1: \begin{abstract}

2:  We propose a novel stochastic algorithm that randomly samples entire rows and columns of the matrix as a way to approximate an arbitrary matrix function using the power series expansion. This contrasts with existing Monte Carlo methods, which only work with one entry at a time, resulting in a significantly better convergence rate than the original approach. To assess the applicability of our method, we compute the subgraph centrality and total communicability of several large networks. In all benchmarks analyzed so far, the performance of our method was significantly superior to the competition, being able to scale up to 64 CPU cores with remarkable efficiency.

3: 

4: 

5:  \keywords{Monte Carlo methods \and randomized algorithms \and matrix functions \and network analysis}

6: \subclass{65C05 \and 68W20 \and 65F60 \and 05C90}

7: 

8: \begin{acknowledgements}

9: This work was supported by national funds through FCT, Fundação para a Ciência e a Tecnologia, under projects URA-HPC PTDC/08838/2022 and UIDB/50021/2020 (DOI:10.54499/UIDB/50021/2020) and grant 2022.11506.BD.

10: JA was funded by  Ministerio de Universidades and specifically the requalification program of the Spanish University System 2021-2023 at the Carlos III University.

11: \end{acknowledgements}

12: 

13: \end{abstract}

14: