The sensitivity of a computational \(L_ 1\)-approximation technique for integral equations (Q1332370): Difference between revisions

Consider a (possibly nonlinear) integral equation \[ \varphi(x)- \lambda \int^ b_ a {\mathcal K}(x, t, \varphi(t)) dt= \psi(x). \] The approximate solution is searched in the form \(\varphi= F(x, A)\) where \(A\in \mathbb{R}^ n\) is the parameter vector determined so that \[ \int^ b_ a \Bigl| F(x, A)- \lambda \int^ b_ a {\mathcal K}(x, t, F(t, A)) dt- \psi(x)\Bigl| dx \] will be minimal. The author demonstrates by numerical examples that in some cases the method may be rather stable in so-called ``sensitive'' problems (noisy data, \(\lambda\) near to a bifurcation point, \(\lambda\) near to an eigenvalue in the linear case).