Блог пользователя broly_1033

Path Queries 2 on CSES
Автор broly_1033, история, 2 года назад, По-английски

Hi everyone, I was studying how to solve query problems on trees using Euler Tour. I was wondering how to solve path query problems on trees. We can solve problems which can be solved by maintaining a prefix array (for e.g. sum of nodes in path from a to b), but how to solve say, node with maximum value in path from a to b. Can anyone guide me on this?
More specifically, sum(a, b) = sum(root, a) + sum(root, b) — 2*sum(root, lca(a, b)) + val(lca). I was using this to solve these problems using Euler Tour. But I cannot find maximum using this (CSES Path Queries 2).
Q1. Can we apply segment trees on Euler Tour array? Q2. How to solve problems like CSES Path Queries 2?

  • Проголосовать: нравится
  • 0
  • Проголосовать: не нравится

»
2 года назад, # |
  Проголосовать: нравится 0 Проголосовать: не нравится

Yah you need to use heavy light decomposition, which basically uses segment tree.

I hope this helps

https://codeforces.com/blog/entry/81317

→ Ответить
  • »
    »
    2 года назад, # ^ |
      Проголосовать: нравится 0 Проголосовать: не нравится

    I have heard of HLD but couldn't seem to grasp it. Thanks, is there any other method though?

    → Ответить
    • »
      »
      »
      2 года назад, # ^ |
        Проголосовать: нравится 0 Проголосовать: не нравится

      It is a fairly high level concept, if I was you I would focus on easier topics first.

      No I don't think there's any other way of solving this problem.

      → Ответить
»
12 месяцев назад, # |
  Проголосовать: нравится 0 Проголосовать: не нравится

My O(Nlog^2N) code is TLEing in the last test, any idea why ?

→ Ответить
  • »
    »
    10 месяцев назад, # ^ |
    ← Rev. 2   Проголосовать: нравится 0 Проголосовать: не нравится

    Hey! I saw your code, eventhough it is asymtotically optimal, there are few optimizations you could do. 1. You are accumulating ranges in a vector and querying the segment tree later. Instead directly query the segment tree. Avoid the intermediate list overhead. 2. This is not needed:
    if (par != -1) { node.leg.erase(find(node.leg.begin(), node.leg.end(), par)); node.depth = 1 + T[par].depth; } 3. Try doing minor optimizations here and there.

    → Ответить