#include <bits/stdc++.h>
#pragma GCC optimize("O3")
#pragma GCC optimize("Ofast")
#pragma GCC optimize("unroll-loops")
#define int long long
using namespace std;
const int mod = (int)1e6 + 3;
signed main(){
ios_base::sync_with_stdio(false);
cin.tie(0);
int n;
cin >> n;
vector<int> a(n), srt(n);
for(int i = 0; i < n; ++i){
cin >> a[i];
srt[i] = a[i];
}
sort(srt.begin(), srt.end());
srt.erase(unique(srt.begin(), srt.end()), srt.end());
int ans = 0;
for(int i = 0; i + 1 < (int)srt.size(); ++i){
vector<int> lmn(n, -1), rmn(n, -1);
set<int> st;
for(int j = 0; j < n; ++j){
auto p = st.lower_bound(a[j] + 1);
if(p != st.end()){
lmn[j] = *p;
}
st.insert(a[j]);
}
st.clear();
for(int j = n - 1; j >= 0; --j){
auto p = st.lower_bound(a[j] + 1);
if(p != st.end()){
rmn[j] = *p;
}
st.insert(a[j]);
}
int cnt = 0;
int mn = -1, mx = n;
for(int j = 0; j < n; ++j){
if(lmn[j] != -1 && rmn[j] != -1 && a[j] == srt[i]){
if(mn == -1) mn = j;
mx = j;
++cnt;
a[j] = srt[i + 1];
}
}
if(cnt){
if(cnt == 1){
(ans += (lmn[mn] + rmn[mn]) * (srt[i + 1] - srt[i]) % mod) %= mod;
assert(ans >= 0 && ans < mod);}
else{
(ans += (lmn[mn] + rmn[mx] + min(rmn[mn], lmn[mx])) * (srt[i + 1] - srt[i]) % mod + (srt[i] + srt[i + 1] + 1) * (srt[i + 1] - srt[i]) / 2 % mod * (cnt * 2 - 3) % mod) %= mod;
assert(ans >= 0 && ans < mod); }
assert(ans >= 0 && ans < mod);
(ans += cnt * ((srt[i] + srt[i + 1] - 1) * (srt[i + 1] - srt[i]) / 2 % mod) % mod) %= mod;
// cout << fir << ' ' << lst << ' ' << cnt << ' ' << ans << endl;
}
assert(ans >= 0 && ans < mod);
}
cout << ans << '\n';
return 0;
}
