結果
| 問題 |
No.5020 Averaging
|
| コンテスト | |
| ユーザー |
 Mao-beta
|
| 提出日時 | 2024-02-25 16:37:39 |
| 言語 | PyPy3 (7.3.15) |
| 結果 |
AC
|
| 実行時間 | 873 ms / 1,000 ms |
| コード長 | 7,124 bytes |
| コンパイル時間 | 285 ms |
| コンパイル使用メモリ | 81,700 KB |
| 実行使用メモリ | 76,696 KB |
| スコア | 29,651,754 |
| 最終ジャッジ日時 | 2024-02-25 16:39:06 |
| 合計ジャッジ時間 | 41,257 ms |
|
ジャッジサーバーID （参考情報） |
judge10 / judge11 |
| 純コード判定しない問題か言語 |
（要ログイン）
| ファイルパターン | 結果 |
|---|---|
| other | AC * 50 |
ソースコード
import sys
import math
import bisect
from heapq import heapify, heappop, heappush
from collections import deque, defaultdict, Counter
from functools import lru_cache
from itertools import accumulate, combinations, permutations, product
sys.setrecursionlimit(1000000)
MOD = 10 ** 9 + 7
MOD99 = 998244353
input = lambda: sys.stdin.readline().strip()
NI = lambda: int(input())
NMI = lambda: map(int, input().split())
NLI = lambda: list(NMI())
SI = lambda: input()
SMI = lambda: input().split()
SLI = lambda: list(SMI())
EI = lambda m: [NLI() for _ in range(m)]
def score(a, b, x):
v1 = abs(5 * 10**17 - a)
v2 = abs(5 * 10**17 - b)
if max(v1, v2) > 0:
return int(2000000 - 100000 * math.log10(max(v1, v2)+1))
else:
return 2000050 - x
def answer(ans):
print(len(ans))
for i, j in ans:
print(i+1, j+1)
def greedy():
N = NI()
AB = EI(N)
ans = []
# 0と何かだけで貪欲
for qi in range(50):
a0, b0 = AB[0]
s = score(a0, b0, qi)
best_s = s
best_v = 1
for v in range(1, N):
av, bv = AB[v]
na, nb = (a0+av) // 2, (b0+bv) // 2
ns = score(na, nb, qi+1)
if ns > best_s:
best_s = ns
best_v = v
if best_s > s:
av, bv = AB[best_v]
na, nb = (a0 + av) // 2, (b0 + bv) // 2
AB[0] = [na, nb]
AB[best_v] = [na, nb]
ans.append([0, best_v])
else:
break
answer(ans)
def means(AB, i, j):
ai, bi = AB[i]
aj, bj = AB[j]
na, nb = (ai + aj) // 2, (bi + bj) // 2
return na, nb
def query(AB, i, j):
na, nb = means(AB, i, j)
AB[i] = [na, nb]
AB[j] = [na, nb]
def total_score(AB):
res = 0
for a, b in AB:
res += score(a, b, 0)
return res
def calc_gap(AB, i, j):
# ABについてi, jを操作したときのスコアの差分
na, nb = means(AB, i, j)
after = score(na, nb, 0) * 2
before = score(AB[i][0], AB[i][1], 0) + score(AB[j][0], AB[j][1], 0)
return after - before
def _main():
N = NI()
AB = EI(N)
total = total_score(AB)
# print(total)
ans = []
# totalが改善するように山登り
for qi in range(50):
max_gap = 0
best_ij = [0, 1]
for i in range(N):
if qi == 49 and i > 0:
break
for j in range(i+1, N):
gap = calc_gap(AB, i, j)
if gap > max_gap:
max_gap = gap
best_ij = [i, j]
if max_gap > 0:
total += max_gap
query(AB, *best_ij)
ans.append(best_ij)
# print(total, score(AB[0][0], AB[0][1], qi))
else:
break
answer(ans)
def manhattan(a, b, ta, tb):
v1 = abs(ta - a)
v2 = abs(tb - b)
return v1 + v2
def chebyshev(a, b, ta, tb):
v1 = abs(ta - a)
v2 = abs(tb - b)
return max(v1, v2)
def twobytwo():
N = NI()
AB = EI(N)
ans = []
CENTER = 5 * 10**17
# 二次元平面上で考える
# 操作は「2点選んで中点につぶす」
# AB[0]の反対側になるべく近いところになるように2点全探索
# →そことAB[0]を選ぶ を繰り返す？
prev_ij = [0, 1]
for qi in range(50):
if qi % 2 == 0:
a0, b0 = AB[0]
ta = CENTER * 2 - a0
tb = CENTER * 2 - b0
minm = CENTER * 2
best_ij = [0, 1]
for i in range(1, N):
for j in range(i + 1, N):
na, nb = means(AB, i, j)
m = manhattan(na, nb, ta, tb)
if m < minm:
best_ij = [i, j]
minm = m
if minm < CENTER * 2:
query(AB, *best_ij)
ans.append(best_ij)
prev_ij = best_ij
else:
j = prev_ij[1]
query(AB, 0, j)
ans.append([0, j])
# print(score(*AB[0], 0))
answer(ans)
def main():
N = NI()
AB = EI(N)
ans = []
CENTER = 5 * 10**17
# 二次元平面上で考える
# 操作は「2点選んで中点につぶす」
# 4点→2点→1点にした点がAB[0]と反対側ならOK
# AB[0]が端の場合無理なのでまず1回やる
minm = CENTER * 2
nj = 1
for j in range(1, N):
m = manhattan(*AB[0], *AB[j])
if m < minm:
nj = j
minm = m
query(AB, 0, nj)
ans.append([0, nj])
for _ in range(3):
# 0以外の4点を全探索
a0, b0 = AB[0]
ta, tb = CENTER*2 - a0, CENTER*2 - b0
minm = CENTER * 2
bestP = []
for P in permutations(range(1, N), 4):
p1, p2, p3, p4 = P
na1, nb1 = means(AB, p1, p2)
na2, nb2 = means(AB, p3, p4)
na, nb = (na1+na2) // 2, (nb1+nb2) // 2
m = manhattan(ta, tb, na, nb)
if m < minm:
minm = m
bestP = list(P)
for P in permutations(range(1, N), 3):
p1, p2, p3 = P
na1, nb1 = means(AB, p1, p2)
na2, nb2 = AB[p3]
na, nb = (na1+na2) // 2, (nb1+nb2) // 2
m = manhattan(ta, tb, na, nb)
if m < minm:
minm = m
bestP = list(P)
# print(score(*AB[0], 0))
if len(bestP) == 3:
p1, p2, p3 = bestP
query(AB, p1, p2)
ans.append([p1, p2])
query(AB, p2, p3)
ans.append([p2, p3])
query(AB, p3, 0)
ans.append([p3, 0])
elif len(bestP) == 4:
p1, p2, p3, p4 = bestP
query(AB, p1, p2)
ans.append([p1, p2])
query(AB, p3, p4)
ans.append([p3, p4])
query(AB, p2, p4)
ans.append([p2, p4])
query(AB, 0, p4)
ans.append([0, p4])
answer(ans)
# print(score(*AB[0], 1))
# print(AB[0])
def sort_ABs():
# いまいち
N = NI()
AB = EI(N)
ans = []
ABI = [[a, b, i] for i, (a, b) in enumerate(AB)]
ABI.sort()
for si in range(N//2):
i, j = ABI[si][2], ABI[N-1-si][2]
query(AB, i, j)
ans.append([i, j])
ABI = [[a, b, i] for i, (a, b) in enumerate(AB)]
AB.sort(key=lambda x: x[1])
for si in range(N // 2):
i, j = ABI[si][2], ABI[N - 1 - si][2]
query(AB, i, j)
ans.append([i, j])
CENTER = 5 * 10**17
a0, b0 = AB[0]
ta = CENTER * 2 - a0
tb = CENTER * 2 - b0
minm = CENTER * 2
best_ij = [0, 1]
for i in range(1, N):
for j in range(i + 1, N):
na, nb = means(AB, i, j)
m = manhattan(na, nb, ta, tb)
if m < minm:
best_ij = [i, j]
minm = m
query(AB, *best_ij)
ans.append(best_ij)
j = best_ij[1]
query(AB, 0, j)
ans.append([0, j])
answer(ans)
if __name__ == "__main__":
main()