2016 ACM/ICPC亚洲区沈阳站
A | B | C | D | E | F | G | H | I | J | K | L | M |
O | O | O | $\varnothing$ | $\varnothing$ | $\varnothing$ | $\varnothing$ | $\varnothing$ | $\varnothing$ | $\varnothing$ |
签到。
签到
$$f[i] = f[i - 1] + 2 * f[i - 2] + i ^ 4$$
$$
\left[
\begin{matrix}
1 & 2 & 1 & 0 & 0 & 0 & 0 \\
1 & 0 & 0 & 0 & 0 & 0 & 0\\
0 &0 & 1 & 4 & 6 &4 & 1 \\
0 & 0 & 0 & 1 & 3 & 3 & 1 \\
0 & 0 & 0 & 0 & 1 & 2 & 1 \\
0 & 0 & 0 & 0 & 0 & 1 & 1 \\
0 & 0 & 0 & 0 & 0 & 0 & 1
\end{matrix}
\right]
\times
\left[
\begin{matrix}
f_{i-1} & 0 & 0 & 0 & 0 & 0 & 0 \\
f_{i-2} & 0 & 0 & 0 & 0 & 0 & 0\\
i^4 &0 & 0 & 0& 0 &0 & 0 \\
i^3 & 0 & 0 & 0 & 0 & 0 & 0 \\
i^2 & 0 & 0 & 0 & 0 & 0 & 0\\
i & 0 & 0 & 0 & 0 & 0 & 0 \\
1 & 0 & 0 & 0 & 0 & 0 & 0
\end{matrix}
\right]=
\left[
\begin{matrix}
f_{i} & 0 & 0 & 0 & 0 & 0 & 0 \\
f_{i} & 0 & 0 & 0 & 0 & 0 & 0\\
(i+1)^4 &0 & 0 & 0& 0 &0 & 0 \\
(i+1)^3 & 0 & 0 & 0 & 0 & 0 & 0 \\
(i+1)^2 & 0 & 0 & 0 & 0 & 0 & 0\\
i+1 & 0 & 0 & 0 & 0 & 0 & 0 \\
1 & 0 & 0 & 0 & 0 & 0 & 0
\end{matrix}
\right]
$$
- #include <bits/stdc++.h>
- #define ll long long
- const ll MOD = ;
- const int N = ;
- struct Mat {
- ll a[][];
- Mat() {
- memset(a, , sizeof(a));
- }
- Mat(int x) {
- memset(a, , sizeof(a));
- for (int i = ; i <= N; i++)
- a[i][i] = ;
- }
- Mat operator * (const Mat &rhs) const {
- Mat c;
- for (int i = ; i <= N; i++)
- for (int j = ; j <= N; j++)
- for (int k = ; k <= N; k++)
- c.a[i][j] = (c.a[i][j] + a[i][k] * rhs.a[k][j] % MOD) % MOD;
- return c;
- }
- };
- Mat qp(Mat a, int b) {
- Mat c();
- while (b) {
- if (b & ) c = a * c;
- a = a * a;
- b >>= ;
- }
- return c;
- }
- int main() {
- int T;
- scanf("%d", &T);
- Mat base;
- base.a[][] = ; base.a[][] = ; base.a[][] = ;
- base.a[][] = ;
- base.a[][] = ; base.a[][] = ; base.a[][] = ; base.a[][] = ; base.a[][] = ;
- base.a[][] = ; base.a[][] = ; base.a[][] = ; base.a[][] = ;
- base.a[][] = ; base.a[][] = ; base.a[][] = ;
- base.a[][] = ; base.a[][] = ;
- base.a[][] = ;
- while (T--) {
- int n, a, b;
- scanf("%d%d%d", &n, &a, &b);
- if (n == ) {
- printf("%d\n", a);
- continue;
- }
- if (n == ) {
- printf("%d\n", b);
- continue;
- }
- n--;
- n--;
- Mat ans;
- ans.a[][] = b; ans.a[][] = a; ans.a[][] = ; ans.a[][] = ; ans.a[][] = ; ans.a[][] = ; ans.a[][] = ;
- ans = qp(base, n) * ans;
- printf("%lld\n", ans.a[][] % MOD);
- }
- return ;
- }
博弈。
$dp[n][a][b]$ 表示当前剩下 $n$ 个物品,第一个人有 $a$ 元钱,第二个人有 $b$ 元钱,第一个人能获得的物品数。
$dp[1][a][b] = [a\leq b]$
消除奇偶性可以通过从 $dp[i - 1][b][a]$ 转移过来。这样就不用考虑奇偶了。
然后枚举两个人分别要出多少钱,当第一个人出 $x$ 块钱,第二个人出 $x+1$ 块钱会使第一个人的收益变小,那么第二个人会继续加价。同理第一个人会继续加价。当无法得到更好的收益的时候就停下来。
- #include <cstdio>
- #include <algorithm>
- #include <cstring>
- const int N = ;
- short dp[N][N][N];
- int n, T, a, b;
- inline int geta(int n, int a) {
- return n / * a + (n - n / ) * (a + );
- }
- int main() {
- for (int a = ; a < N; a++)
- for (int b = ; b < N; b++)
- if (a >= b) dp[][a][b] = ;
- for (int i = ; i < N; i++) {
- for (int a = ; a < N; a++) {
- int limit = std::min(N, geta(i, a));
- for (int b = ; b < limit; b++) {
- if (a == b) {
- dp[i][a][b] = (i + ) / ;
- continue;
- }
- int vala = i - dp[i - ][b][a];
- int valb = ;
- for (int u = ; ; u++) {
- if (b < u || (valb = (i - - dp[i - ][b - u][a])) >= vala) {
- dp[i][a][b] = vala;
- break;
- }
- if (a < u || (vala = (i - dp[i - ][b][a - u])) <= valb) {
- dp[i][a][b] = valb;
- break;
- }
- }
- }
- }
- }
- scanf("%d", &T);
- while (T--) {
- scanf("%d%d%d", &n, &a, &b);
- int ans1 = dp[n][a][b], ans2 = n - ans1;
- printf("Alice %d Bob %d\n", ans1, ans2);
- }
- return ;
- }
经过暑假牛客多校的洗礼,看到团就想到暴搜...
首先用了bfs+bitset。T了。
改成vector判,又T了。
看了一份题解是dfs,存团的是数组。
我把那份代码的数组改成vector,还是T。
这么卡STL的吗...
- #include <bits/stdc++.h>
- const int N = ;
- std::vector<int> G[N];
- bool mp[N][N];
- int ans, n, m, s;
- void solve(int *a, int u) {
- if (a[] == s) {
- ans++;
- return;
- }
- for (int v: G[u]) {
- if (v <= u) continue;
- bool flag = ;
- for (int j = ; j <= a[]; j++) {
- if (!mp[a[j]][v]) {
- flag = ;
- break;
- }
- }
- if (!flag) continue;
- a[++a[]] = v;
- solve(a, v);
- --a[];
- }
- }
- int a[N];
- int main() {
- int T;
- scanf("%d", &T);
- while (T--) {
- scanf("%d%d%d", &n, &m, &s);
- for (int i = ; i <= n; i++) {
- G[i].clear();
- for (int j = ; j <= n; j++)
- mp[i][j] = ;
- a[i] = ;
- }
- for (int i = ; i < m; i++) {
- int u, v;
- scanf("%d%d", &u, &v);
- if (u > v) std::swap(u, v);
- G[u].push_back(v);
- mp[u][v] = mp[v][u] = ;
- }
- ans = ;
- for (int i = ; i <= n; i++) {
- a[] = ;
- a[] = i;
- solve(a, i);
- }
- printf("%d\n", ans);
- }
- return ;
- }
积分题。
看的这篇 https://www.cnblogs.com/chen9510/p/7635679.html
- #include <bits/stdc++.h>
- const double pi = acos(-1.0);
- const double eps = 1e-;
- inline int dcmp(double x) {
- if (fabs(x) < eps) return ;
- return x < ? - : ;
- }
- double cal(double x) {
- return sin(x) - x * cos(x) - / 3.0 * sin(x) * sin(x) * sin(x);
- }
- double calV(double mid) {
- double V = cal(acos(1.0)) - cal(acos(1.0 - mid));
- V *= -2.0 / mid;
- return V;
- }
- double cal2(double x) {
- return x + sin( * x) / ;
- }
- double area(double a, double x) {
- return (cal2(pi / 2.0) - cal2(asin(x / a))) * a;
- }
- int main() {
- int T;
- scanf("%d", &T);
- while (T--) {
- double d;
- scanf("%lf", &d);
- if (dcmp(d - ) >= ) {
- double h = - * d;
- double a = sqrt( * + h * h) / ;
- printf("%.5f\n", pi * a);
- continue;
- }
- if (dcmp(d) == ) {
- puts("0.00000");
- continue;
- }
- double l = , r = 2.0;
- for (int i = ; i < ; i++) {
- double mid = (l + r) / 2.0;
- double V = calV(mid);
- if (dcmp(V - pi * d) == ) break;
- if (dcmp(V - pi * d) < ) l = mid;
- else r = mid;
- }
- double mid = l;
- int flag = ;
- if (mid < ) flag = -;
- double len = sqrt(mid * mid + );
- double h = sqrt( - ( - mid) * ( - mid));
- double a = len / ( + flag * sqrt( - h * h));
- double x = a - len;
- printf("%.5f\n", area(a, x));
- }
- }
先把AC自动机建出来。在AC自动机上DP。
$dp[i] = \sum \frac{1}{6} \times dp[from]$
$from$ 为能转到 $i$ 且不为某个串的结尾的结点
因为会存在环,所以高斯消元就行了。
- #include <bits/stdc++.h>
- const int N = + ;
- double mat[N][N];
- const double eps = 1e-;
- inline void gauss(int n) {
- for(int i = ; i <= n; i++) {
- int r = i;
- for(int j = i + ; j <= n; j++)
- if(std::fabs(mat[r][i]) < std::fabs(mat[j][i]))
- r = j;
- if(r != i) std::swap(mat[i], mat[r]);
- for(int j = ; j <= n; j++) {
- if(j == i) continue;
- double t = mat[j][i] / mat[i][i];
- for(int k = i; k <= n + ; k++)
- mat[j][k] -= mat[i][k] * t;
- }
- }
- for(int i = ; i <= n; i++) {
- mat[i][n + ] /= mat[i][i];
- }
- }
- struct Aho {
- static const int sz = ;
- int ch[N][sz], last[N], fail[N], tol;
- bool end[N];
- void init() {
- tol = ;
- newnode();
- }
- inline int newnode() {
- memset(ch[tol], , sizeof(ch[tol]));
- last[tol] = fail[tol] = end[tol] = ;
- return tol++;
- }
- void insert(int *a, int n) {
- int u = ;
- for (int i = ; i < n; i++) {
- int id = a[i] - ;
- if (!ch[u][id]) ch[u][id] = newnode();
- u = ch[u][id];
- }
- end[u] = ;
- }
- void build() {
- std::queue<int> que;
- for (int i = ; i < sz; i++)
- if (ch[][i]) que.push(ch[][i]), fail[ch[][i]] = last[ch[][i]] = ;
- while (!que.empty()) {
- int u = que.front(); que.pop();
- end[u] |= end[last[u]];
- for (int i = ; i < sz; i++) {
- int &v = ch[u][i];
- if (v) {
- fail[v] = ch[fail[u]][i];
- que.push(v);
- last[v] = end[fail[v]] ? fail[v] : last[fail[v]];
- } else {
- v = ch[fail[u]][i];
- }
- }
- }
- }
- void solve() {
- memset(mat, , sizeof(mat));
- mat[][tol] = -1.0;
- for (int i = ; i < tol; i++) {
- mat[i][i] = -1.0;
- if (end[i]) continue;
- for (int j = ; j < sz; j++)
- mat[ch[i][j]][i] += 1.0 / ;
- }
- gauss(tol - );
- bool flag = ;
- for (int i = ; i < tol; i++)
- if (end[i]) {
- if (flag) putchar(' ');
- printf("%.6f", mat[i][tol]);
- flag = ;
- }
- puts("");
- }
- } ac;
- int a[];
- int main() {
- int T;
- scanf("%d", &T);
- for (; T--; ) {
- int n, l;
- scanf("%d%d", &n, &l);
- ac.init();
- for (int i = ; i <= n; i++) {
- for (int j = ; j < l; j++)
- scanf("%d", a + j);
- ac.insert(a, l);
- }
- ac.build();
- ac.solve();
- }
- return ;
- }
$dp[u] = min(dp[anc] + (sum[u] - sum[anc])^2 + p)$
$anc$ 为 $u$ 到根的路径上的结点。
斜率优化DP一下。在进入一个结点时存储一下对当前队列的修改,离开一个结点时改回去,这样就能保证进入一个结点时,队列存储的都是它的祖先。
- #include <bits/stdc++.h>
- #define pii pair<int, int>
- #define ll long long
- #define fi first
- #define se second
- const int N = 1e5 + ;
- const double eps = 1e-;
- ll dp[N];
- int n, p, que[N], l, r;
- ll sum[N];
- std::vector<std::pii> G[N];
- inline ll sqr(ll x) {
- return x * x;
- }
- inline ll up(int i, int j) {
- return dp[i] + sqr(sum[i]) - dp[j] - sqr(sum[j]);
- }
- inline ll down(int i, int j) {
- return sum[i] - sum[j];
- }
- ll ans;
- void dfs(int u, int fa = ) {
- std::vector<std::pii> vec;
- int x = l, y = r;
- while (l < r && up(que[l + ], que[l]) <= down(que[l + ], que[l]) * * sum[u]) {
- vec.push_back(std::pii(l, que[l]));
- l++;
- }
- if (u != ) {
- dp[u] = dp[que[l]] + sqr(sum[u] - sum[que[l]]) + p;
- ans = std::max(ans, dp[u]);
- }
- while (l < r && up(que[r], que[r - ]) * down(u, que[r]) >= up(u, que[r]) * down(que[r], que[r - ])) {
- vec.push_back(std::pii(r, que[r]));
- r--;
- }
- que[++r] = u;
- for (auto v: G[u]) {
- if (v.fi == fa) continue;
- sum[v.fi] = sum[u] + v.se;
- dfs(v.fi, u);
- }
- l = x, r = y;
- for (auto p: vec)
- que[p.fi] = p.se;
- }
- int main() {
- int T;
- scanf("%d", &T);
- while (T--) {
- scanf("%d%d", &n, &p);
- for (int i = ; i <= n; i++)
- dp[i] = sum[i] = , G[i].clear();
- for (int i = , u, v, w; i < n; i++) {
- scanf("%d%d%d", &u, &v, &w);
- G[u].push_back(std::pii(v, w));
- G[v].push_back(std::pii(u, w));
- }
- dp[] = -p;
- que[l = r = ] = ;
- ans = ;
- dfs();
- printf("%lld\n", ans);
- }
- return ;
- }
基环树先找出环,用一个数组 $a$ 记录位置。然后以环上的每个点为根,去bfs非环上的点得到bfs序.
那么对于非环上的点,与他距离不大于 $k$ 的点bfs序连续,对于环上的点,与他距离不大于 $k$ 的点在 $a$ 数组里连续
这样就可以用线段树维护了。
然后就是恶心的细节了。
- #include <bits/stdc++.h>
- #define ll long long
- inline void checkmax(int &a, int b) {
- if (a < b) a = b;
- }
- inline void checkmin(int &a, int b) {
- if (a > b) a = b;
- }
- const int N = 1e5 + ;
- std::vector<int> vec[N];
- int fa[N], son[N], n, id[N], BCC[N], a[N];
- int tol;
- bool vis[N];
- void getBCC(int u, int v) {
- for (int i = u; i != v; i = fa[i])
- a[BCC[i] = ++a[]] = i;
- a[BCC[v] = ++a[]] = v;
- }
- void dfs(int u, int pre) {
- vis[u] = ;
- for (int v: vec[u]) {
- if (v == pre) continue;
- if (!vis[v]) {
- fa[v] = u;
- dfs(v, u);
- } else if (!BCC[u]) {
- getBCC(u, v);
- }
- }
- }
- int que[N];
- int ls[N], rs[N], lg[N], rg[N];
- void bfs(int s) {
- int h = , t = ;
- que[++t] = s;
- id[s] = ++tol;
- while (h != t) {
- int u = que[++h]; ls[u] = tol + ;
- for (int v: vec[u])
- if (!id[v] && !BCC[v]) {
- fa[v] = u;
- que[++t] = v;
- id[v] = ++tol;
- }
- rs[u] = tol;
- }
- for (int i = ; i <= t; i++) {
- int u = que[i];
- lg[u] = N; rg[u] = -N;
- for (int v: vec[u])
- if (id[v] > id[u] && !BCC[v])
- checkmin(lg[u], ls[v]), checkmax(rg[u], rs[v]);
- }
- }
- struct Seg {
- #define lp p << 1
- #define rp p << 1 | 1
- static const int NN = N * ;
- ll sum[NN], lazy[NN];
- void build(int p, int l, int r) {
- sum[p] = lazy[p] = ;
- if (l == r) return;
- int mid = l + r >> ;
- build(lp, l, mid);
- build(rp, mid + , r);
- }
- inline void pushup(int p) {
- sum[p] = sum[lp] + sum[rp];
- }
- inline void tag(int p, int len, ll w) {
- lazy[p] += w;
- sum[p] += len * w;
- }
- inline void pushdown(int p, int llen, int rlen) {
- if (lazy[p]) {
- tag(lp, llen, lazy[p]);
- tag(rp, rlen, lazy[p]);
- lazy[p] = ;
- }
- }
- void update(int p, int l, int r, int x, int y, int w) {
- if (x > r || l > y) return;
- if (x <= l && y >= r) {
- tag(p, r - l + , w);
- return;
- }
- int mid = l + r >> ;
- pushdown(p, mid - l + , r - mid);
- if (x <= mid) update(lp, l, mid, x, y, w);
- if (y > mid) update(rp, mid + , r, x, y, w);
- pushup(p);
- }
- ll query(int p, int l, int r, int x, int y) {
- if (x > r || l > y) return ;
- if (x <= l && y >= r) return sum[p];
- int mid = l + r >> ;
- pushdown(p, mid - l + , r - mid);
- ll ans = ;
- if (x <= mid) ans += query(lp, l, mid, x, y);
- if (y > mid) ans += query(rp, mid + , r, x, y);
- return ans;
- }
- inline int getl(int x) { return (x > ) ? x - : a[]; }
- inline int getr(int x) { return (x < a[]) ? x + : ; }
- inline void update(int u, int w) { update(, , tol, u, u, w); }
- inline ll query(int u) { return query(, , tol, u, u); }
- void update(int u, int k, int w) {
- update(id[u], w);
- if (k >= ) {
- update(, , tol, ls[u], rs[u], w);
- if (!BCC[u]) update(id[fa[u]], w);
- else {
- update(id[a[getl(BCC[u])]], w);
- update(id[a[getr(BCC[u])]], w);
- }
- }
- if (k >= ) {
- update(, , tol, lg[u], rg[u], w);
- if (!BCC[u]) {
- update(id[u], -w);
- update(, , tol, ls[fa[u]], rs[fa[u]], w);
- int p = BCC[fa[u]];
- if (!p) update(id[fa[fa[u]]], w);
- else update(id[a[getl(p)]], w), update(id[a[getr(p)]], w);
- } else {
- int bl = getl(BCC[u]), br = getr(BCC[u]);
- update(, , tol, ls[a[bl]], rs[a[bl]], w);
- update(, , tol, ls[a[br]], rs[a[br]], w);
- if (getl(bl) != br) update(id[a[getl(bl)]], w);
- if (getr(br) != bl && getr(br) != getl(bl)) update(id[a[getr(br)]], w);
- }
- }
- }
- ll query(int u, int k) {
- ll ans = query(id[u]);
- if (k >= ) {
- ans += query(, , tol, ls[u], rs[u]);
- if (!BCC[u]) ans += query(id[fa[u]]);
- else ans += query(id[a[getl(BCC[u])]]) + query(id[a[getr(BCC[u])]]);
- }
- if (k >= ) {
- ans += query(, , tol, lg[u], rg[u]);
- if (!BCC[u]) {
- ans += query(, , tol, ls[fa[u]], rs[fa[u]]) - query(id[u]);
- int p = BCC[fa[u]];
- if (!p) ans += query(id[fa[fa[u]]]);
- else ans += query(id[a[getl(p)]]) + query(id[a[getr(p)]]);
- } else {
- int bl = getl(BCC[u]), br = getr(BCC[u]);
- ans += query(, , tol, ls[a[bl]], rs[a[bl]]) + query(, , tol, ls[a[br]], rs[a[br]]);
- if (getl(bl) != br) ans += query(id[a[getl(bl)]]);
- if (getr(br) != bl && getr(br) != getl(bl)) ans += query(id[a[getr(br)]]);
- }
- }
- return ans;
- }
- } seg;
- int main() {
- freopen("in.txt", "r", stdin);
- int T;
- scanf("%d", &T);
- while (T--) {
- scanf("%d", &n);
- tol = a[] = ;
- for (int i = ; i <= n; i++) {
- fa[i] = son[i] = BCC[i] = id[i] = vis[i] = ;
- vec[i].clear();
- }
- for (int i = , u, v; i <= n; i++) {
- scanf("%d%d", &u, &v);
- vec[u].push_back(v);
- vec[v].push_back(u);
- }
- dfs(, );
- for (int i = ; i <= a[]; i++)
- bfs(a[i]);
- seg.build(, , tol);
- int q;
- scanf("%d", &q);
- while (q--) {
- int u, k;
- char s[];
- scanf("%s%d%d", s, &u, &k);
- if (s[] == 'M') {
- int d;
- scanf("%d", &d);
- seg.update(u, k, d);
- } else {
- printf("%lld\n", seg.query(u, k));
- }
- }
- }
- return ;
- }
终于把这道题给补了...比赛前我居然还在搞这些不考的东西
裸 K 短路,并且是有向无环图,求 $T$ 到其他点的最短路拓扑排序就能解决。
求出 $T$ 到其他所有点的最短路树,记 $d[i]$ 为 $i$ 到 $T$ 的最短路。给每一个点分配一个前趋,如果多个相同则选其中一个。(注意有重边时要记录边而不是记录前趋的点!!!)
走 $S$ 到 $T$ 上的树边即为最短路,走一条非树边 $(u, v, c)$ 则会使费用增大 $d[v] + c - d[u]$。记该花费为非树边的费用,显然树边的费用为 $0$。
将一条从 $S$ 到 $T$ 的路径记为其经过的非树边序列 $p$,那么这条路径的权值和即为 $d[s] + \sum_{(u,v,c) \in p} (d[v] + c - d[u])$
求 $k$ 短路即求第 $k$ 小的合法非树边序列费用之和。
合法的非树边序列为相邻两条非树边 $e$,$f$,$e$ 在 $f$ 之前,$head(f)$ 需为 $tail(e)$ 在 $T$ 上的祖先或相同。
用一个堆来存储搜索的状态,当前的非树边权值和为优先级,再存储最后一条非树边的起点。
往后可以有两个决策,一为直接加上最后一条非树边的终点之后的非树边中,权值最小的那个。
二为将最后一条非树边替换为 $u$ 之后的非树边下一条比这条非树边大的。
发现需要用另一个堆维护每个点往后所有的非树边。发现 $u$ 和 $pre[u]$ 大部分非树边相同,只是多了一些以自身为起点的非树边,那么可以可持久化地添加非树边。
可以用可持久化左偏树,虽然论文里说的是它不可完全可持久化。
还是挺好写的,就是在合并的过程中用新的节点来合并。像线段树合并。
- #include <bits/stdc++.h>
- #define pii pair<int, int>
- #define fi first
- #define se second
- const int N = 4e5 + ;
- const int INF = 0x3f3f3f3f;
- template<class T>inline bool chkmin(T &a, const T &b) { return a > b ? a = b, true : false; }
- template<class T>inline bool chkmax(T &a, const T &b) { return a < b ? a = b, true : false; }
- namespace Heap {
- struct Node {
- int lp, rp, v, dis, val;
- } tree[N * ];
- int tol, root[N];
- int newnode(int x = , int y = ) {
- int p = ++tol;
- tree[p].val = x, tree[p].v = y;
- tree[p].lp = tree[p].rp = ; tree[p].dis = ;
- return p;
- }
- int merge(int p, int q) {
- if (!p || !q) return p + q;
- if (tree[p].val < tree[q].val) std::swap(p, q);
- int x = newnode();
- tree[x] = tree[p];
- tree[x].rp = merge(tree[x].rp, q);
- if (tree[tree[x].lp].dis < tree[tree[x].rp].dis) std::swap(tree[x].lp, tree[x].rp);
- tree[x].dis = tree[tree[x].rp].dis + ;
- return x;
- }
- inline void init() {
- tol = ;
- tree[].dis = -;
- }
- } using namespace Heap;
- int n, k, cnt, id[N][], s, t, deg[N];
- int head1[N], head2[N], e, d[N];
- struct Ed {
- int v, ne, c;
- } E[N];
- void add(int u, int v, int c) {
- E[cnt].v = v; E[cnt].ne = head1[u]; E[cnt].c = c; head1[u] = cnt++;
- E[cnt].v = u; E[cnt].ne = head2[v]; E[cnt].c = c; head2[v] = cnt++;
- deg[u]++;
- }
- int Q[N], pre[N];
- int topo(int S, int *head) {
- int l = , r = ;
- Q[r++] = S;
- d[S] = ;
- while (l <= r) {
- int u = Q[l++];
- for (int i = head[u]; ~i; i = E[i].ne) {
- int v = E[i].v;
- if (chkmax(d[v], d[u] + E[i].c)) pre[v] = i;
- if (!--deg[v]) Q[r++] = v;
- }
- }
- return r;
- }
- int solve() {
- scanf("%d%d", &n, &k);
- cnt = e = ;
- init();
- for (int i = ; i <= n; i++)
- id[i][] = ++e, id[i][] = ++e;
- s = ++e, t = ++e;
- for (int i = ; i <= e; i++) {
- root[i] = ;
- d[i] = -INF;
- pre[i] = -;
- deg[i] = ;
- head1[i] = head2[i] = -;
- }
- for (int i = ; i < n; i++) {
- int a, b, c;
- scanf("%d%d%d", &a, &b, &c);
- add(id[i][], id[i + ][c], a);
- add(id[i][], id[i + ][c], b);
- add(id[i][], id[i + ][], );
- add(id[i][], id[i + ][], );
- }
- add(s, id[][], );
- add(id[n][], t, );
- add(id[n][], t, );
- int m = topo(t, head2);
- if (k == ) return d[s];
- for (int j = ; j < m; j++) {
- int u = Q[j];
- for (int i = head1[u]; ~i; i = E[i].ne) {
- int v = E[i].v;
- if (d[v] == INF || (i ^ ) == pre[u]) continue;
- root[u] = merge(root[u], newnode(d[v] + E[i].c - d[u], v));
- }
- if (~pre[u]) root[u] = merge(root[u], root[E[pre[u] ^ ].v]);
- }
- std::priority_queue<std::pii> que;
- que.push(std::pii(d[s] + tree[root[s]].val, root[s]));
- for (k -= ; k; k--) {
- auto p = que.top(); que.pop();
- int i = p.se, j = root[tree[i].v];
- if (j) que.push(std::pii(p.fi + tree[j].val, j));
- if (tree[i].lp) que.push(std::pii(p.fi - tree[i].val + tree[tree[i].lp].val, tree[i].lp));
- if (tree[i].rp) que.push(std::pii(p.fi - tree[i].val + tree[tree[i].rp].val, tree[i].rp));
- }
- return que.top().fi;
- }
- int main() {
- freopen("in.txt", "r", stdin);
- int T;
- scanf("%d", &T);
- while (T--) printf("%d\n", solve());
- return ;
- }
2016 ACM/ICPC亚洲区沈阳站的更多相关文章
- 2016 ACM/ICPC亚洲区青岛站现场赛(部分题解)
摘要 本文主要列举并求解了2016 ACM/ICPC亚洲区青岛站现场赛的部分真题,着重介绍了各个题目的解题思路,结合详细的AC代码,意在熟悉青岛赛区的出题策略,以备战2018青岛站现场赛. HDU 5 ...
- 2016 ACM/ICPC亚洲区大连站-重现赛 解题报告
任意门:http://acm.hdu.edu.cn/showproblem.php?pid=5979 按AC顺序: I - Convex Time limit 1000 ms Memory li ...
- 2016 ACM/ICPC亚洲区大连站 F - Detachment 【维护前缀积、前缀和、二分搜索优化】
F - Detachment In a highly developed alien society, the habitats are almost infinite dimensional spa ...
- 2016 ACM/ICPC亚洲区青岛站
A B C D E F G H I J K L M O O O O $\varnothing$ $\varnothing$ $\varnothing$ $\varnothing$ ...
- HDU 5950 Recursive sequence 【递推+矩阵快速幂】 (2016ACM/ICPC亚洲区沈阳站)
Recursive sequence Time Limit: 2000/1000 MS (Java/Others) Memory Limit: 65536/65536 K (Java/Other ...
- HDU 5952 Counting Cliques 【DFS+剪枝】 (2016ACM/ICPC亚洲区沈阳站)
Counting Cliques Time Limit: 8000/4000 MS (Java/Others) Memory Limit: 65536/65536 K (Java/Others) ...
- HDU 5948 Thickest Burger 【模拟】 (2016ACM/ICPC亚洲区沈阳站)
Thickest Burger Time Limit: 2000/1000 MS (Java/Others) Memory Limit: 65536/65536 K (Java/Others)T ...
- HDU 5949 Relative atomic mass 【模拟】 (2016ACM/ICPC亚洲区沈阳站)
Relative atomic mass Time Limit: 2000/1000 MS (Java/Others) Memory Limit: 65536/65536 K (Java/Oth ...
- 2015ACM/ICPC亚洲区沈阳站 Pagodas
Pagodas Time Limit: 2000/1000 MS (Java/Others) Memory Limit: 65536/65536 K (Java/Others)Total Sub ...
随机推荐
- http与tcp,udp的区别
1.网络协议的概念 (1)在学习网络课程的时候,老师会讲iso七层模型,有应用层 表示层 会话层 传输层 网络层 数据链路层 物理层,其中http就属于应用层,tcp与udp是属于传输层,如图1.1( ...
- .NET Core:过滤器
过滤器的作用是在 Action 方法执行前或执行后做一些加工处理.使用过滤器可以避免Action方法的重复代码.功能上更贴合业务的使用过滤器. 在Startup中的ConfigureServices方 ...
- 深入理解C语言 - 指针使用的常见错误
在C语言中,指针的重要性不言而喻,但在很多时候指针又被认为是一把双刃剑.一方面,指针是构建数据结构和操作内存的精确而高效的工具.另一方面,它们又很容易误用,从而产生不可预知的软件bug.下面总结一下指 ...
- python数据分析开发中的常用整理
Pandas操作 python使用pandas读取csv import pandas as pd #数据筛选 usetTable = pd.read_csv(filename,header = 0) ...
- Unity Shader 屏幕后效果——高斯模糊
高斯模糊是图像模糊处理中非常经典和常见的一种算法,也是Bloom屏幕效果的基础. 实现高斯模糊同样用到了卷积的概念,关于卷积的概念和原理详见我的另一篇博客: https://www.cnblogs.c ...
- AbstractExecutorService源码
public class RunnableFutureTask { static FinalizableDelegatedExecutorService executorService = (Fina ...
- 关于wordpress4.8中的Twenty Seventeen主题的主题选项增加章节的实现
我这里的wordpress版本是4.8 默认的主题是 Twenty Seventeen 我想实现的事 主题选项的首页 多增加2个章节 默认是只有4个章节 我想在增加2个 到6个 看下实现后的效果 ...
- loadrunner:Action.c(4): Error -27796: Failed to connect to server "10.8.251.101:10086": [10060] Connection timed out
Action.c(4): Error -27796: Failed to connect to server "10.8.251.101:10086": [10060] Conne ...
- Form之action提交不刷新不跳转
<div class="file-box"> <form action="/File/fileUpLoad" id="form1&q ...
- centos7 配置nginx vim语法高亮
看了Nginx核心知识100讲,按照他的做法,没有配置成功,可以使用下面的方法: 下载nginx源码,http://nginx.org/en/download.html 这里下载的是:nginx-1. ...