Submit | All submissions | Best solutions | Back to list |
RATING - Coder Ratings |
Some of the more elite (and not-so-elite) coders around take part in a certain unnamed programming contest. In said contest, there are multiple types of competitions. Here, we consider the Open and High School competition types. For each type, each competitor receives a rating, an integer between 1 and 100000, inclusive. A coder's rating is based upon his or her level of performance in matches and is calculated using a complicated formula which, thankfully, you will not be asked to implement.
Although the Open and High School ratings for a coder who has participated in both competition types lately are usually close, this is not always the case. In particular, High School matches are more about speed, since many coders are able to solve all the problems, whereas Open matches require more thinking and there is a steeper curve in terms of problem difficulty.
Problem Statement
You are given N coders (1 ≤ N ≤ 300000), conveniently numbered from 1 to N. Each of these coders participates in both High School and Open matches. For each coder, you are also given an Open rating Ai and a High School rating Hi. Coder i is said to be better than coder j if and only if both of coder i's ratings are greater than or equal to coder j's corresponding ratings, with at least one being greater. For each coder i, determine how many coders coder i is better than.
Input Format
On the first line of input is a single integer N, as described above.
N lines then follow. Line i+1 contains two space-separated integers, Ai and Hi.
Output Format
Line i should contain the number of coders that coder i is better than.
Sample Input
8 1798 1832 862 700 1075 1089 1568 1557 2575 1984 1033 950 1656 1649 1014 1473
Sample Output
6 0 2 4 7 1 5 1
Added by: | Brian Bi |
Date: | 2009-04-10 |
Time limit: | 1s-1.846s |
Source limit: | 50000B |
Memory limit: | 1536MB |
Cluster: | Cube (Intel G860) |
Languages: | All except: ASM64 NODEJS OBJC PERL6 SQLITE VB.NET |
Resource: | own problem |
hide comments
|
|||||||
2009-12-13 18:54:23 SALVO
@ Critical Thinking... will learn n try :P Last edit: 2009-04-13 19:14:28 |
|||||||
2009-12-13 18:54:23 Brian Bi
I don't know what you need binary search for. Also, the test data is not by any means contrived, so ordinary BSTs should work (I set the time limit liberally for the sake of slower languages.) Of course, BITs are easier to code and faster besides. Last edit: 2009-04-13 03:08:41 |
|||||||
2009-12-13 18:54:23 ~!(*(@*!@^&
@SALVO : Follow me, it requires some knowledge about interval/segment tree or binary index tree + binary search to AC it. @Omar : Data range: 1..100000 |
|||||||
2009-12-13 18:54:23 Omar ElAzazy
what is the limit of a rating Ai & Hi ? |
|||||||
2009-12-13 18:54:23 Brian Bi
An aside: The sample data contains actual coders' ratings, but this problem is not meant to single out any of them as actually "better" than any other! |
|||||||
2009-12-13 18:54:23 cpphamza
From the problem limits it's clear that it should be N*log(N) or better. |
|||||||
2009-12-13 18:54:23 SALVO
what should be the complexity get AC ? |
|||||||
2009-12-13 18:54:23 [Trichromatic] XilinX
After solving this problem, you can try problem NICEDAY. |