Vortex dynamics in the presence of excess energy for the Landau-Lifshitz-Gilbert equation

DOI10.1007/S00526-013-0609-5MaRDI QIDQ2445453zbMATH OpenOpenAlexWikidataFDO

Authors Christof Melcher, Roger Moser, Matthias Kurzke, Daniel Spirn

Publication date 14 April 2014

Published in Calculus of Variations and Partial Differential Equations (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1203.4426

renormalized energy thin ferromagnetic film

Singular perturbations in context of PDEs (35B25) Ginzburg-Landau equations (35Q56) PDEs in connection with optics and electromagnetic theory (35Q60) Statistical mechanics of magnetic materials (82D40)

Abstract: We study the Landau-Lifshitz-Gilbert equation for the dynamics of a magnetic vortex system. We present a PDE-based method for proving vortex dynamics that does not rely on strong well-preparedness of the initial data and allows for instantaneous changes in the strength of the gyrovector force due to bubbling events. The main tools are estimates of the Hodge decomposition of the supercurrent and an analysis of the defect measure of weak convergence of the stress energy tensor. Ginzburg-Landau equations with mixed dynamics in the presence of excess energy are also discussed.

Review text

Let \(\Omega\) be a domain in \(\mathbb R^3\). The present paper deals with the magnetization of a thin ferromagnetic film represented by \(\Omega\) (cf. the authors' paper [Arch. Ration. Mech. Anal. 199, No. 3, 843--888 (2011; Zbl 1229.35283)]). The magnetization is modeled by an equation of the form \[ (1) \qquad\frac{\partial \mathbf m}{\partial t}=\mathbf m\times\bigg(\alpha_{\epsilon}\frac{\partial\mathbf m}{\partial t}-f_{\epsilon}(\mathbf m)\bigg) \] associated with the initial condition \(\mathbf m(0)=\mathbf m^{0}_{\epsilon}\), and Dirichlet or Neumann boundary conditions on the boundary \(\partial\Omega\). Here, \(\mathbf m=(m_1,m_2,m_3):\Omega\to\mathbb S^2\) is the magnetization vector, the symbol \(\times\) denotes the vector product in \(\mathbb R^3\), \[ f_{\epsilon}(\mathbf m)=\Delta \mathbf m+|\nabla\mathbf m|^2\mathbf m-\frac{1}{\epsilon^2}(m_3 \mathbf e_3-m_3^2\mathbf m),\quad e_3=(0,0,1), \] and it is assumed that \(\alpha_{\epsilon} \log\frac{1}{\epsilon}\to \alpha_o\) as \(\epsilon\to 0.\) The authors study the asymptotic behavior of equation (1) as \(\epsilon\to 0\). In the limit, a time-dependent set of vortices, whose centers are denoted by \(a_1(t),\dots,a_N(t)\), are obtained. A number of quantities are associated to \(\mathbf m\), in particular, the energy density \[ e_{\epsilon}(\mathbf m)=\frac{1}{2}|\nabla\mathbf m|^2+\frac{1}{2\epsilon^2}m_3^2, \] the vorticity \[ \omega(\mathbf m)=\biggl\langle\mathbf m,\frac{\partial \mathbf m}{\partial x_1}\times \frac{\partial \mathbf m}{\partial x_2}\biggl\rangle, \] where the bracket stands for the scalar product in \(\mathbb R^3,\) and the planar Jacobian \[ J(\mathbf m)=\frac{\partial m_1}{\partial x_1}\frac{\partial m_2}{\partial x_2}-\frac{\partial m_1}{\partial x_2}\frac{\partial m_2}{\partial x_1}. \] The following assumptions are made on the initial data: There exists \(C_o\) such that \[ \int_{\Omega} e_{\epsilon}(\mathbf m_{\epsilon}^{o})dx\leq\pi N\log\frac{1}{\epsilon}+C_o,\text{ for every }\epsilon\in (0,\frac{1}{2}]. \] There exist \(a^{o}=(a_1^{o},\dots,a_N^{o})\in\Omega^N\), \(d=(d_1,\dots,d_N)\), with \(d_n=\pm1\) and \(q^{o}=(q^{o}_1,\dots,q^{o}_N)\in \big(\frac{1}{2}+\mathbb Z\big)^N\) such that \[ \frac{e_{\epsilon}(\mathbf m_{\epsilon}^{o})}{\log\frac{1}{\epsilon}}\to\pi \sum_{n=1}^{N}\delta_{a_n^{o}}, \quad J(\mathbf m_{\epsilon^{o}})\to\pi \sum_{n=1}^{N}d_n \delta_{a_n^{o}}, \quad \text{ and } \quad \omega(\mathbf m_{\epsilon}^{o})\to 4\pi \sum_{n=1}^{N} q_n^{o}\delta_{a_n^{o}}, \] in the distributional sense, as \(\epsilon\to 0.\) Under these conditions, it is proved that, for every \(\epsilon\in (0,\frac{1}{2}]\), there exists a weak solution \(\mathbf m_{\epsilon}\) of equation (1), with \(\mathbf m_{\epsilon}(0)=\mathbf m_{\epsilon}^{o} \) satisfying the required boundary condition on \(\partial\Omega.\) Moreover, there exist \(T>0,\) a piecewise constant function \(q:(0,T)\to (\frac{1}{2}+\mathbb Z)^N\), and a subsequence \(\epsilon_k\to 0\) such that, for almost every \(t\in(0,T)\), \[ \frac{e_{\epsilon_k}(\mathbf m_{\epsilon_k }(t))}{\log\frac{1}{\epsilon_k}},\quad \omega(\mathbf m_{\epsilon_k}(t)), \quad \text{and }\quad J(\mathbf m_{\epsilon_k }(t)) \] converge to weighed sums of Dirac measures located at the vortex centers \(a_n(t)\), \(n=1, \dots,N\). The motion of the vortices is described by the system of ordinary differential equations \[ 4\pi q_n i\dot{a}_n+\pi\alpha_o\dot{a}_n+\frac{\partial W}{\partial a_n}(a,d)=0, \quad\quad n=1,\dots,N, \] in \((0,T)\) with the initial condition \(a(0)=a^{o},\) where \(W\) is the renormalized energy [\textit{F. Bethuel} et al., Ginzburg-Landau vortices. Boston, MA: Birkhäuser (1994; Zbl 0802.35142)]. The case of the Ginzburg-Landau equations with mixed dynamics is also investigated in Sections 1.2 and 7.

Recommendations

Cites work

Cited in

(4)

This page was built for publication: Vortex dynamics in the presence of excess energy for the Landau-Lifshitz-Gilbert equation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2445453)