#include <bits/stdc++.h> using namespace std; const int N = 500 + 7; const int A = 20000 + 7; // https://judge.yosupo.jp/submission/195925 const int MOD = 998244353; const long long MOD2 = (long long) MOD * MOD; const int root = 3; const int alim = 64; // Bound for using O(n^2) polynomial mult int modpow(int b, int e) { int ans = 1; for (; e; b = (long long) b * b % MOD, e /= 2) if (e & 1) ans = (long long) ans * b % MOD; return ans; } const int MODinv = 2 - MOD; // pow(-MOD, -1, 2**32) inline int m_reduce(long long x) { int m = x * MODinv; return (x>>32) - (((long long) m * MOD) >> 32); } const int r2 = modpow(2, 64); inline int m_transform(int x) { return m_reduce((long long)x * r2); } inline int m_add(int x, int y) { int z = x + y; return z < 0 ? z + MOD : z - MOD; } inline int m_sub(int x, int y) { int z = x - y; return z < 0 ? z + MOD : z - MOD; } inline int m_mult(int x, int y) { return m_reduce((long long) x * y); } vector<int> rt = {1}; vector<int> transformed_rt; vector<int> transformed_rt2; template<int a> void transform(vector<int> &P) { int m = P.size(); int n = m / a; int size = rt.size(); while (2 * size < n) { rt.resize(n / 2); int r = modpow(root, MOD / (4 * size)); for (int i = 0; i < size; ++i) rt[i + size] = (long long) r * rt[i] % MOD; size *= 2; } // For montgomery for (int i = transformed_rt.size(); i < rt.size(); ++i) { transformed_rt.resize(rt.size()); transformed_rt[i] = m_transform(rt[i]); transformed_rt2.resize(rt.size()); transformed_rt2[i] = (unsigned int) MODinv * transformed_rt[i]; } // Radix 4 recursive NTT auto dfs = [&](auto &&self, int i, int k) -> void { if (k == 1) return; int step = k * a; int quarter_step = step / 4; int R20 = transformed_rt2[2 * i]; int RR0 = transformed_rt[2 * i]; int R21 = transformed_rt2[2 * i + 1]; int RR1 = transformed_rt[2 * i + 1]; int R2 = transformed_rt2[i]; int RR = transformed_rt[i]; int *P1 = &P[i * step]; int *P2 = P1 + quarter_step; int *P3 = P2 + quarter_step; int *P4 = P3 + quarter_step; #pragma GCC ivdep for (int j = 0; j < quarter_step; ++j) { int z0; { int z = P3[j]; int m = (unsigned int) R2 * z; z0 = ((long long) z * RR - (long long) m * MOD) >> 32; } int z1; { int z = P4[j]; int m = (unsigned int) R2 * z; z1 = ((long long) z * RR - (long long) m * MOD) >> 32; } int sum0 = m_add(P1[j], z0); int diff0 = m_sub(P1[j], z0); int sum1 = P2[j] + z1; int diff1 = P2[j] - z1; // [sum0, sum1, diff0, diff1] int zz0; { int z = sum1; int m = (unsigned int) R20 * z; zz0 = ((long long) z * RR0 - (long long) m * MOD) >> 32; } int zz1; { int z = diff1; int m = (unsigned int) R21 * z; zz1 = ((long long) z * RR1 - (long long) m * MOD) >> 32; } P1[j] = m_add(sum0, zz0); P2[j] = m_sub(sum0, zz0); P3[j] = m_add(diff0, zz1); P4[j] = m_sub(diff0, zz1); } self(self, 4*i+0, k/4); self(self, 4*i+1, k/4); self(self, 4*i+2, k/4); self(self, 4*i+3, k/4); }; int k = n; while (k >= 4) k /= 4; if (k == 2) { int step = n * a; int half_step = step / 2; for (int j1 = 0; j1 < half_step; ++j1) { int j2 = j1 + half_step; int diff = m_sub(P[j1], P[j2]); P[j1] = m_add(P[j1], P[j2]); P[j2] = diff; } k = n/2; dfs(dfs, 0, k); dfs(dfs, 1, k); } else { k = n; dfs(dfs, 0, k); } for (int i = 0; i < m; ++i) if (P[i] < 0) P[i] += MOD; } template<int a> void inverse_transform(vector<int> &P) { int m = P.size(); int n = m / a; int n_inv = m_transform(modpow(n, MOD - 2)); vector<int> rev(n); for (int i = 1; i < n; ++i) { rev[i] = rev[i / 2] / 2 + (i & 1) * n / 2; } // P = [p * n_inv for p in P] for (int i = 0; i < m; ++i) P[i] = m_mult(n_inv, P[i]); // P = [P[a * rev[i // a] + (i % a)] for i in range(m)] for (int i = 1; i < n; ++i) if (i < rev[i]) swap_ranges(P.begin() + a * i, P.begin() + a * i + a, P.begin() + a * rev[i]); // P = [P[-a * (i // a) + (i % a)] for i in range(m)] for (int i = 1; i < n/2; ++i) swap_ranges(P.begin() + a * i, P.begin() + a * i + a, P.begin() + a * (n - i)); transform<a>(P); // P = [P[a * rev[i // a] + (i % a)] for i in range(m)] for (int i = 1; i < n; ++i) if (i < rev[i]) swap_ranges(P.begin() + a * i, P.begin() + a * i + a, P.begin() + a * rev[i]); } template<int a> void fast_polymult_mod(vector<int> &P, vector<int> &Q) { int m = P.size(); int n = m / a; transform<a>(P); transform<a>(Q); vector<int> &PQ = P; for (int i = 0; i < n; ++i) { vector<unsigned long long> res(2 * a); for (int j = 0; j < a; ++j) { if (j >= 10 && j % 9 == 8) for (int k = j; k < j + a - 10; ++k) res[k] -= (res[k] >> 63) * 9 * MOD2; for (int k = 0; k < a; ++k) res[j + k] += (long long) P[i * a + j] * Q[i * a + k]; } int c = rt[i/2]; if (i & 1) c = MOD - c; for (int j = 0; j < a; ++j) PQ[i * a + j] = (res[j] + c * (res[j + a] % MOD)) % MOD; } inverse_transform<a>(PQ); } template <size_t... N> void work(std::index_sequence<N...>, int x, std::vector<int>& a, std::vector<int>& b) { static void (*ptrs[])(std::vector<int>&, std::vector<int>&) = {&fast_polymult_mod<N+1>...}; ptrs[x - 1](a, b); } void fast_polymult(vector<int> &P, vector<int> &Q) { int m1 = P.size(); int m2 = Q.size(); int res_len = m1 + m2 - 1; int b = 1; while ((alim << b) < res_len) ++b; int a = ((res_len - 1) >> b) + 1; int m = a << b; P.resize(m); Q.resize(m); // Call fast_polymult_mod<a>(P, Q); work(std::make_index_sequence<alim>{}, a, P, Q); P.resize(res_len); } int n, a[N]; vector<int> square(vector<int> P) { vector<int> Q = P; fast_polymult(P, Q); return P; /* vector<int> Q(2 * P.size() - 1); for (int i = 0; i < (int) P.size(); i++) for (int j = 0; j < (int) P.size(); j++) Q[i + j] += P[i] * P[j]; return Q; */ } int main() { ios::sync_with_stdio(false), cin.tie(nullptr); cin >> n; for (int i = 1; i <= n; i++) { cin >> a[i]; } vector<int> b; int zero = 0; for (int i = 1; i <= n; i++) { int sum = 0; for (int j = i; j <= n; j++) { sum += a[j]; if (sum == 0) { zero++; } else { b.push_back(sum); } } } int shift = (b.empty() ? 0 : *min_element(b.begin(), b.end())); int mx = (b.empty() ? 0 : *max_element(b.begin(), b.end())); long long ans = 1LL * zero * (zero - 1) * (zero - 2) / 6; if (shift > 0 || mx < 0) { cout << ans << '\n'; return 0; } vector<int> Q(mx - shift + 1); for (int x : b) { Q[x - shift]++; } cerr << "sz: " << Q.size() << '\n'; vector<int> P = square(Q); for (int x : b) { P[2 * (x - shift)]--; } for (int i = 0; i < (int) P.size(); i++) P[i] /= 2; ans += 1LL * P[2 * (0 - shift)] * zero; long long three = 0; for (int x : b) if (0 <= -x - 2 * shift && -x - 2 * shift < (int) P.size()) { three += P[-x - 2 * shift]; if (0 <= -2 * x - shift && -2 * x - shift < (int) Q.size()) { three -= Q[-2 * x - shift]; } } ans += three / 3; cout << ans << '\n'; }
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 | #include <bits/stdc++.h> using namespace std; const int N = 500 + 7; const int A = 20000 + 7; // https://judge.yosupo.jp/submission/195925 const int MOD = 998244353; const long long MOD2 = (long long) MOD * MOD; const int root = 3; const int alim = 64; // Bound for using O(n^2) polynomial mult int modpow(int b, int e) { int ans = 1; for (; e; b = (long long) b * b % MOD, e /= 2) if (e & 1) ans = (long long) ans * b % MOD; return ans; } const int MODinv = 2 - MOD; // pow(-MOD, -1, 2**32) inline int m_reduce(long long x) { int m = x * MODinv; return (x>>32) - (((long long) m * MOD) >> 32); } const int r2 = modpow(2, 64); inline int m_transform(int x) { return m_reduce((long long)x * r2); } inline int m_add(int x, int y) { int z = x + y; return z < 0 ? z + MOD : z - MOD; } inline int m_sub(int x, int y) { int z = x - y; return z < 0 ? z + MOD : z - MOD; } inline int m_mult(int x, int y) { return m_reduce((long long) x * y); } vector<int> rt = {1}; vector<int> transformed_rt; vector<int> transformed_rt2; template<int a> void transform(vector<int> &P) { int m = P.size(); int n = m / a; int size = rt.size(); while (2 * size < n) { rt.resize(n / 2); int r = modpow(root, MOD / (4 * size)); for (int i = 0; i < size; ++i) rt[i + size] = (long long) r * rt[i] % MOD; size *= 2; } // For montgomery for (int i = transformed_rt.size(); i < rt.size(); ++i) { transformed_rt.resize(rt.size()); transformed_rt[i] = m_transform(rt[i]); transformed_rt2.resize(rt.size()); transformed_rt2[i] = (unsigned int) MODinv * transformed_rt[i]; } // Radix 4 recursive NTT auto dfs = [&](auto &&self, int i, int k) -> void { if (k == 1) return; int step = k * a; int quarter_step = step / 4; int R20 = transformed_rt2[2 * i]; int RR0 = transformed_rt[2 * i]; int R21 = transformed_rt2[2 * i + 1]; int RR1 = transformed_rt[2 * i + 1]; int R2 = transformed_rt2[i]; int RR = transformed_rt[i]; int *P1 = &P[i * step]; int *P2 = P1 + quarter_step; int *P3 = P2 + quarter_step; int *P4 = P3 + quarter_step; #pragma GCC ivdep for (int j = 0; j < quarter_step; ++j) { int z0; { int z = P3[j]; int m = (unsigned int) R2 * z; z0 = ((long long) z * RR - (long long) m * MOD) >> 32; } int z1; { int z = P4[j]; int m = (unsigned int) R2 * z; z1 = ((long long) z * RR - (long long) m * MOD) >> 32; } int sum0 = m_add(P1[j], z0); int diff0 = m_sub(P1[j], z0); int sum1 = P2[j] + z1; int diff1 = P2[j] - z1; // [sum0, sum1, diff0, diff1] int zz0; { int z = sum1; int m = (unsigned int) R20 * z; zz0 = ((long long) z * RR0 - (long long) m * MOD) >> 32; } int zz1; { int z = diff1; int m = (unsigned int) R21 * z; zz1 = ((long long) z * RR1 - (long long) m * MOD) >> 32; } P1[j] = m_add(sum0, zz0); P2[j] = m_sub(sum0, zz0); P3[j] = m_add(diff0, zz1); P4[j] = m_sub(diff0, zz1); } self(self, 4*i+0, k/4); self(self, 4*i+1, k/4); self(self, 4*i+2, k/4); self(self, 4*i+3, k/4); }; int k = n; while (k >= 4) k /= 4; if (k == 2) { int step = n * a; int half_step = step / 2; for (int j1 = 0; j1 < half_step; ++j1) { int j2 = j1 + half_step; int diff = m_sub(P[j1], P[j2]); P[j1] = m_add(P[j1], P[j2]); P[j2] = diff; } k = n/2; dfs(dfs, 0, k); dfs(dfs, 1, k); } else { k = n; dfs(dfs, 0, k); } for (int i = 0; i < m; ++i) if (P[i] < 0) P[i] += MOD; } template<int a> void inverse_transform(vector<int> &P) { int m = P.size(); int n = m / a; int n_inv = m_transform(modpow(n, MOD - 2)); vector<int> rev(n); for (int i = 1; i < n; ++i) { rev[i] = rev[i / 2] / 2 + (i & 1) * n / 2; } // P = [p * n_inv for p in P] for (int i = 0; i < m; ++i) P[i] = m_mult(n_inv, P[i]); // P = [P[a * rev[i // a] + (i % a)] for i in range(m)] for (int i = 1; i < n; ++i) if (i < rev[i]) swap_ranges(P.begin() + a * i, P.begin() + a * i + a, P.begin() + a * rev[i]); // P = [P[-a * (i // a) + (i % a)] for i in range(m)] for (int i = 1; i < n/2; ++i) swap_ranges(P.begin() + a * i, P.begin() + a * i + a, P.begin() + a * (n - i)); transform<a>(P); // P = [P[a * rev[i // a] + (i % a)] for i in range(m)] for (int i = 1; i < n; ++i) if (i < rev[i]) swap_ranges(P.begin() + a * i, P.begin() + a * i + a, P.begin() + a * rev[i]); } template<int a> void fast_polymult_mod(vector<int> &P, vector<int> &Q) { int m = P.size(); int n = m / a; transform<a>(P); transform<a>(Q); vector<int> &PQ = P; for (int i = 0; i < n; ++i) { vector<unsigned long long> res(2 * a); for (int j = 0; j < a; ++j) { if (j >= 10 && j % 9 == 8) for (int k = j; k < j + a - 10; ++k) res[k] -= (res[k] >> 63) * 9 * MOD2; for (int k = 0; k < a; ++k) res[j + k] += (long long) P[i * a + j] * Q[i * a + k]; } int c = rt[i/2]; if (i & 1) c = MOD - c; for (int j = 0; j < a; ++j) PQ[i * a + j] = (res[j] + c * (res[j + a] % MOD)) % MOD; } inverse_transform<a>(PQ); } template <size_t... N> void work(std::index_sequence<N...>, int x, std::vector<int>& a, std::vector<int>& b) { static void (*ptrs[])(std::vector<int>&, std::vector<int>&) = {&fast_polymult_mod<N+1>...}; ptrs[x - 1](a, b); } void fast_polymult(vector<int> &P, vector<int> &Q) { int m1 = P.size(); int m2 = Q.size(); int res_len = m1 + m2 - 1; int b = 1; while ((alim << b) < res_len) ++b; int a = ((res_len - 1) >> b) + 1; int m = a << b; P.resize(m); Q.resize(m); // Call fast_polymult_mod<a>(P, Q); work(std::make_index_sequence<alim>{}, a, P, Q); P.resize(res_len); } int n, a[N]; vector<int> square(vector<int> P) { vector<int> Q = P; fast_polymult(P, Q); return P; /* vector<int> Q(2 * P.size() - 1); for (int i = 0; i < (int) P.size(); i++) for (int j = 0; j < (int) P.size(); j++) Q[i + j] += P[i] * P[j]; return Q; */ } int main() { ios::sync_with_stdio(false), cin.tie(nullptr); cin >> n; for (int i = 1; i <= n; i++) { cin >> a[i]; } vector<int> b; int zero = 0; for (int i = 1; i <= n; i++) { int sum = 0; for (int j = i; j <= n; j++) { sum += a[j]; if (sum == 0) { zero++; } else { b.push_back(sum); } } } int shift = (b.empty() ? 0 : *min_element(b.begin(), b.end())); int mx = (b.empty() ? 0 : *max_element(b.begin(), b.end())); long long ans = 1LL * zero * (zero - 1) * (zero - 2) / 6; if (shift > 0 || mx < 0) { cout << ans << '\n'; return 0; } vector<int> Q(mx - shift + 1); for (int x : b) { Q[x - shift]++; } cerr << "sz: " << Q.size() << '\n'; vector<int> P = square(Q); for (int x : b) { P[2 * (x - shift)]--; } for (int i = 0; i < (int) P.size(); i++) P[i] /= 2; ans += 1LL * P[2 * (0 - shift)] * zero; long long three = 0; for (int x : b) if (0 <= -x - 2 * shift && -x - 2 * shift < (int) P.size()) { three += P[-x - 2 * shift]; if (0 <= -2 * x - shift && -2 * x - shift < (int) Q.size()) { three -= Q[-2 * x - shift]; } } ans += three / 3; cout << ans << '\n'; } |