Development of a parallel algorithmic language (abstract only)

Untitled

“…The DEAP library allows simultaneous evaluation of subjects, however, in our case the sequential execution of the genetic algorithm proved to be less time-consuming, due to the communication overhead between different processes. Communication overhead is especially prominent in Python since the global interpreter lock (GIL) prevents parallel execution of multiple threads [65]. SAILoR required on average approximately 7 minutes and 41 seconds to infer a Boolean network with 64 nodes on 32 allocated AMD EPYC 7453 CPU cores.…”