Well done, would you join our Clan!

if we use rerooting in this problem, won't we have to use modular inverse at some point, and it is not neccesary that the modular inverse would exist with respect to m.

(please correct me if I am wrong)

In rerooting, what we basically do is first calculate the required data by considering some node as the root of the tree.

In rerooting type problems, it is required to calculate the same data by considering each and every node as the root, using naive approach would result in tle.

So, what we try do is using the calculated values(say by considering 1 to be the root), we try to calculate the answer for each node(if we assume that node to be the root) in O(1) time.

Consider a case in which node is the current root and it is one of it's child, so it would had contributed something in the answer of node, so now we try to make it as the new root, so first we remove the contribution of it from node, then node is now a child for it, so we add the contribution of the subtree with root being node to the answer of it and now it has become the new root of the tree.

But when we need to calculate the answer for say it1(second child of node) , we have to rollback the changes we made.

Hello,

Actually I can't understand how to use merge and apply function

Like in your second question I am using

Data merge(Data a, Data b) { return a + b; }

and apply as

Data apply(Data a, int c, int d, Cost w) { return a + w; }

But this is wrong