1: \begin{abstract}

2: In this paper we study the relaxed primal-dual 

3: algorithm for solving composite 

4: monotone inclusions in real Hilbert spaces with

5: critical preconditioners. 

6: Our approach is based in new results on the asymptotic behaviour of

7: Krasnosel'ski\u{\i}-Mann (KM) iterations defined in the range of 

8: monotone self-adjoint linear operators. These results generalize

9: the convergence of classical KM iterations aiming at approximating 

10: fixed points. 

11: We prove that the relaxed primal-dual algorithm 

12: with critical preconditioners define KM iterations in the range of a 

13: particular monotone self-adjoint linear operator with non-trivial kernel.

14: We then deduce from our fixed point approach that the shadows of 

15: primal-dual iterates on the range 

16: of the linear operator converges weakly to some point in this vector 

17: subspace from which we obtain a 

18: solution. This generalizes \cite[Theorem~3.3]{condat} to infinite 

19: dimensional relaxed primal-dual monotone inclusions involving critical 

20: preconditioners.

21: The Douglas-Rachford splitting (DRS) is interpreted as a particular 

22: instance of the primal-dual algorithm 

23: when the step-sizes are critical and we recover classical results from 

24: this new perspective. We 

25: implement the relaxed primal-dual algorithm with critical 

26: preconditioners in total variation reconstruction and we 

27: illustrate its flexibility and efficiency.

28: \par

29: \bigskip

30: 

31: \noindent \textbf{Keywords.} {\it convex 

32: optimization \and Douglas--Rachford splitting \and 

33: Krasnosel'ski\u{\i}-Mann iterations \and

34: monotone operator  theory \and primal-dual algorithm \and 

35: quasinonexpansive operators.}

36: \end{abstract}

37: