Final Answer (Combining down and up)

For each node $v$, the final answer combines two pieces of information.

The down contribution is what you computed when $v$ was in a subtree during Phase $1$. This captures everything below $v$.

The up contribution is what comes from the rest of the tree when $v$ acts as the root during Phase $2$. This captures everything above $v$.

For tree diameter, the answer at node $v$ is the maximum depth you can reach going down into any subtree OR going up through the parent and then down into sibling subtrees.

answer[v] := max(dp_down[v], dp_up[v])

This gives you every node's "view" of the tree. You compute all $n$ answers in $O(n)$ total time instead of running $n$ separate DFS traversals. That is the power of rerooting: you reuse information instead of recomputing it.

Space complexity is $O(n)$ for the data structures used.