• [PA2013] Filary


    (m = 2),可以得到答案的一个下界是 (lceil {n over 2} ceil)

    直接做很难处理,考虑随机一个值必选。根据答案下界,随机一次能得到最优解的概率大于 (1 over 2),因此随机 (20) 次左右即可。

    同余条件即 (m mid x - y)(x) 已经通过随机固定,那么枚举每个数 (y),将 (x - y) 的所有 素因子 用桶维护,取最大值。

    但可能存在合数 (m),使 (k) 同样可以取到最大,可以考虑将两个对应 (a_i) 的位置完全相同 的素因子 “合并”。

    如果暴力维护,那么单次尝试 (mathcal O(n log^2 n)),因为合法的素因子只有 (mathcal O(log n)) 个,我们可以考虑给每 (a_i) 赋随机权值,对每个素因子维护 xor-Hash 即可。

    线性筛出每个数最小素因子,复杂度 (mathcal O(V + Tcdot n log n))(T) 为随机次数。

    #include <cstdio>
    #include <cctype>
    #include <algorithm>
    #include <vector>
    #include <random>
    #include <chrono>
    using namespace std;
    using ll = long long;
    using u64 = unsigned long long;
    
    namespace io {
    #define File(s) freopen(s".in", "r", stdin), freopen(s".out", "w", stdout)
      const int SIZE = (1 << 21) + 1;
      char ibuf[SIZE], *iS, *iT, obuf[SIZE], *oS = obuf, *oT = oS + SIZE - 1;
      inline char getc () {return (iS == iT ? (iT = (iS = ibuf) + fread (ibuf, 1, SIZE, stdin), (iS == iT ? EOF : *iS ++)) : *iS ++);}
      inline void flush () {fwrite (obuf, 1, oS - obuf, stdout); oS = obuf;}
      inline void putc (char x) {*oS ++ = x; if (oS == oT) flush ();}
      template<class T>
      inline void read(T &x) {
        char ch; int f = 1;
        x = 0;
        while(isspace(ch = getc()));
        if(ch == '-') ch = getc(), f = -1;
        do x = x * 10 + ch - '0'; while(isdigit(ch = getc()));
        x *= f;
      }
      template<class T, class ...Args>
      inline void read(T &x, Args&... args) {read(x); read(args...);}
      template<class T>
      inline void write(T x) {
        static char stk[128];
        int top = 0;
        if(x == 0) {putc('0'); return;}
        if(x < 0) putc('-'), x = -x;
        while(x) stk[top++] = x % 10, x /= 10;
        while(top) putc(stk[--top] + '0');
      }
      template<class T, class ...Args>
      inline void write(T x, Args... args) {write(x); putc(' '); write(args...);}
      inline void space() {putc(' ');}
      inline void endl() {putc('
    ');}
      struct _flush {~_flush() {flush();}} __flush;
    };
    using io::read; using io::write; using io::flush; using io::space; using io::endl; using io::getc; using io::putc;
    
    mt19937_64 rnd(chrono::steady_clock::now().time_since_epoch().count());
    
    const int V = 10000004, P = 664600, N = 1000005;
    bool np[V];
    int p[P], pc = 0, lpf[V];
    
    void prime_sieve(int n) {
      for (int i = 2; i <= n; ++i) {
        if (!np[i])
          p[++pc] = i, lpf[i] = pc;
        for (int j = 1; j <= pc && i * p[j] <= n; ++j) {
          np[i * p[j]] = true;
          lpf[i * p[j]] = j;
          if (i % p[j] == 0) break;
        }
      }
    }
    
    int k = 0, m = -1;
    int cnt[N], val[N], minexp[N], cv = 0;
    u64 id[N], Hash[P];
    int occ[P];
    void work(int select) {
      fill(Hash + 1, Hash + 1 + pc, 0);
      fill(occ + 1, occ + 1 + pc, 0);
      fill(minexp, minexp + 1 + pc, numeric_limits<int>::max());
      int nowc = 0;
      for (int i = 1; i <= cv; ++i) {
        if (val[i] == select) {
          nowc = cnt[i];
          continue;
        }
        int x = abs(val[i] - select);
        int last = 0, ex = 0;
        while (x != 1) {
          if (lpf[x] != last) {
            occ[lpf[x]] += cnt[i];
            if (last) minexp[last] = min(minexp[last], ex);
            ex = 1;
            Hash[lpf[x]] ^= id[i];
            last = lpf[x];
          }
          else ++ex;
          x /= p[lpf[x]];
        }
        if (last) minexp[last] = min(minexp[last], ex);
      }
      int K = *max_element(occ + 1, occ + 1 + pc), M = 0, fc = 0;
      static int fac[128], prod[128];
      if (K + nowc < k) return;
    
      for (int i = 1; i <= pc; ++i)
        if (occ[i] == K) fac[++fc] = i;
      for (int i = 1; i <= fc; ++i) {
        prod[i] = 1;
        for (int t = minexp[fac[i]]; t; --t) prod[i] *= p[fac[i]];
        for (int j = 1; j < i; ++j)
          if (Hash[fac[i]] == Hash[fac[j]]) {
            prod[j] *= prod[i];
            break;
          }
        }
      K += nowc;
      M = *max_element(prod + 1, prod + 1 + fc);
      if (K > k) m = M, k = K;
      else if (K == k && M > m) m = M;
    }
    
    int a[N];
    
    int main() {
      prime_sieve(10000000);
      int n, cmod2[2] = {0, 0};
      read(n);
      m = 2;
      for (int i = 1; i <= n; ++i)
        read(a[i]), ++cmod2[a[i] & 1];
      k = max(cmod2[0], cmod2[1]);
    
      sort(a + 1, a + 1 + n);
      for (int i = 1; i <= n; ++i) {
        if (a[i] != a[i - 1]) {
          ++cv;
          val[cv] = a[i]; cnt[cv] = 1;
          id[cv] = rnd();
        }
        else ++cnt[cv];
      }
      uniform_int_distribution<int> D(1, n);
      for (int T = 1; T <= 20; ++T) {
        int p = D(rnd);
        work(a[p]);
      }
      write(k, m), endl();
      return 0;
    }
    
  • 相关阅读:
    AIoT 2020 年分析
    TensorFlow解析常量、变量和占位符
    TensorFlow编程结构
    对端边缘云网络计算模式:透明计算、移动边缘计算、雾计算和Cloudlet
    Harmony生命周期
    立体显示与BCN双稳态手性向列相
    显示技术示例
    SystemML大规模机器学习,优化算子融合方案的研究
    改天有空看看 1
    gradle 123123
  • 原文地址:https://www.cnblogs.com/RiverHamster/p/sol-lg6817.html
Copyright © 2020-2023  润新知