Double-slit Markov Darwinism

More about Double-slit Markov Darwinism

See also previous Markov FSM versus Neural Net artwork. We have two cases when the node emits two threads : Firstly, both threads enter the same next state as each other. In this case no other node is involved, and we have merely continued the issue one step further on, and create no new threads. The other case is where the two threads exit through different branches, to different new states. In this case we will have a superposition of two next states, and we need to make these interfere (like the "double slit" quantum experiment) by merging the outputs (the thread "particles" become a "wave" superposition of probabilities (Markov weightings).) Both threads persist independently to encounter the next round, where one of the two cases may happen again. So in summary, if the two threads go through the same "slot" they continue as the two threads on to the next stage; if the two threads go through different "slots" we track them both. If one of those threads is about to split into two as it leaves the next node, we apply a Darwin algorithm to terminate two of the four potential threads, in order to keep only two. This may be totally random, or we keep only the two with the highest weightings.