On the collocation method for a nonlinear boundary integral equation (Q1825615)

The following potential problem with nonlinear boundary condition is studied. For given \(g\in H^{-1/2}(\Gamma)\) find \(\Phi \in H^ 1(\Omega)\) such that \((1)\quad \Delta \Phi =0\) in \(\Omega\) and \(- \frac{\partial \Phi}{\partial n}=f(x,\Phi)-g\) on \(\Gamma =\partial \Omega\). Under some suitable conditions problem (1) is equivalent to the solution of a nonlinear integral equation \((2)\quad u-Ku+VF(u)=Vg\) where V, K are the single layer and double layer operator, respectively, and F is the Nemitskyi operator corresponding to f. The use of collocation methods is then studied for equation (2) using nodal point collocation for odd-degree splines and midpoint collocation for even-degree splines. The main theorem states that for \(g\in H^{s- 1}(\Gamma),\) \(s>1/2,\) the collocation equation has a unique solution for sufficiently small mesh size and this solution converges to the solution of (2) in the \(1/2\)-norm. Error estimates are also obtained.