Fig. 2: Diagrammatic representation of the IE solver procedure.
From: Learning integral operators via neural integral equations
The solver is initialized with the free function y0 ≔ f. The integral operator is applied to y0, and a new guess y1 is obtained. This is repeated until convergence to a solution. The left panel shows the solution as a function of solver steps. The right panel shows the error of the solution as a function of solver steps.
