Decremental dynamic algorithm to trace mutually connected clusters

The structure and dynamics of interdependent networks model catastrophic failures in complex systems that are interdependent. Percolation transitions on these networks exhibit hybrid phase transitions, which have significant practical implications for the early detection of large-scale failures. While the computer simulation of the percolation transitions and related dynamics can effectively be reduced to the computation of mutually connected clusters, such a computation is nontrivial, and several algorithms to handle the task have been proposed. Here we introduce a C + + implementation of one of the algorithms. This implementation uses intrusive data structures and thus provides a greater flexibility for applications in which efficient memory access is critical. The data structures, which we provide as a part of the library, are also useful for general percolation problems.