#include <bits/stdc++.h>
//#pragma GCC optimize("O3")
#define int long long
#define fi first
#define se second
#define pb push_back
#define lb lower_bound
using namespace std;
const int MAXN = 1e5+10;
const int SQRT = 700;
const int MOD = 998244353;
const int LOG = 21;
const int INF = 1e13+10;
typedef pair<int,int> pii;
typedef pair<pii,int> ipii;
int n;
vector <int> adj[MAXN];
int dp[MAXN][2][2];
int b[MAXN];
void dfs(int nw, int par){
if(par!=-1 && adj[nw].size() == 1){
if(b[nw]){
dp[nw][0][1] = 0;
dp[nw][1][0] = 1;
dp[nw][1][1] = INF;
dp[nw][0][0] = INF;
} else {
dp[nw][0][0] = 0;
dp[nw][1][1] = 1;
dp[nw][1][0] = INF;
dp[nw][0][1] = INF;
}
return;
}
vector <int> vec[2][2];
for(auto nx : adj[nw]){
if(nx == par) continue;
dfs(nx, nw); //build dp bawah
}
//klik 1, nyala 1
for(auto nx : adj[nw]){
if(nx == par) continue;
dp[nw][1][1] += dp[nx][0][1];
vec[1][1].pb(dp[nx][1][1]-dp[nx][0][1]); //switch gk klik nyala -> klik nyala
}
sort(vec[1][1].begin(), vec[1][1].end());
int sum = 0, mx = 0, mx2 = 0, cnt = 0;
for(int i=0; i<vec[1][1].size(); i++){
int in = vec[1][1][i];
if(cnt==0){
sum += in; mx2 = mx; mx = in; cnt++; continue;
}
sum += in; mx2 = mx; mx = in; cnt++;
if(in > 0) break;
}
int te = sum;
if(vec[1][1].size()>=2) te = min(sum, sum-mx-mx2);
if(b[nw] == 1){ //odd->odd klik nyala, optimal
if(cnt%2==1) dp[nw][1][1] += te;
else dp[nw][1][1] += (sum-mx);
} else {
if(cnt%2==1) dp[nw][1][1] += (sum-mx);
else dp[nw][1][1] += te;
}
dp[nw][1][1]++;
//klik 1, nyala 0
for(auto nx : adj[nw]){
if(nx == par) continue;
dp[nw][1][0] += dp[nx][0][1];
vec[1][0].pb(dp[nx][1][1]-dp[nx][0][1]); //switch gk klik nyala -> klik nyala
}
sort(vec[1][0].begin(), vec[1][0].end());
sum = 0, mx = 0, cnt = 0;
for(int i=0; i<vec[1][0].size(); i++){
int in = vec[1][0][i];
if(cnt==0){
sum += in; mx2 = mx; mx = in; cnt++; continue;
}
sum += in; mx2 = mx; mx = in; cnt++;
if(in > 0) break;
}
te = sum;
if(vec[1][0].size()>=2) te = min(sum, sum-mx-mx2);
if(b[nw] == 1){ //odd->even klik nyala
if(cnt%2==1) dp[nw][1][0] += (sum-mx);
else dp[nw][1][0] += te;
} else {
if(cnt%2==1) dp[nw][1][0] += te;
else dp[nw][1][0] += (sum-mx);
}
dp[nw][1][0]++;
//klik 0, nyala 0
for(auto nx : adj[nw]){
if(nx == par) continue;
dp[nw][0][0] += dp[nx][0][0];
vec[0][0].pb(dp[nx][1][0]-dp[nx][0][0]); //switch gk klik mati -> klik mati
}
sort(vec[0][0].begin(), vec[0][0].end());
sum = 0, mx = 0, cnt = 0;
for(int i=0; i<vec[0][0].size(); i++){
int in = vec[0][0][i];
if(cnt==0){
sum += in; mx2 = mx; mx = in; cnt++; continue;
}
sum += in; mx2 = mx; mx = in; cnt++;
}
te = sum;
if(vec[0][0].size()>=2) te = min(sum, sum-mx-mx2);
if(b[nw] == 1){ //odd->odd klik nyala
if(cnt%2==1) dp[nw][0][0] += te;
else dp[nw][0][0] += (sum-mx);
} else {
if(cnt%2==1) dp[nw][0][0] += (sum-mx);
else dp[nw][0][0] += te;
}
//klik 0, nyala 1
for(auto nx : adj[nw]){
if(nx == par) continue;
dp[nw][0][1] += dp[nx][0][0];
vec[0][1].pb(dp[nx][1][0]-dp[nx][0][0]); //switch gk klik mati -> klik mati
}
sort(vec[0][1].begin(), vec[0][1].end());
sum = 0, mx = 0, cnt = 0;
for(int i=0; i<vec[0][1].size(); i++){
int in = vec[0][1][i];
if(cnt==0){
sum += in; mx2 = mx; mx = in; cnt++; continue;
}
sum += in; mx2 = mx; mx = in; cnt++;
}
te = sum;
if(vec[0][1].size()>=2) te = min(sum, sum-mx-mx2);
if(b[nw] == 1){ //odd->even klik nyala
if(cnt%2==1) dp[nw][0][1] += (sum-mx);
else dp[nw][0][1] += te;
} else {
if(cnt%2==1) dp[nw][0][1] += te;
else dp[nw][0][1] += (sum-mx);
}
}
signed main(){
ios_base::sync_with_stdio(0);cin.tie(0);cout.tie(0);
cin >> n;
for(int i=1; i<=n-1; i++){
int x, y; cin >> x >> y;
adj[x].pb(y); adj[y].pb(x);
}
for(int i=1; i<=n; i++) cin >> b[i];
dfs(1, -1);
int ans = min(dp[1][1][0], dp[1][0][0]);
if(ans>=n) cout << "impossible\n";
else cout << ans << '\n';
// for(int i=1; i<=n; i++){
// cout << i << ' '<< dp[i][1][1] << ' ' << dp[i][1][0] << ' ' << dp[i][0][1] << ' ' << dp[i][0][0] << '\n';
// }
}
