Some not fluent thoughts

Go with the Flow Challenge
1

I will write on one of the rules weakness by my understanding.
By my understanding we have two reasons for agent became dead:

  1. reach the top
  2. nowhere to move (it mean the other molecules is surrounded by agent )

imagine we are lucky and we have the accumulation. Inside accumulation molecules cannot move.
But… the upper layer of the accumulation is not surrounded by molecules, so it can “evaporate upwards”, after time, layer by layer accumulation can “evaporate”.
Ok… I think we have rule- if agent is dead- it moves no longer
But imagine that our accumulation is in the initial stage and its molecules is “almost dead”- their have still somewhere to move. So above said is still valid.
I think extreme values of environment signal should stop the movement (like new rule).
so there was some thoughts in my not fluent English :slight_smile:

2

That is a good idea, did you build this into your submission? We like the idea of contestants coming up with their own rules (as long as there’s a docstring). Could you give us an example of an extreme value that would cause a different agent reaction?

3

I tried it but didn implement in final final solution. Now I think that wis idea is not good for because the oil behave not localy and not causaly. I need to find local rules. I tried one of the eule rhe bigger amaunt of molecules in the box (3x3x3) the slower speed of molecules. I think it is physical, but didnt implement in the final solution…

4

2 days ago Google AI team published paper on marriage deep learning and cellular automata. Cellular automata is that we did in this competition. https://arxiv.org/pdf/2008.04965v1.pdf