According to the result in previous slide, as we know the ray must go through the big square, then the probability that ray will intersect with the small inside square is the ratio of area of small square/area of big square. So the formula of cell cost comes from we know this ray will go through this cell, then it will have different cost on left child and right child. On the one hand, the cost depends on the number of triangles in child. On the other hand, the cost also depends on the area of child part(as it has larger area, the higher probability that we will intersect with it).
Why is it optimizing the algorithm seems to increase C_trav and our cost(cell)? (1/8 < 1/1.5) Am I reading this slide wrong