1: \begin{abstract}

2: We introduce subgradient-based Lavrentiev regularisation of the form

3: \begin{equation*}

4: \mathcal{A}(u) + \alpha \partial \mathcal{R}(u) \ni f^\delta

5: \end{equation*}

6: for linear and nonlinear ill-posed problems with monotone operators~\({ \mathcal{A} }\) and general regularisation functionals~\({ \mathcal{R} }\). In contrast to Tikhonov regularisation, this approach perturbs the equation itself and avoids the use of the adjoint of the derivative of~\({ \mathcal{A} }\). It is therefore especially suitable for time-causal problems that only depend on information in the past, governed for instance by Volterra integral operators of the first kind. We establish a general well-posedness theory in Banach spaces, prove convergence-rate results using variational source conditions, and include some illustrating examples.

7: \end{abstract}

8: