#include using namespace std; using Int = long long; const char newl = '\n'; template inline void chmin(T1 &a,T2 b){if(a>b) a=b;} template inline void chmax(T1 &a,T2 b){if(a void drop(const T &x){cout< vector read(size_t n){ vector ts(n); for(size_t i=0;i>ts[i]; return ts; } template V compress(V vs){ sort(vs.begin(),vs.end()); vs.erase(unique(vs.begin(),vs.end()),vs.end()); return vs; } template map dict(const vector &vs){ map res; for(int i=0;i<(int)vs.size();i++) res[vs[i]]=i; return res; } map dict(const string &s){ return dict(vector(s.begin(),s.end())); } template vector fusion(vector bs,Ts... ts){ auto append=[&](auto vs){for(auto v:vs) bs.emplace_back(v);}; initializer_list{(void(append(ts)),0)...}; return bs; } //INSERT ABOVE HERE template struct Slope{ using P = pair; multiset

L,R; T offL,offR,entire; Slope():offL(0),offR(0),entire(0){} inline T relu(T x){return max(0,x);} void pushL(T pos,T num){if(num>T(0)) L.emplace(pos-offL,num);} void pushR(T pos,T num){if(num>T(0)) R.emplace(pos-offR,num);} T posL(){return L.rbegin()->first+offL;} T posR(){return R.begin() ->first+offR;} void add_x_minus_a(T a,T cnt=T(1)){ T use(0); while(usesecond; L.erase(--L.end()); T tmp=min(cnt-use,num); pushR(pos,tmp); pushL(a,tmp); pushL(pos,relu(num-tmp)); entire+=relu(pos-a)*tmp; use+=tmp; } pushR(a,cnt-use); } void add_a_minus_x(T a,T cnt=T(1)){ T use(0); while(usesecond; R.erase(R.begin()); T tmp=min(cnt-use,num); pushL(pos,tmp); pushR(a,tmp); pushR(a,relu(num-tmp)); entire+=relu(a-pos)*tmp; use+=tmp; } pushL(a,cnt-use); } // f_{new}(x) = f_{old}(x + diff) void shift(T diff){ offL-=diff; offR-=diff; } // f_{new}(x) = min_{y<=x} f_{old}(y) void apply_cumulative_min(){ R.clear(); } T get_val(T x){ T res=entire; for(auto[pos,num]:L) res+=relu((pos+offL)-x)*num; for(auto[pos,num]:R) res+=relu(x-(pos+offR))*num; return res; } }; signed main(){ cin.tie(0); ios::sync_with_stdio(0); int m,n; cin>>m>>n; auto as=read(m); auto bs=read(n); auto cs=compress(fusion(as,bs)); auto dc=dict(cs); for(int &a:as) a=dc[a]; for(int &b:bs) b=dc[b]; const int sz = cs.size(); vector num(sz,0); for(int a:as) num[a]--; using ll = long long; ll last=-1; for(int k=1;k<=m;k++){ for(int b:bs) num[b]++; int pos=0; Slope S; S.add_a_minus_x(pos,1e9); for(int i=0;i