OJ Board

buty of arifmetica is
function swap(var a,b:T):T;begin
a:=-(a+b);
b:=-(a+b);
a:=-(a+b);
end;

or on array a[0..1]

for i:=0 to 2 do a:=-(a[0]+a[1]);

Artikali wrote:http://www.comp.nus.edu.sg/~stevenha/pr ... ing_Change

Yours advice bad cose S very big {TLE I think}

(x-1)*(x-2)*(x-3) = x^3-6x^2+11x-6
a + b + c =6
ab+ac+bc= 11
abc=6
it got ac in 1.2 seconds during contest
Time complexity O(n^2)
Space Complexity O(n)
in contest my prog with same algo as yours writed in pascal get me
tle 2. sec
now I send it , and get acc with 0:02.830 sec (by the way i only one ...

i think better is:
displaying just the time the problem was solved and the number of attempts

BUT then runklist page will be more expensive

3.5 Compute all A[n] by making use of factor[n], O(n)

I think you compute A[n] and sumA[n] same time

>> p =s -sum(p[j]| i mod j=0, 0<j<i);
Is this part O(n sqrt(n))?
it depent from implimentation
In my implement not.
cose count action dec(p[j*i],p ) i=1,2,3..mn and j=2i,3i,..flor(m/i)j
get ...

Krzysztof Duleba wrote:Could you explain why?

first of all my engl bad.
here cose :
here answer they who have rating here and on topcoder and they
answer from point of view uva contest.
not in topcoder forum cose i think in tc forum many "cool haker" so flood

1.subj?
2. how big need download enviroment for contest?
ps pls ansver here. not send to topcoder's forum.

The bottle neck of my algorithm is the sieve,
my too (for j:=2 to m div i do dec(p[i*j],p );
afterwards, everything can be done in linear time.
yep!
Since our approaches are different, it's not surprised that there is a constant factor difference in the running times.
I think not so dif.
Cho ...

1. cho can you explain?(your sol very fast-0.064sec)
2. my sol is(0.152sec more then 2 time slow :( ) :
w =(2*i+1)^3
s =w -w[i-1];
p =s -sum(p[j]| i mod j=0, 0<j<i);
r =sum(p ,1..i);
where w - all atoms in cube 2i
s -new atoms (|x| or |y| or |z| ==i)
p -atoms in level i from wich can view center (0 ...

here 2 solution(not full ) for 11014 problem:
sol1:
{$N+}const mn=100000;{100}
var r:array[0..mn] of comp;c12,t:comp;i,j,n:longint;
begin
{secret part 1-make all r[1..10000]}
i:=0;repeat read(n);if n=0 then break;
inc(i);n:=n div 2;writeln('Crystal ',i,': ',r[n]:0:0);
until false;
end.

sol2:
{$N ...

now all ok (I get accepted)
but i think exist some bug in fpc
cose my 1 try(wrong) use var r:comp; and writeln(r:0:0)
and 2 try(accepted) use var r:int64; and writeln(r);

in home i use:
for 1try bp(turbo/borland pascal)
for 2 try dcc( delphi )
for all value they get same result
but on judge ...

my prog(wich judge think wrong) for:
2
4
6
8
10
100
1000
10000
100000
200000
0
return :
Crystal 1: 26
Crystal 2: 98
Crystal 3: 290
Crystal 4: 578
Crystal 5: 1154
Crystal 6: 854786
Crystal 7: 834185090
Crystal 8: 832136021570
Crystal 9: 831930692993858
Crystal 10: 6655340768473154

It's correct? wich ...

you 3 test is WRONG:
Rome Chittagong
...
Chittagong Bandorban
Rome Barisal
HERE Bandorban Barisal with same 1 letter It can't be!

I think in some test was error.
in string where must only 1 or more '_'
as state in problem: each grid is terminated by a separation line full of underscores `_'.
judge (if see carefull test file for 785 ) may wotch
that situeted ' '(I think on end of string full of underscores `_' )

base on that ...

Maybe you can add new service
for one who get AC(or PE) on problem XXX
phe can get solution other man's who get AC(or PE) on problem XXX
with for example 1) .bigger time
example 2) bigger time and bigger memory?

WHAT THINK about IT the judge admin?