Generate 2D grid with alternation probability -
i want generate 2d grid environment rewards distributed (1s rewards, 0s no rewards) based on alternation probability defined in this paper falk , konald.
the basic idea once random square seeded (top left, say) 1 or 0, probability of adjacent square staying same previous, or switching - given probability of alternation.
the paper describes generation process going left right
, top bottom
. not clear on how authors intended implement it.
algorithm:
- seed top left square
- left right: starting 1,1 -- alternate set probability
- top bottom: starting 1,1 -- alternate set probability
- obtain reward matrix
Comments
Post a Comment