Representation of the exact solution and a stability analysis on the Fredholm integral equation of the first kind in reproducing kernel space (Q858829)

In an attempt to give a numerical method for the ill-posed probelm of solving a Fredholm integral equation of the first kind in \(C[a,b]\), the authors first look at the problem in the Sobolev space \(W_2^1[a,b] = \{ u: u\) is absolutely continuous in \([a,b]\) and \(u' \in L^2[a,b]\}\). The idea is to work in this space and then transfer the results back to the original space. The theory is based on the fact that the integral operator \(Au(x) := \int_a^b K(x,s) u(s) \,ds\) is a bounded linear operator from \(W^1_2[a,b]\) to itself. The authors claim this to be true if \(\int_a^b \int_a^b k(x,s) \,dx \,ds \leq M\) for some real number \(M\). Since a function \(k\) is not mentioned anywhere else in the paper, the integrand is probably supposed to be \(K\) and not \(k\), in which case the reviewer does not understand the statement. After all, one could choose \(K(x,s) = 0\) for \(a < s < (a+b)/2\) and \(K(x,s) = 1\) else. For the function \(u \equiv 1 \in W_2^1[a,b]\) this would then produce \(Au(x) = 0\) for \(a < s < (a+b)/2\) and \(A u(x) = b-a \neq 0\) else, i.e.\ \(Au\) is not even continuous, let alone absolutely continuous.