On the difference of two maximal monotone operators: Regularization and algorithmic approaches (Q942362): Difference between revisions

Let be a nonmonotonic operator on a real Hilbert space \(H\) described as the difference of two maximal monotone operators \(B\) and \(C\). The topic of the paper is to find its zeros relying on the corresponding regularized problem as well as the associated dual problem. The attention is focused in finding \(x\in H\) such that \[ B(x)-C(x)\ni 0. \] One fundamental approach is to consider a regularized version by using Yosida approximates of the operator and of its resolvent, an other tool considers the dual problem. This dual problem may be well posed, often in contrast to the original one. A splitting algorithm involving the resolvents of the operators is suggested via fixed-point formulation of the regularized problem. Essential properties of the operators are studied. The algorithmic aspect is emphasised, convergence is proved in the DC programming case. The problem under consideration has many applications such as multicommodity network, image restoration processing, discrete tomography, clustering and several nonconvex industrial problems.