Vtoriqt mi purvi post ;-)

Sled kato probvah da vidq kak shte izglejda i mi haresa reshih da popisha malko , vupreki che nikoi nqma da go chete ;-) No vse pak tova si e dnevnik , private property ;-)

Uf da si izkaja mukata ;-) Teq rumanci za napulno poludeli. Davat nqkakvi nenormalni zadachi (campion.edu.ro) i posle taka hubavo gi opisvat che napravo .. ;-) Ta da si kaja napravo mukata .. dneska se muchih da resha poslednata im zadacha ( leaves , training #5 ) i nishto ne uspqh da napravq ;-) Reshenieto im beshe prosto .. slojnostta na dp-to e N^2*K (hehe,tova vseki moje da go izmisli , obache time limit-va oshte na 3iq test ;-))) ) , obache kato se polzva neznam si kva funkciq tq stava N*K , za podrobnosti vijte reshenieto na batch. Otivash na reshenieto na batch ( IOI 2002 ) , tam se zabatachvash oshte poveche .. It's too complicated for me ! :-( Obache obqsnenieto na rumancite prosto schupva ot vsqkade , samo malak paste ,,, za po-qsno :

"

We start by observing if a father is no longer good for A[i][j], as i grows, the next father will be to the right.

We then write that a father k1 is better than a different father k2 (k1 <>

A[k1-1][j-1] + sum(x from k1 to i, g[x] * (x-k1)) <>

This can be rewritten as:

(A[k1-1][j-1] - A[k2-1][j-1] + cost(k1.k2-1)) / (k2-k1) + sum(x from 1 to k2-1, g[x]) < -sum(x from 1 to i, g[i])

meaning a g(k1, k2) < f(i) (g is an integer function) // hehe predefining G :PPPP

"

Oppa , nali G beshe G[x] = goleminata na listoto na X , pyk stana integer function .. abe rumanci ..

Striga tolkova gluposti za dneska ;-)