#pragma GCC optimize ("Ofast")
#include<bits/stdc++.h>
using namespace std;
void*wmem;
char memarr[96000000];
template<class T> inline void walloc1d(T **arr, int x, void **mem = &wmem){
  static int skip[16] = {0, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1};
  (*mem) = (void*)( ((char*)(*mem)) + skip[((unsigned long long)(*mem)) & 15] );
  (*arr)=(T*)(*mem);
  (*mem)=((*arr)+x);
}
inline int my_getchar_unlocked(){
  static char buf[1048576];
  static int s = 1048576;
  static int e = 1048576;
  if(s == e && e == 1048576){
    e = fread_unlocked(buf, 1, 1048576, stdin);
    s = 0;
  }
  if(s == e){
    return EOF;
  }
  return buf[s++];
}
inline void rd(int &x){
  int k;
  int m=0;
  x=0;
  for(;;){
    k = my_getchar_unlocked();
    if(k=='-'){
      m=1;
      break;
    }
    if('0'<=k&&k<='9'){
      x=k-'0';
      break;
    }
  }
  for(;;){
    k = my_getchar_unlocked();
    if(k<'0'||k>'9'){
      break;
    }
    x=x*10+k-'0';
  }
  if(m){
    x=-x;
  }
}
struct MY_WRITER{
  char buf[1048576];
  int s;
  int e;
  MY_WRITER(){
    s = 0;
    e = 1048576;
  }
  ~MY_WRITER(){
    if(s){
      fwrite_unlocked(buf, 1, s, stdout);
    }
  }
}
;
MY_WRITER MY_WRITER_VAR;
void my_putchar_unlocked(int a){
  if(MY_WRITER_VAR.s == MY_WRITER_VAR.e){
    fwrite_unlocked(MY_WRITER_VAR.buf, 1, MY_WRITER_VAR.s, stdout);
    MY_WRITER_VAR.s = 0;
  }
  MY_WRITER_VAR.buf[MY_WRITER_VAR.s++] = a;
}
inline void wt_L(char a){
  my_putchar_unlocked(a);
}
inline void wt_L(long long x){
  int s=0;
  int m=0;
  char f[20];
  if(x<0){
    m=1;
    x=-x;
  }
  while(x){
    f[s++]=x%10;
    x/=10;
  }
  if(!s){
    f[s++]=0;
  }
  if(m){
    my_putchar_unlocked('-');
  }
  while(s--){
    my_putchar_unlocked(f[s]+'0');
  }
}
template<class S, class T> inline S chmin(S &a, T b){
  if(a>b){
    a=b;
  }
  return a;
}
template<class T> struct fenwick{
  int size;
  int memory;
  T*data;
  void malloc(int mem);
  void malloc(int mem, int fg);
  void walloc(int mem, void **workMemory = &wmem);
  void walloc(int mem, int fg, void **workMemory = &wmem);
  void free(void);
  void init(int N);
  void add(int k, T val);
  T get(int k);
  T range(int a, int b);
  int kth(T k);
}
;
int N;
int Q;
int A[20000];
int L;
int R;
long long x[20000+1];
long long y[20000+1];
long long h1[20000+1];
long long h2[20000+1];
fenwick<int> f;
int main(){
  int i, xr20shxY;
  wmem = memarr;
  long long res;
  rd(N);
  rd(Q);
  {
    int Lj4PdHRW;
    for(Lj4PdHRW=(0);Lj4PdHRW<(N);Lj4PdHRW++){
      rd(A[Lj4PdHRW]);A[Lj4PdHRW] += (-1);
    }
  }
  f.malloc(N);
  f.init(N);
  for(i=(0);i<(N);i++){
    x[i+1] = x[i] + f.range(A[i]+1, N-1);
    f.add(A[i], 1);
  }
  f.init(N);
  for(i=(0);i<(N);i++){
    y[i+1] = y[i] + f.get(A[N-1-i]-1);
    f.add(A[N-1-i], 1);
  }
  for(xr20shxY=(0);xr20shxY<(Q);xr20shxY++){
    rd(L);L += (-1);
    rd(R);R += (-1);
    res = 4611686016279904256LL;
    for(i=(0);i<(N+1);i++){
      h1[i] = h2[i] = 0;
    }
    for(i=(0);i<(L);i++){
      h1[A[i]+1]++;
    }
    for(i=(R+1);i<(N);i++){
      h2[A[i]+1]++;
    }
    for(i=(0);i<(N);i++){
      h1[i+1] += h1[i];
    }
    for(i=(0);i<(N);i++){
      h2[i+1] += h2[i];
    }
    for(i=(0);i<(N);i++){
      chmin(res, ((L-h1[i+1]) + h2[i]) * (R - L + 1));
    }
    for(i=(0);i<(L);i++){
      res += h2[A[i]];
    }
    res += x[L] + y[N-R-1];
    wt_L(res);
    wt_L('\n');
  }
  return 0;
}
template<class T> void fenwick<T>::malloc(int mem){
  memory = mem;
  data = (T*)std::malloc(sizeof(T)*mem);
}
template<class T> void fenwick<T>::malloc(int mem, int fg){
  memory = mem;
  data = (T*)std::malloc(sizeof(T)*mem);
  if(fg){
    init(mem);
  }
}
template<class T> void fenwick<T>::walloc(int mem, void **workMemory /* = &wmem*/){
  memory = mem;
  walloc1d(&data, mem, workMemory);
}
template<class T> void fenwick<T>::walloc(int mem, int fg, void **workMemory /* = &wmem*/){
  memory = mem;
  walloc1d(&data, mem, workMemory);
  if(fg){
    init(mem);
  }
}
template<class T> void fenwick<T>::free(void){
  memory = 0;
  free(data);
}
template<class T> void fenwick<T>::init(int N){
  size = N;
  memset(data,0,sizeof(T)*N);
}
template<class T> void fenwick<T>::add(int k, T val){
  while(k < size){
    data[k] += val;
    k |= k+1;
  }
}
template<class T> T fenwick<T>::get(int k){
  T res = 0;
  while(k>=0){
    res += data[k];
    k = (k&(k+1))-1;
  }
  return res;
}
template<class T> T fenwick<T>::range(int a, int b){
  if(b==-1){
    b=size-1;
  }
  return get(b) - get(a-1);
}
template<class T> int fenwick<T>::kth(T k){
  int i=0;
  int j=size;
  int c;
  T v;
  while(i<j){
    c = (i+j)/2;
    v = get(c);
    if(v <= k){
      i=c+1;
    }
    else{
      j=c;
    }
  }
  return i==size?-1:i;
}
// cLay varsion 20201018-2

// --- original code ---
// int N, Q, A[2d4], L, R;
// ll x[2d4+1], y[2d4+1];
// ll h1[2d4+1], h2[2d4+1];
// fenwick<int> f;
// {
//   ll res;
//   rd(N,Q,(A--)(N));
//   f.malloc(N);
// 
//   f.init(N);
//   rep(i,N) x[i+1] = x[i] + f.range(A[i]+1, N-1), f.add(A[i], 1);
//   f.init(N);
//   rep(i,N) y[i+1] = y[i] + f.get(A[N-1-i]-1), f.add(A[N-1-i], 1);
// 
//   rep(Q){
//     rd(L--, R--);
//     res = ll_inf;
//     rep(i,N+1) h1[i] = h2[i] = 0;
//     rep(i,L) h1[A[i]+1]++;
//     rep(i,R+1,N) h2[A[i]+1]++;
//     rep(i,N) h1[i+1] += h1[i];
//     rep(i,N) h2[i+1] += h2[i];
//     rep(i,N) res <?= ((L-h1[i+1]) + h2[i]) * (R - L + 1);
//     rep(i,L) res += h2[A[i]];
//     res += x[L] + y[N-R-1];
//     wt(res);
//   }
// }