A fast solver for integral equations with convolution-type kernel (Q2392715)

The paper is devoted to study the data redundancy of the coefficient matrix of the corresponding discrete system which forms a basis for fast algorithms of solving the integral equations whose kernel includes a convolution function factor as \[ u(x) - \int\limits_E K(x-y)L(x,y)u(y) dy = f(x),~ \, x \in E, \] where \(E \subset \mathbb R^d\) \((d \geq 1)\) is a bounded rectangular domain; \(f\), \(K\) and \(L\) are given functions defined on \(E\), \(\mathbb R^d\) and \(E \times E\), respectively, and \(u\) is the unknown to be determind. The authors develop lossless matrix compression strategies, which reduce the cost of integral evaluations and the storage to linear complexity, i.e. the same order of the approximation space dimensions. They establish that this algorithm preserves the convergence order of the approximate solution and propose a hardware-aware parallel algorithm for these strategies.