In order to score half of the points, we must first determine an array ~dist[a][b]~ which stores the shortest distances between pairs of nodes. We can do that by starting BFS from each node towards all other nodes. After that, we can do a special BFS from starting nodes that expands in the following way: if we are currently in node a with distance ~d~, then we add all nodes to the queue which are not yet visited and are distant from ~a~ by ~\le (d-1) \times K~. We can find these nodes by traversing ~dist[a]~.

To score all points it was enough to conclude that, if we know the distance from node ~x~ to some of the starting nodes, then we can deduce the solution for that node (either using a formula or a simple simulation). Distance from each node to the set of the starting nodes can be determined using BFS with all starting nodes being initially emplaced in our queue.