高斯消去

#pragma GCC optimzize("Ofast,no-stack-protector")
#include<bits/stdc++.h>
#define int long long
#define quick ios::sync_with_stdio(0);cin.tie(0);
#define rep(x,a,b) for(int x=a;x<=b;x++)
#define repd(x,a,b) for(int x=a;x>=b;x--)
#define lowbit(x) (x&-x)
#define sz(x) (int)(x.size())
#define F first
#define S second
#define all(x) x.begin(),x.end()
#define mp make_pair
#define eb emplace_back
using namespace std;
typedef pair<int,int> pii;
void debug(){
    cout<<"\n";
}
template <class T,class ... U >
void debug(T a, U ... b){
    cout<<a<<" ",debug(b...);
}
const int N=1e3+7;
const int INF=1e18;
vector<int> v[N];
int M;
int qpow(int a,int b){
    int ret=1;
    while(b>0){
        if(b&1) ret=ret*a%M;
        a=a*a%M;
        b>>=1;
    }
    return ret;
}
int inv(int x){
    return qpow(x,M-2);
}
signed main(){
    quick
    int n;
    cin>>n>>M;
    vector<int> st(n);
    rep(i,0,n-1) cin>>st[i];
    rep(i,0,n-1){
        rep(j,0,n-1){
            int x;
            cin>>x;
            v[j].eb(x);
        }
    }
    rep(i,0,n-1){
        v[i].eb(st[i]);
    }

    rep(i,0,n-1){
        if(!v[i][i]){
            rep(j,i+1,n-1){
                if(v[j][i]){
                    swap(v[j],v[i]);
                    break;
                }
            }
        }
        rep(j,i+1,n-1){
            int mul=(M-v[j][i]%M*inv(v[i][i])%M)%M;
            rep(x,i,n){
                v[j][x]=(v[j][x]+mul*v[i][x])%M;
            }
        }
    }
    vector<int> ans(n);
    repd(i,n-1,0){
        int s=v[i][n];
        rep(j,i+1,n-1){
            s=s-v[i][j]*ans[j]%M;
            s=(s+M)%M;
        }
        ans[i]=s*inv(v[i][i])%M;
    }
    rep(i,0,n-1){
        cout<<ans[i]<<" \n"[i==n-1];
    }

    return 0;
}