if you make an account, you are automatically registered.

This has been resolved.

EDIT: I'm passing along a request from the contest admin, who would also like to request that for any sorts of technical issues, to contact directly as opposed to posting about them on social media. The admin isn't actively tracking all forms of social media where these issues might be discussed but will respond to any direct queries, and hopes that contestants who notice issues will escalate them properly so everyone can have a smooth contest.

You need to wait until the entire contest is finished. However, you can still view the problem statements.

DP On Trees, DP[i][j] is how many diferent coloring can we generate in the subtree of i if we give i the color j , dp[i][j] *= (dp[x][(j+1)%3] + dp[x][(j+2)%3]) , x is a son of i, do this for every x son of i, take care of the nodes who already have a color. Code

Here's my Java solution for reference as well.

Simply binary search on the first cow which does not get a gift.

Find a prefix such that each cow in that prefix will enter the queue inside the prefix. If this happens all remaining cows will not reach the front of the queue. You need to find such smallest prefix. I solved it in O(n log n) because I use max-segment tree but it can be solved with two sweeps in O(n) I believe.

Suffix Array + LCP Array

I have a solution with suffix automaton.

DP[X][Y][4] — is cell (X,Y) reachable from B if the player was last is one of the 4 adjacent cells. Then the transition can be done by seeing if we can push the box to the four directions (cells of type (X+dirx[p],Y+diry[p])). That's possible if from the current player's cell we can reach cell (X-dirx[p],Y-diry[p]) without stepping on the cell with the box (cell (X, Y)). Well this is equivalent to checking if these two cells are in the same biconnected component (I'm not sure if that's the correct name, but I mean components such that by removing each vertex the component remains connected). During the contest I wasn't able to pass the problem as I implement Tarjan's algorithm in a wrong way, but I'm almost certain this solution should work.

Could somebody please explain why bridge tree / finding 2-edge connected components isn't sufficient for solving this problem?

"The USACO December 2017 contest has just ended. Results are being tabulated and will be posted soon." — On the front page as of now.

I assume that means within a day.