CC

[CC @CaltechCountry]

@CaltechEconThry @BarbaraFantechi https://t.co/g8JV0JJHJM

[CC @CaltechCountry]

@NGL97230373 @KhomutovAn49011 @docmilanfar The median and mean of the true distribution may differ so it will not converge to zero with more samples. E.g., if the distribution is uniform over {0,1,5}, with high samples the median will be 1 and the mean will be 2

[CC @CaltechCountry]

@KhomutovAn49011 @NGL97230373 @docmilanfar Right, the picture is about samples, but there are several proofs of this statement as applied to distributions:
https://t.co/vGxQFvYB7C

[CC @CaltechCountry]

@NGL97230373 @docmilanfar It’s a fact about distributions (rather than samples), so it applies to sample distributions of any size as well

[CC @CaltechCountry]

@deedydas 2b. If at end, take staircase route above the diagonal from the found monster to the last row until win or hit monster 2.
3. Go vertical down next to monster 2, then horizontal to behind monster 1 and down to end.
n=3.

Easy to see that n=2 can’t be guaranteed.

[CC @CaltechCountry]

@deedydas #5 is not too hard.

1. Scour second row for a monster
2a. If not at end, get behind it via an adjacent column, n = 3 (or 2).