This submission is migrated from previous version of oj.uz, which used different machine for grading. This submission may have different result if resubmitted.
#include "squad.h"
#include <vector>
#include <algorithm>
#include <cassert>
struct PT {
typedef long long T;
T x, y;
PT(T xx = 0, T yy = 0) : x(xx), y(yy){}
PT operator +(const PT &p) const { return PT(x+p.x,y+p.y); }
PT operator -(const PT &p) const { return PT(x-p.x,y-p.y); }
PT operator *(T c) const { return PT(x*c,y*c); }
T operator *(const PT &p) const { return x*p.x+y*p.y; }
T operator %(const PT &p) const { return x*p.y-y*p.x; }
// be careful using this without eps!
bool operator<(const PT &p)const { return x != p.x ? x < p.x : y < p.y; }
bool operator==(const PT &p)const{ return x == p.x && y == p.y; }
};
std::vector<PT> splitHull(const std::vector<PT> &hull) {
std::vector<PT> ans(hull.size());
for(int i = 0, j = (int) hull.size()-1, k = 0; k < (int) hull.size(); k++) {
if(hull[i] < hull[j]) {
ans[k] = hull[i++];
} else {
ans[k] = hull[j--];
}
}
return ans;
}
std::vector<PT> ConvexHull (std::vector<PT> pts, bool needs = true) {
if(needs) {
std::sort(pts.begin(), pts.end());
}
pts.resize(std::unique(pts.begin(), pts.end()) - pts.begin());
if(pts.size() <= 1) return pts;
std::vector<PT> ans(pts.size() + 2);
int s = 0;
for(int i = 0; i < (int) pts.size(); i++) {
while(s > 1 && (pts[i] - ans[s - 2]) % (ans[s - 1] - ans[s - 2]) >= 0) {
s--;
}
ans[s++] = pts[i];
}
for(int i = (int) pts.size() - 2, t = s + 1; i >= 0; i--) {
while(s >= t && (pts[i] - ans[s - 2]) % (ans[s - 1] - ans[s - 2]) >= 0) {
s--;
}
ans[s++] = pts[i];
}
ans.resize(s-1);
return ans;
}
std::vector<PT> ConvexHull(const std::vector<PT> &a, const std::vector<PT> &b) {
auto A = splitHull(a);
auto B = splitHull(b);
std::vector<PT> C(A.size() + B.size());
std::merge(A.begin(), A.end(), B.begin(), B.end(), C.begin());
return ConvexHull(C, false);
}
int maximizeScalarProduct(const std::vector<PT> &hull, PT vec) {
int ans = 0;
int n = hull.size();
if(n < 20) {
for(int i = 0; i < n; i++) {
if(hull[i] * vec > hull[ans] * vec) {
ans = i;
}
}
} else {
int diff = 1;
if(hull[0] * vec == hull[1] * vec) {
int l = 2, r = n - 1;
while(l != r) {
int mid = (l + r) / 2;
if((hull[1] - hull[0]) * (hull[mid] - hull[0]) > 0 && (hull[1] - hull[0]) % (hull[mid] - hull[0]) == 0) {
l = mid + 1;
} else {
r = mid;
}
}
diff = l;
//diff = 2;
}
if(hull[0] * vec < hull[diff] * vec) {
int l = diff, r = n - 1;
while(l != r) {
int mid = (l + r + 1) / 2;
if(hull[mid] * vec >= hull[mid - 1] * vec && hull[mid] * vec >= hull[0] * vec) {
l = mid;
} else {
r = mid - 1;
}
}
if(hull[0] * vec < hull[l] * vec) {
ans = l;
}
} else {
int l = diff, r = n - 1;
while(l != r) {
int mid = (l + r + 1) / 2;
if(hull[mid] * vec >= hull[mid - 1] * vec || hull[mid - 1] * vec < hull[0] * vec) {
l = mid;
} else {
r = mid - 1;
}
}
if(hull[0] * vec < hull[l] * vec) {
ans = l;
}
}
}
return ans;
}
bool comp(PT a, PT b){
int hp1 = (a.x < 0 || (a.x==0 && a.y<0));
int hp2 = (b.x < 0 || (b.x==0 && b.y<0));
if(hp1 != hp2) return hp1 < hp2;
long long R = a%b;
if(R) return R > 0;
return a*a < b*b;
}
std::vector<PT> minkowskiSum(const std::vector<PT> &a, const std::vector<PT> &b){
if(a.empty() || b.empty()) return std::vector<PT>(0);
std::vector<PT> ret;
if(std::min(a.size(), b.size()) < 2){
for(int i = 0; i < (int) a.size(); i++) {
for(int j = 0; j < (int) b.size(); j++) {
ret.push_back(a[i]+b[j]);
}
}
}
ret.push_back(a[0]+b[0]);
PT p = ret.back();
int pa = 0, pb = 0;
while(pa + pb + 1 < a.size()+b.size()){
if(pb == (int) b.size() || (pa < (int) a.size() && comp((a[(pa+1)%a.size()]-a[pa]),(b[(pb+1)%b.size()]-b[pb]))))
p = p + (a[(pa+1)%a.size()]-a[pa]), pa++;
else p = p + (b[(pb+1)%b.size()]-b[pb]), pb++;
while(ret.size() >= 2 && (p-ret.back()) % (p-ret[(int)ret.size()-2]) == 0) {
ret.pop_back();
}
ret.push_back(p);
}
return ret;
}
const int ms = 300300;
PT h1[ms], h2[ms], tmp[ms];
std::vector<PT> solve(int l, int r) {
if(r - l <= 1) {
return std::vector<PT>(0);
}
int mid = (l + r) / 2;
std::vector<PT> ans = ConvexHull(solve(l, mid), solve(mid, r));
std::vector<PT> other;
{
std::vector<PT> hull[2];
for(int i = l; i < mid; i++) {
hull[0].push_back(h1[i]);
}
for(int i = mid; i < r; i++) {
hull[1].emplace_back(h2[i]);
}
other = minkowskiSum(ConvexHull(hull[0], false), ConvexHull(hull[1], false));
}
{
std::vector<PT> hull[2];
for(int i = l; i < mid; i++) {
hull[0].push_back(h2[i]);
}
for(int i = mid; i < r; i++) {
hull[1].emplace_back(h1[i]);
}
other = ConvexHull(other, minkowskiSum(ConvexHull(hull[0], false), ConvexHull(hull[1], false)));
}
std::merge(h1 + l, h1 + mid, h1 + mid, h1 + r, tmp + l);
for(int i = l; i < r; i++) {
h1[i] = tmp[i];
}
std::merge(h2 + l, h2 + mid, h2 + mid, h2 + r, tmp + l);
for(int i = l; i < r; i++) {
h2[i] = tmp[i];
}
return ConvexHull(ans, other);
}
std::vector<PT> tot;
void Init(std::vector<int> A, std::vector<int> D, std::vector<int> P){
int N = A.size();
for(int i = 0; i < N; i++) {
h1[i] = PT(A[i], P[i]);
h2[i] = PT(D[i], P[i]);
}
tot = solve(0, N);
}
long long BestSquad(int X, int Y){
PT cur(X, Y);
return cur * tot[maximizeScalarProduct(tot, cur)];
}
Compilation message (stderr)
squad.cpp: In function 'std::vector<PT> minkowskiSum(const std::vector<PT>&, const std::vector<PT>&)':
squad.cpp:143:20: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
while(pa + pb + 1 < a.size()+b.size()){
~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~
# | Verdict | Execution time | Memory | Grader output |
---|
Fetching results... |
# | Verdict | Execution time | Memory | Grader output |
---|
Fetching results... |
# | Verdict | Execution time | Memory | Grader output |
---|
Fetching results... |