OJ Board

The bug is here:

void BigNum::operator =(BigNum &num)
{

memcpy(digit,num.digit,sizeof(digit));

}


Your operator '=' doesn't return a reference to itself - it's useless. My G++ compiler also reports compilation error.


This is the proper use of that operator:

BigNum& BigNum::operator ...

In this task you have to be careful about the precision errors. Even the 1e-5 precision may not work.. There is a way to solve this task using only integer numbers - just to implement some mathematical operations using ints.
And there is no need to use big num.

It is not about a big number but the proper solution for this problem (for example some DP...), if you find it, using the big number won't cause TLE.

JackBauer wrote: TopCoder problems seam to be a little better in discribing input.

That's right but consider that TopCoder matches have quite different rules about "getting accepted".

1s? When I was doing USACO training it was 5 seconds.
O(nm) is the model solution, but it is smartly improved so that O(nm) occurs only in the worst case.

You can try these:

http://spoj.sphere.pl
http://acm.zju.edu.cn/

The problems are not divided into categories like DP, graphs etc.
It is your task to determine whether the problem can be solved by DP or greedy, or in other way. You just need to read some of them and think about the method which ...

http://www.topcoder.com/tc?module=Stati ... d2=dynProg

I think this may be interesting for you.

My O(nm) solution passed all test cases (it worked 2.12 second at worse).
You should optimize your code to get it accepted.

Finding all prime factors of the form 4m+3 works pretty fast even for the numbers like 10^12. If you find all prime numbers not greater than sqrt(n) first, you only have to check pi(sqrt(n)) factors for given number later (where pi(x) denotes the count of prime numbers not greater than x).

For ...

My algorithm works something like that:
I sort rectangles' edges ('opening' and 'closing' edges) by x-coordinate. Then I start to sweep from left to right - for every event (edge parallel to y axis) I count the area (and also the colors) of rectangles which intersect or cover this event and next ...

Yes, sweeping worked out. My algorithm runs in O(n^2*lg n) time and it passed all tests.

Hi,

I think in this problem you have to create some kind of sweeping (at the moment I'm thinking about that . It's similar to calculating common area of rectangles, if you perform that counting colors will be easy.

yaro

OK, I found a mistake (switched int to long long) and got AC.

Hi,

My code got WA, I don't know why. Are there any tricky cases?

Thanks for any help.

I tried many diffrent kinds of multiple input readings and nothing. Now I'm wondering about the type of coordinate, but I got upset with this task and now I'm doing something else.

Thanks