to \(1\) or \(0\) if \(B\) is positive or negative on \(\partial X_j\), respectively. In the next extension, the functions \(u^\pm\) have values in \(\mathbb{C}^N\), and the values of \(B\) are complex self-adjoint invertible \(N\times N\) matrices. Then the formula of the spectral flow is equal, the only difference being that the definition of \(b_j\) takes into account whether \(B\) is positive or negative definite. In the most general result, the author considers general Dirac type operators for arbitrary non-flat metrics whose principal symbol may not be given by the metric. The resulting formula is the same. The author also proves that \(c_2=1\), and conjectures that \(c_m=1\) for all \(m\). This conjecture was recently proved in a preprint of Katsnelson and Nazaikinski.
0 references
Dirac operator
0 references
spectral flow
0 references
planar domain
0 references
local boundary condition
0 references
0 references
0 references