#include <bits/stdc++.h>
// created [2019/11/29] 22:16:35
#pragma GCC diagnostic ignored "-Wsign-compare"
#pragma GCC diagnostic ignored "-Wsign-conversion"

using i32 = int32_t;
using i64 = int64_t;
using u32 = uint32_t;
using u64 = uint64_t;
using uint = unsigned int;
using usize = std::size_t;
using ll = long long;
using ull = unsigned long long;
using ld = long double;
template<typename T> constexpr T popcount(const T u) { return u ? static_cast<T>(__builtin_popcountll(static_cast<u64>(u))) : static_cast<T>(0); }
template<typename T> constexpr T log2p1(const T u) { return u ? static_cast<T>(64 - __builtin_clzll(static_cast<u64>(u))) : static_cast<T>(0); }
template<typename T> constexpr T msbp1(const T u) { return log2p1(u); }
template<typename T> constexpr T lsbp1(const T u) { return __builtin_ffsll(u); }
template<typename T> constexpr T clog(const T u) { return u ? log2p1(u - 1) : static_cast<T>(u); }
template<typename T> constexpr bool ispow2(const T u) { return u and (static_cast<u64>(u) & static_cast<u64>(u - 1)) == 0; }
template<typename T> constexpr T ceil2(const T u) { return static_cast<T>(1) << clog(u); }
template<typename T> constexpr T floor2(const T u) { return u == 0 ? static_cast<T>(0) : static_cast<T>(1) << (log2p1(u) - 1); }
template<typename T> constexpr bool btest(const T mask, const usize ind) { return static_cast<bool>((static_cast<u64>(mask) >> ind) & static_cast<u64>(1)); }
template<typename T> void bset(T& mask, const usize ind) { mask |= (static_cast<T>(1) << ind); }
template<typename T> void breset(T& mask, const usize ind) { mask &= ~(static_cast<T>(1) << ind); }
template<typename T> void bflip(T& mask, const usize ind) { mask ^= (static_cast<T>(1) << ind); }
template<typename T> void bset(T& mask, const usize ind, const bool b) { (b ? bset(mask, ind) : breset(mask, ind)); }
template<typename T> constexpr T bcut(const T mask, const usize ind) { return ind == 0 ? static_cast<T>(0) : static_cast<T>((static_cast<u64>(mask) << (64 - ind)) >> (64 - ind)); }
template<typename T> bool chmin(T& a, const T& b) { return (a > b ? a = b, true : false); }
template<typename T> bool chmax(T& a, const T& b) { return (a < b ? a = b, true : false); }
constexpr unsigned int mod                  = 1000000007;
template<typename T> constexpr T inf_v      = std::numeric_limits<T>::max() / 4;
template<typename Real> constexpr Real pi_v = Real{3.141592653589793238462643383279502884};

template<typename T>
T read()
{
    T v;
    return std::cin >> v, v;
}
template<typename T, typename... Args>
auto read(const usize size, Args... args)
{
    std::vector<decltype(read<T>(args...))> ans(size);
    for (usize i = 0; i < size; i++) { ans[i] = read<T>(args...); }
    return ans;
}
template<typename... Types>
auto reads() { return std::tuple<std::decay_t<Types>...>{read<Types>()...}; }
#    define SHOW(...) static_cast<void>(0)

template<typename T>
std::vector<T> make_v(const usize size, const T v) { return std::vector<T>(size, v); }
template<typename... Args>
auto make_v(const usize size, Args... args) { return std::vector<decltype(make_v(args...))>(size, make_v(args...)); }
template<typename T>
std::vector<T> divisors(const T n)
{
    std::vector<T> head, tail;
    for (T i = 1; i * i <= n; i++) {
        if (n % i == 0) {
            head.push_back(i);
            if (i * i != n) { tail.push_back(n / i); }
        }
    }
    for (auto it = tail.rbegin(); it != tail.rend(); it++) { head.push_back(*it); }
    return head;
}

template<usize column>
class bit_matrix
{
public:
    bit_matrix(const usize row) : row{row}, table(row) {}
    bit_matrix(const bit_matrix& m) : row{m.row}, table{m.table} {}
    bit_matrix& operator=(const bit_matrix& m)
    {
        assert(row == m.row), assert(column == m.column);
        for (usize i = 0; i < row; i++) { table[i] = m[i]; }
        return *this;
    }
    const std::bitset<column>& operator[](const usize r) const { return assert(r < row), table[r]; }
    std::bitset<column>& operator[](const usize r) { return assert(r < row), table[r]; }
    friend bit_matrix operator+(const bit_matrix& m) { return m; }
    friend bit_matrix operator^(const bit_matrix& m1, const bit_matrix& m2)
    {
        assert(m1.row == m2.row);
        bit_matrix ans(m1.row);
        for (usize i = 0; i < m1.row; i++) { ans[i] = m1.table[i] ^ m2.table[i]; }
        return ans;
    }
    template<usize col2>
    friend bit_matrix operator*(const bit_matrix& m1, const bit_matrix<col2>& m2)
    {
        assert(column == m2.row);
        bit_matrix<col2> ans(m1.row);
        for (usize i = 0; i < m1.row; i++) {
            for (usize j = 0; j < column; j++) {
                if (not m1.table[i][j]) { continue; }
                ans[i] ^= m2.table[j];
            }
        }
        return ans;
    }
    friend bit_matrix operator^(const bit_matrix& m, const unsigned long long n) { return assert(m.row == m.column), n == 0 ? bit_matrix::id(m.row) : n % 2 == 1 ? m*(m ^ (n - 1)) : ((m * m) ^ (n / 2)); }
    friend bit_matrix& operator^=(bit_matrix& m1, const bit_matrix& m2)
    {
        assert(m1.row == m2.row), assert(m1.column == m2.column);
        for (usize i = 0; i < m1.row; i++) { m1.table[i] ^= m2.table[i]; }
        return m1;
    }
    friend bit_matrix& operator*=(bit_matrix& m1, const bit_matrix& m2) { return m1 = m1 * m2; }
    friend bit_matrix& operator^=(bit_matrix& m, const unsigned long long n) { return m = m ^ n; }
    friend std::ostream& operator<<(std::ostream& os, const bit_matrix& m)
    {
        os << "[\n";
        for (usize i = 0; i < m.row; i++) { os << "[" << m[i] << "]\n"; }
        return (os << "]\n");
    }
    static bit_matrix id(const usize n)
    {
        bit_matrix ans(n);
        for (usize i = 0; i < n; i++) { ans[i].set(i); }
        return ans;
    }
    usize row;

private:
    std::vector<std::bitset<column>> table;
};
template<usize column>
bit_matrix<column> gauss_jordan(bit_matrix<column> mat)
{
    for (usize r = 0, ci = 0; ci < column; ci++) {
        const usize c = column - ci - 1;
        if (r == mat.row) { break; }
        usize piv = r;
        for (; piv < mat.row and not mat[piv][c]; piv++) {}
        if (piv == mat.row) { continue; }
        std::swap(mat[piv], mat[r]);
        for (usize j = 0; j < mat.row; j++) {
            if (j == r) { continue; }
            if (mat[j][c]) { mat[j] ^= mat[r]; }
        }
        r++;
    }
    return mat;
}
int main()
{
    const int N       = read<int>();
    const auto A      = read<int>(N);
    constexpr uint lg = 17;
    std::vector<uint> as;
    for (uint i = 0; i < N; i++) {
        for (uint a = A[i]; a <= 100000; a *= 2) { as.push_back(a); }
    }
    bit_matrix<lg> m(as.size());

    for (int r = 0; r < as.size(); r++) { m[r] = as[r]; }
    const auto base = gauss_jordan(m);
    std::vector<uint> bs;
    for (int i = 0; i < as.size(); i++) {
        if (base[i].to_ullong() == 0) { break; }
        bs.push_back(base[i].to_ulong());
    }
    //    SHOW(bs);
    const int bn = bs.size();
    std::set<int> ok;
    const int mask = 1 << bn;
    for (int m = 0; m < mask; m++) {
        uint a = 0;
        for (int i = 0; i < bn; i++) {
            if (not btest(m, i)) { continue; }
            a ^= bs[i];
        }
        if (a <= 100000) { ok.insert(a); }
    }
    //    SHOW(ok);
    std::vector<ll> S(100001, 0);
    for (int i = 0; i < N; i++) { S[A[i]] += A[i]; }
    std::vector<ll> minus(100001, 0);
    for (const int i : ok) {
        if (i == 0) { continue; }
        for (int j = i; j < 100000; j += i) {
            minus[i] += S[j] - S[j] / i;
        }
    }
    const ll sum = std::accumulate(A.begin(), A.end(), 0LL);
    std::cout << sum - *std::max_element(minus.begin(), minus.end()) << std::endl;
    return 0;
}