1: \begin{abstract}

2: We introduce a variant of Multicut Decomposition Algorithms (MuDA), called CuSMuDA (Cut Selection for Multicut Decomposition Algorithms), for solving 

3: multistage stochastic linear programs that incorporates a class of  cut selection

4: strategies to choose the most relevant cuts of the approximate recourse functions.

5: This class contains Level 1 \cite{dpcuts0} and Limited Memory Level 1 \cite{guiguesejor2017} cut selection strategies, initially

6: introduced for respectively Stochastic Dual Dynamic Programming (SDDP) and Dual Dynamic Programming (DDP).

7: We prove the almost sure convergence of the method in a finite number

8: of iterations and obtain as a by-product the almost sure convergence in a finite number of iterations of SDDP combined with our class of cut selection strategies.

9: 

10: We compare the performance of MuDA, SDDP, and their variants with cut selection (using Level 1 and Limited Memory Level 1) on

11: several instances of a portfolio problem and of an inventory problem.

12: On these experiments, in general, SDDP is quicker (i.e., satisfies the stopping criterion quicker) than 

13: MuDA and cut selection allows us to decrease the computational bulk with Limited Memory Level 1 being more efficient (sometimes

14: much more) than Level 1.\\

15: \end{abstract}

16: