Existence of radial mixed type solutions in Chern-Simons theories of rank 2 in \(\mathbb{R}^2\) (Q2323525)

The authors consider non-abelian Chern-Simons gauge theories corresponding to gauge groups of rank two in \(\mathbb{R}^2\), namely the \(\mathrm{SU}(3)\), the \(\mathrm{SO}(5)\), and the \(G_2\) Chern-Simons models. In the static situation, the associated Euler-Lagrange equations can be reduced to a simpler \(2\times 2\) system of first order semi-linear elliptic PDEs with critical non-linearity which are called self-dual equations. Their solutions are called (self-dual) vortices and serve as a theoretical basis for exotic excitations such as anyons. There are three types of solutions which are classified according to their asymptotic behaviors for large distances, namely topological, non-topological, and mixed-type solutions. The latter satisfy \((u_1(x),u_2(x))\to(-\infty,\log(1/2))\) or \((u_1(x),u_2(x))\to(\log(1/2),-\infty)\) as \(|x|\to\infty\). Their main results are an improvement of the existence result of radial mixed-type solutions of the \(\mathrm{SU}(3)\) model in [the authors, Calc. Var. Partial Differ. Equ. 56, No. 2, Paper No. 17, 30 p. (2017; Zbl 1370.35109)] and establishing the existence of radial mixed-type solutions of the \(\mathrm{SO}(5)\) and \(G_2\) models in a radial setting. The starting point in all of these models is the system \[ \begin{cases} \Delta u_1 & = 4\mathrm{e}^{2u_1} - 2\mathrm{e}^{u_1} -2a \mathrm{e}^{2u_2} + a\mathrm{e}^{u_2} + a(b-2)\mathrm{e}^{u_1+u_2} + 4\pi N_1\delta_0\\ \Delta u_2 & = 4\mathrm{e}^{2u_2} - 2\mathrm{e}^{u_2} -2b \mathrm{e}^{2u_1} + b\mathrm{e}^{u_1} + b(a-2)\mathrm{e}^{u_1+u_2} + 4\pi N_2\delta_0 \end{cases}.\tag{1} \] Their main result (Theorem 1) says that for any non-negative integers \(N_1\) and \(N_2\), any \(a,b>0\) with \(ab<4\), and all \(\beta>2N_1+aN_2+4\), the above system admits a radial solution \((u_1,u_2)\) obeying the mixed-type boundary condition \[ u_1(r) = -\beta \log r + \mathcal{O}(1) \quad \text{and} \quad u_2(r) = -\log 2 + o(1) \quad \text{as}\ r\to\infty \] where \(r=|x|\). \medskip Their proof mainly relies on degree theory. First, they propose auxiliary functions \(f_1,f_2\) (see (2.7)--(2.8)) which are in spirit close to the true solutions of (1). In Theorem 2 they state a priori estimates for the difference \(|u_j(r)-f_j(r,t)|\) where \(u_j\) are the solutions to a certain deformation of (1) (see (2.1)--(2.3)). Its proof mainly relies on the maximum principle and a bubbling arguments. This is the content of Section 2. In Section 3, they analyze the error \(u_j-f_j\equiv w_j\) more closely. For this purpose they construct a suitable solution space \(E=\mathcal{D}_r\times X\) (see p. 1011) where \(\mathcal{D}_r\) is the Dirichlet space and \(X\) a weighted Sobolev space. They then construct an operator \(G:E\times[0,1]\) (see p. 1014) in such a way that \(u_j=f_j+w_j\) solves the deformed system if and only if every fixed point of \(G\) solves the corresponding system for the error \(w_j\) (see (3.7)--(3.8)). The operator \(G\) can be seen as a certain weighted version of \((-\Delta+1)^{-1}\) and is in particular a compact operator in \(E\) (Lemma 7). This fact is the main motivation for their definition of \(E\). In particular, \(G\) is used to construct a homotopy which connects the initial problem to a decoupled problem. Using the bounds on the difference \(|u_j(r)-f_j(r,t)|\), they conclude that the degree of the homotopy is well defined and determine it to be \(-1\) in the solution space (Lemma 8). This concludes the prove of existence of radial solutions to (1).