1: \begin{abstract}

2:   In this paper we introduce a randomized version of the

3:   backward Euler method,

4:   that is applicable to stiff ordinary differential equations and

5:   nonlinear evolution equations with time-irregular

6:   coefficients. In the finite-dimensional  case,

7:   we consider Carath\'eodory type functions satisfying

8:   a one-sided Lipschitz condition. 

9:   After investigating the well-posedness and the stability properties

10:   of the randomized scheme, we prove the convergence to the exact solution

11:   with a rate of $0.5$ in the root-mean-square norm assuming only that

12:   the coefficient function is square

13:   integrable with respect to the temporal parameter.

14:   

15:   These results are then extended to the numerical solution of

16:   infinite-dimen\-sional  evolution equations under monotonicity and

17:   Lipschitz conditions. Here we consider a combination of the randomized

18:   backward Euler scheme with a Galerkin finite element method. 

19:   We obtain error estimates that correspond to the regularity of the exact 

20:   solution. 

21:   The practicability of the randomized scheme is also illustrated through

22:   several numerical experiments.

23: \end{abstract}

24: