#ifndef CLASS_MODINT
#define CLASS_MODINT

#include <cstdint>

template <std::uint32_t mod>
class modint {
private:
	std::uint32_t n;
public:
	modint() : n(0) {};
	modint(std::int64_t n_) : n((n_ >= 0 ? n_ : mod - (-n_) % mod) % mod) {};
	static constexpr std::uint32_t get_mod() { return mod; }
	std::uint32_t get() const { return n; }
	bool operator==(const modint& m) const { return n == m.n; }
	bool operator!=(const modint& m) const { return n != m.n; }
	modint& operator+=(const modint& m) { n += m.n; n = (n < mod ? n : n - mod); return *this; }
	modint& operator-=(const modint& m) { n += mod - m.n; n = (n < mod ? n : n - mod); return *this; }
	modint& operator*=(const modint& m) { n = std::uint64_t(n) * m.n % mod; return *this; }
	modint operator+(const modint& m) const { return modint(*this) += m; }
	modint operator-(const modint& m) const { return modint(*this) -= m; }
	modint operator*(const modint& m) const { return modint(*this) *= m; }
	modint inv() const { return (*this).pow(mod - 2); }
	modint pow(std::uint64_t b) const {
		modint ans = 1, m = modint(*this);
		while (b) {
			if (b & 1) ans *= m;
			m *= m;
			b >>= 1;
		}
		return ans;
	}
};

#endif // CLASS_MODINT

#include <vector>
#include <iostream>
#include <algorithm>
#include <functional>
using namespace std;
using mint = modint<998244353>;
long long inversion(int N, vector<int> A) {
	vector<int> perm(N);
	for (int i = 0; i < N; ++i) {
		perm[i] = i;
	}
	sort(perm.begin(), perm.end(), [&](int i, int j) { return A[i] != A[j] ? A[i] > A[j] : i > j; });
	long long ans = 0;
	vector<int> bit(N + 1);
	for (int i : perm) {
		for (int j = i; j >= 1; j -= j & (-j)) {
			ans += bit[j];
		}
		for (int j = i + 1; j <= N; j += j & (-j)) {
			++bit[j];
		}
	}
	return ans;
}
int main() {
	int N;
	cin >> N;
	vector<int> A(N);
	for (int i = 0; i < N; ++i) {
		cin >> A[i];
	}
	vector<mint> fact(N + 1), factinv(N + 1);
	fact[0] = 1;
	for (int i = 1; i <= N; ++i) fact[i] = fact[i - 1] * i;
	factinv[N] = fact[N].inv();
	for (int i = N; i >= 1; --i) factinv[i - 1] = factinv[i] * i;
	function<mint(int, int)> comb = [&](int x, int y) {
		if (x < y || y < 0) return mint(0);
		return fact[x] * factinv[y] * factinv[x - y];
	};
	long long p1 = inversion(N, A);
	vector<int> B = A; B.insert(B.end(), A.begin(), A.end());
	long long p2 = inversion(2 * N, B) - p1 * 2;
	mint ans = 1;
	mint coeff = 0; long long sum = 0;
	for (int i = 1; i <= N; ++i) {
		coeff += mint(p1) * mint(1LL * i * (i - 1) / 2) * mint(1LL * N * (N - 1) / 2).inv();
		coeff += mint(p2) * mint(1LL * sum * i) * mint(1LL * N * N).inv();
		sum += i;
		ans *= comb(N, i);
	}
	ans *= coeff;
	cout << ans.get() << endl;
	return 0;
}