An inexact Spingarn's partial inverse method with applications to operator splitting and composite optimization (Q1695823)

The problem of finding a point in the graph of a maximal monotone operator on a Hilbert space such that the primal variable belongs to a closed subspace whereas the dual is required to belong to its orthogonal complement, is sometimes referred to as a partial inverse method. This framework encompasses certain convex minimization problems and inclusion problems defined by the sum of finitely many maximal monotone operators. The iteration-complexity of a proposed inexact version of the partial inverse method is analysed, aided by recent developments in hybrid proximal extragradient methods. A~certain approximate solution notion is proposed and pointwise and ergodic iteration-complexity results for the new inexact partial inverse method are obtained. Applications include a proposal to study an inexact operator splitting method for solving monotone inclusion problems, stated earlier.