OJ Board

I can't get a few lines here. Is there a way to use some buildin hash function. And how do you se the memory limits for the hash_set? What about that "string" in the init part. Why is that? And what is that

return hash< const char* >()( x.s );

Hello,

Can somebody give some example source code how to use them (and some comments... are they reliable, how can you change the size of the hash array (if you can...)) . I tried to use them some time ago but I couldn't even compile...

Thank you!

You are wrong. Think again:)

Hello,

You can use the stdio.h instead of iostream. With stdio.h you can print long double in the following way:

long double d = 100.0;

printf("%Lf",d); -> This wil print something like 100.00000000
To cut the 0's in the end you can modify the %lf in the following way
printf("%.3Lf",d); -> The ...

misof,
As far as I know good hash is the one which uses nice primes... 2^32 is not one of them as far as I know. In addition Cho mentioned that you can use two different hashings and make a pair. This way you minimise the chance of wrong answer. Is it smart to use int and unsigned int as pair of mods?

Hello,

David,
What do you mean by saying this is slow? Do you use double hashing or something more complex that makes it slow? I can imagine how it can bi done faster...

Hi,

I'll try to help:)
There are a number of algorithms for min-const-maxflow (and weighted matching). The best (fastest) (as far as I know) is the Hungarian algorithm. However it is very complex and hard to impelement. Well, not "very" complex but hard for me. You can also do min-cost-max-flow ...

Very interesting...

So if we don't get AC with one prime, we change it... in the we will hit the right prime number which would give us AC:)))

Hello,

So the solution is based on hashing the patern? What if we have the same hash for a different set of letters ot this is impossible?

Hello,

Kalinov,
Thank you very much! Now I get it:)))

Hello,

Wook,
Can you explain a bit more about your idea. Where can I find some information for the trees you mentioned?

Hello,

Cho,
You have very nice idea but I have some doubts that this will give TLE. After you find the same hash as the patern you will need to do the slow compare to see if the matching is valid. For the test case that krijger mentioned (a lot of 'A's) this will give TLE. Do you do something else ...

Thanks kalinov,

However, if someone doesn't know STL what should he/she do?
I'm just curious if there is some other O(N lg N) approach.

Very nice... what if you don't use bitmasks. Will it give TLE?

After I spend a lot of time I found that it is useless here:) So what is the approach?