#include <atcoder/all>
using namespace atcoder;
#include <atcoder/internal_math>
using namespace internal;
#include <bits/stdc++.h>
using namespace std;
#include <boost/multiprecision/cpp_int.hpp>
#pragma GCC target("avx2")
#pragma GCC optimize("O3")
#pragma GCC optimize("unroll-loops")
#define rep(i, n) for (int i = 0; i < (int)(n); i++)
#define rep2(i,a,b) for (int i = (int)(a); i < (int)(b); i++)
#define all(v) v.begin(),v.end()
#define inc(x,l,r) ((l)<=(x)&&(x)<(r)) 
#define Unique(x) sort(all(x)), x.erase(unique(all(x)), x.end())
#define pcnt __builtin_popcountll
typedef long long ll;
#define int ll
using ld = long double;
using vi = vector<int>;
using vs = vector<string>;
using P = pair<int,int>;
using vp = vector<P>;
using ull = unsigned long long;
using Bint = boost::multiprecision::cpp_int;
template<typename T1,typename T2> bool chmax(T1 &a, const T2 b) {if (a < b) {a = b; return true;} else return false; }
template<typename T1,typename T2> bool chmin(T1 &a, const T2 b) {if (a > b) {a = b; return true;} else return false; }
template<typename T> using priority_queue_greater = priority_queue<T, vector<T>, greater<T>>;
template<typename T1,typename T2> ostream &operator<< (ostream &os, const pair<T1,T2> &p){os << p.first <<" "<<p.second;return os;}
ostream &operator<< (ostream &os, const modint1000000007 &m){os << m.val();return os;}
istream &operator>> (istream &is, modint1000000007 &m){ll in;is>>in;m=in;return is;}
ostream &operator<< (ostream &os, const modint998244353 &m){os << m.val();return os;}
istream &operator>> (istream &is, modint998244353 &m){ll in;is>>in;m=in;return is;}
template<typename T> istream &operator>>(istream& is,vector<T> &v){for(T &in:v)is>>in;return is;}
template<class... T> void input(T&... a){(cin>> ... >> a);}
#ifdef LOCAL
template<typename T> ostream &operator<<(ostream &os,const vector<T> &v){os<<"\x1b[32m";rep(i,v.size())os<<v[i]<<(i+1!=v.size()?" ":"");os<<"\x1b[0m";return os;}
template<class T> int print(T& a){cout << "\x1b[32m"<< a<< '\n' << "\x1b[0m";return 0;}
template<class T,class... Ts> int print(const T&a, const Ts&... b){cout << "\x1b[32m" << a;(cout<<...<<(cout<<' ',b));cout<<'\n' << "\x1b[0m";return 0;}
#else
template<typename T> ostream &operator<<(ostream &os,const vector<T> &v){rep(i,v.size())os<<v[i]<<(i+1!=v.size()?" ":"");return os;}
template<class T> int print(T& a){cout << a<< '\n';return 0;}
template<class T,class... Ts> int print(const T&a, const Ts&... b){cout << a;(cout<<...<<(cout<<' ',b));cout<<'\n';return 0;}
#endif
#define VI(v,n) vi v(n); input(v)
#define INT(...) int __VA_ARGS__; input(__VA_ARGS__)
#define STR(...) string __VA_ARGS__; input(__VA_ARGS__)
#define CHAR(...) char __VA_ARGS__; input(__VA_ARGS__)
int sign(ll x){return x>0?1:x<0?-1:0;}
ll ceil(ll x,ll y){assert(y!=0);if(sign(x)==sign(y))return (x+y-1)/y;return -((-x/y));}
ll floor(ll x,ll y){assert(y!=0);if(sign(x)==sign(y))return x/y;if(y<0)x*=-1,y*=-1;return x/y-(x%y<0);}
ll abs(ll x,ll y){return abs(x-y);}
ll bit(int n){return 1ll<<n;}
template<class T> bool ins(string s,T t){return s.find(t)!=string::npos;}
P operator+ (const P &p, const P &q){ return P{p.first+q.first,p.second+q.second};}
P operator- (const P &p, const P &q){ return P{p.first-q.first,p.second-q.second};}
int yesno(bool ok,string y="Yes",string n="No"){ cout<<(ok?y:n)<<endl;return 0;}
int YESNO(bool ok,string y="YES",string n="NO"){ cout<<(ok?y:n)<<endl;return 0;}
int di[]={0,1,0,-1,-1,-1,1,1};
int dj[]={1,0,-1,0,-1,1,-1,1};
const ll INF = 8e18;
//using mint = modint1000000007;
using mint = modint998244353;
//mint stom(const string &s,int b=10){mint res = 0;for(auto c:s)res *= b,res += c-'0';return res;}
template<typename T=int>struct edge{int to;T cost;int idx;};
template<typename T=int>class Graph{
public:
    int n;
    bool weighted,directed;
    vector<pair<int,edge<T>>> edges;
    vector<vector<edge<T>>> E;

    Graph(int n,bool weighted=false,bool directed=false):n(n),weighted(weighted),directed(directed),E(n){}
    void add_edge(int u,int v,T w=1){
        int idx = edges.size();
        E[u].push_back({v,w,idx});
        if(!directed)E[v].push_back({u,w,idx});
        edges.push_back({u,{v,w,idx}});
    }
    void read_edge(int m,int index=1){
        rep(i,m){
            int u,v;
            cin>>u>>v;
            u-=index,v-=index;
            if(weighted){
                T w;cin>>w;
                add_edge(u,v,w);
            }
            else add_edge(u,v);
        }
    }
    inline const vector<edge<T>> &operator[](int u) const { return E[u]; }
    inline vector<edge<T>> &operator[](int u) { return E[u]; }
    pair<int,edge<T>> get_edge(int i){return edges[i];}

    vector<bool> namori(){
        
        assert(n==edges.size());
        vi deg(n);
        rep(u,n)for(auto &e:E[u])deg[e.to]++;
 
        vector<bool> is_cycle(n,true);
        queue<int> Q;
        rep(u,n)if(deg[u]==1)Q.push(u);
        while(Q.size()){
            auto u = Q.front();Q.pop();
            is_cycle[u] = false;
            for(auto e:E[u]){
                if(deg[e.to]>=2){
                    deg[e.to]--;
                    if(deg[e.to]==1)Q.push(e.to);
                }
            }
        }
        return is_cycle;
    }
};
signed main() {
    cin.tie(0);
    ios_base::sync_with_stdio(false);
    cout << fixed << setprecision(20);

    INT(n,k);
    Graph G(n);
    G.read_edge(n);
    auto l = G.namori();
    int loop = 0;
    for(auto e:l)if(e)loop++;
    vector<mint> dp(2);
    dp[1] = k;
    rep(i,loop-1){
        vector<mint> pd(2);
        rep(j,2){
            pd[0] += dp[j]*(k-1-!j);
            pd[1] += dp[j]*(!j);
        }
        swap(dp,pd);
    }
    print(dp[0]*mint(k-1).pow(n-loop));
    return 0;
}