OJ Board

as it turns out Gauss elimination using bitmask also solves this problem in ~1s

yeah, it's quite annoying when it's happens over and over again....

but on the other hand it's a little bit funny, when you consider this post http://forums.topcoder.com/?module=Mess ... ID=1028297

on which elements won't move:
first we choose which k elements won't move at first m positions (m choose k)
then we want to count how many there are permutations of n' = n - k elements that all of m' = m - k first elements will move
we use in/out
+ all permutations
- all permutations that have at ...

do bfs starting from 'A', then from 'B', ....
you can easily put counting number of shortest paths in bfs function

One question to understand the problem statement, what is answer for this case


1
7 3
0 0
2 0
2 1
1 1
1 -1
4 -1
4 0


i.e. i think this case doesn't have largest gap, because 5 is supremum but you can't achieve that (tree located at intersection), and for each E>0 you can achieve 5-E (move tree ...

can anyone verify my output, thx

Could anyone verify my outputs, i keep getting wa over and over again ;-)


13
0 0
10 0
10 2
9 1
8 2
7 1
6 4
5 1
4 9
3 1
2 4
1 10
0 1
5 11
4356 1
8
-10000 -10000
10000 -10000
10000 10000
5000 10000
5000 5000
-5000 5000
-5000 10000
-10000 10000
0 6000
4356 1
14
0 0
6 0 ...

Ah, so you mean using T as the nodes, and from S we get the edges? Then we need a mincost maxflow algorithm?

I won't say anything new, but it much helps when you think about reduction to minimum cost path cover of DAG. In CLR there is solution to minimum path cover of DAG and when someone ...

What is answer for
are there any tricky testcases?

EDIT: bug found

you want to compute sth like this

T[n][k] = number of different numbers that can be written using exactly n matches, and if k==0 all this numbers has the most significant digit equal to 0. When k==1 is the other case.

I think it should be very easy to people who starts in TC. At TC there are often tasks similar to this. But when someone starts only in acm-style contests these type of tasks are rare...

answer for n=2000 ...

Could anyone give some i/o ? ;-)

EDIT:
verify?

100 100
1 1 0 1
10 2 -2 -20

100 45
-100 -1 100 2
3 100 -4 -100

100 65
24 -43 24 -44
-45 78 -46 78



1046872.19 524243.69
648388.69 65709.36
962958.92 58058.69


EDIT':

correct one, after finding very stupid bug...

1100617.88 450178.78
369184 ...

Hi, could anyone help me?

http://online-judge.uva.es/board/viewtopic.php?t=21548

thanks in advance

satej wrote:@goodwill I sent an e-mail to Miguel yesterday. I think they will fix it as soon as they can.

at new server i've still wa, it's becouse the tests aren't fixed yet or shall i look for bug in my code?