// -*- coding:utf-8-unix -*-
// #![feature(map_first_last)]
#![allow(dead_code)]
#![allow(unused_imports)]
#![allow(unused_macros)]
// use core::num;
use std::cmp::*;
use std::fmt::*;
use std::hash::*;
use std::io::BufRead;
use std::iter::FromIterator;
use std::*;
use std::{cmp, collections, fmt, io, iter, ops, str};
const INF: i64 = 1223372036854775807;
const UINF: usize = INF as usize;
const LINF: i64 = 2147483647;
const INF128: i128 = 1223372036854775807000000000000;
const MOD1: i64 = 1000000007;
const MOD9: i64 = 998244353;
const MOD: i64 = MOD9;
// const MOD: i64 = MOD2;
const UMOD: usize = MOD as usize;
const M_PI: f64 = 3.14159265358979323846;

// use proconio::input;
// const MOD: i64 = INF;

use cmp::Ordering::*;
use std::collections::*;

use std::io::stdin;
use std::io::stdout;
use std::io::Write;

macro_rules! p {
    ($x:expr) => {
        //if expr
        println!("{}", $x);
    };
}

macro_rules! vp {
    // vector print separate with space
    ($x:expr) => {
        println!(
            "{}",
            $x.iter()
                .map(|x| x.to_string())
                .collect::<Vec<_>>()
                .join(" ")
        );
    };
}

macro_rules! d {
    ($x:expr) => {
        eprintln!("{:?}", $x);
    };
}
macro_rules! yn {
    ($val:expr) => {
        if $val {
            println!("Yes");
        } else {
            println!("No");
        }
    };
}

macro_rules! map{
    // declear btreemap
    ($($key:expr => $val:expr),*) => {
        {
            let mut map = ::std::collections::BTreeMap::new();
            $(
                map.insert($key, $val);
            )*
            map
        }
    };
}

macro_rules! set{
    // declear btreemap
    ($($key:expr),*) => {
        {
            let mut set = ::std::collections::BTreeSet::new();
            $(
                set.insert($key);
            )*
            set
        }
    };
}

//input output
#[allow(dead_code)]
fn read<T: std::str::FromStr>() -> T {
    let mut s = String::new();
    std::io::stdin().read_line(&mut s).ok();
    s.trim().parse().ok().unwrap()
}

#[allow(dead_code)]
fn read_vec<T: std::str::FromStr>() -> Vec<T> {
    read::<String>()
        .split_whitespace()
        .map(|e| e.parse().ok().unwrap())
        .collect()
}

#[allow(dead_code)]
fn read_mat<T: std::str::FromStr>(n: u32) -> Vec<Vec<T>> {
    (0..n).map(|_| read_vec()).collect()
}

#[allow(dead_code)]
fn readii() -> (i64, i64) {
    let mut vec: Vec<i64> = read_vec();
    (vec[0], vec[1])
}

#[allow(dead_code)]
fn readiii() -> (i64, i64, i64) {
    let mut vec: Vec<i64> = read_vec();
    (vec[0], vec[1], vec[2])
}
#[allow(dead_code)]
fn readuu() -> (usize, usize) {
    let mut vec: Vec<usize> = read_vec();
    (vec[0], vec[1])
}

#[allow(dead_code)]
fn readff() -> (f64, f64) {
    let mut vec: Vec<f64> = read_vec();
    (vec[0], vec[1])
}

fn readcc() -> (char, char) {
    let mut vec: Vec<char> = read_vec();
    (vec[0], vec[1])
}

fn readuuu() -> (usize, usize, usize) {
    let mut vec: Vec<usize> = read_vec();
    (vec[0], vec[1], vec[2])
}
#[allow(dead_code)]
fn readiiii() -> (i64, i64, i64, i64) {
    let mut vec: Vec<i64> = read_vec();
    (vec[0], vec[1], vec[2], vec[3])
}

#[allow(dead_code)]
fn readuuuu() -> (usize, usize, usize, usize) {
    let mut vec: Vec<usize> = read_vec();
    (vec[0], vec[1], vec[2], vec[3])
}

use std::collections::VecDeque;

macro_rules! M {
    (a :expr ) => {
        M::new({ a })
    };
}
#[derive(Copy, Clone, Debug)]
pub struct M(i64);
impl M {
    fn new(x: i64) -> Self {
        M(x.rem_euclid(MOD))
    }
    fn pow(self, n: usize) -> Self {
        match n {
            0 => M::new(1),
            _ => {
                let mut a = self.pow(n >> 1);
                a *= a;
                if n & 1 == 1 {
                    a *= self;
                }
                a
            }
        }
    }
    fn inv(self) -> Self {
        self.pow((MOD - 2) as usize)
    }
}
impl std::ops::Neg for M {
    type Output = M;
    fn neg(self) -> Self::Output {
        Self::new(-self.0)
    }
}
impl std::ops::AddAssign<M> for M {
    fn add_assign(&mut self, rhs: Self) {
        self.0 += rhs.0;
        self.0 %= MOD;
    }
}
impl std::ops::AddAssign<i64> for M {
    fn add_assign(&mut self, rhs: i64) {
        *self += M::new(rhs);
    }
}
impl std::ops::AddAssign<usize> for M {
    fn add_assign(&mut self, rhs: usize) {
        *self += M::new(rhs as i64);
    }
}
impl<T> std::ops::Add<T> for M
where
    M: std::ops::AddAssign<T>,
{
    type Output = Self;
    fn add(self, other: T) -> Self {
        let mut res = self;
        res += other;
        res
    }
}
impl std::ops::SubAssign<M> for M {
    fn sub_assign(&mut self, rhs: Self) {
        self.0 -= rhs.0;
        if self.0 < 0 {
            self.0 %= MOD;
            self.0 += MOD;
        }
    }
}
impl std::ops::SubAssign<i64> for M {
    fn sub_assign(&mut self, rhs: i64) {
        *self -= M::new(rhs);
        if (*self).0 < 0 {
            self.0 %= MOD;
            self.0 += MOD;
        }
    }
}
impl std::ops::SubAssign<usize> for M {
    fn sub_assign(&mut self, rhs: usize) {
        *self -= M::new(rhs as i64);
        if (*self).0 < 0 {
            self.0 %= MOD;
            self.0 += MOD;
        }
    }
}
impl<T> std::ops::Sub<T> for M
where
    M: std::ops::SubAssign<T>,
{
    type Output = Self;
    fn sub(self, other: T) -> Self {
        let mut res = self;
        res -= other;
        res
    }
}
impl std::ops::MulAssign<M> for M {
    fn mul_assign(&mut self, rhs: Self) {
        self.0 %= MOD;

        self.0 *= (rhs.0 % MOD);
        self.0 %= MOD;
    }
}
impl std::ops::MulAssign<i64> for M {
    fn mul_assign(&mut self, rhs: i64) {
        *self *= M::new(rhs);
    }
}
impl std::ops::MulAssign<usize> for M {
    fn mul_assign(&mut self, rhs: usize) {
        *self *= M::new(rhs as i64);
    }
}
impl<T> std::ops::Mul<T> for M
where
    M: std::ops::MulAssign<T>,
{
    type Output = Self;
    fn mul(self, other: T) -> Self {
        let mut res = self;
        res *= other;
        res
    }
}
impl std::ops::DivAssign<M> for M {
    fn div_assign(&mut self, rhs: Self) {
        *self *= rhs.inv();
    }
}
impl std::ops::DivAssign<i64> for M {
    fn div_assign(&mut self, rhs: i64) {
        *self /= M::new(rhs);
    }
}
impl std::ops::DivAssign<usize> for M {
    fn div_assign(&mut self, rhs: usize) {
        *self /= M::new(rhs as i64);
    }
}
impl<T> std::ops::Div<T> for M
where
    M: std::ops::DivAssign<T>,
{
    type Output = Self;
    fn div(self, other: T) -> Self {
        let mut res = self;
        res /= other;
        res
    }
}
impl std::fmt::Display for M {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        write!(f, "{}", self.0)
    }
}
impl std::ops::Deref for M {
    type Target = i64;
    fn deref(&self) -> &Self::Target {
        &self.0
    }
}
impl std::ops::DerefMut for M {
    fn deref_mut(&mut self) -> &mut Self::Target {
        &mut self.0
    }
}

#[allow(dead_code)]
pub fn gcd(a: usize, b: usize) -> usize {
    if b == 0 {
        a
    } else {
        gcd(b, a % b)
    }
}
#[allow(dead_code)]
pub fn lcm(a: usize, b: usize) -> usize {
    a / gcd(a, b) * b
}
#[allow(dead_code)]
/// (gcd, x, y)
pub fn extgcd(a: i64, b: i64) -> (i64, i64, i64) {
    if b == 0 {
        (a, 1, 0)
    } else {
        let (gcd, x, y) = extgcd(b, a % b);
        (gcd, y, x - (a / b) * y)
    }
}
#[allow(dead_code)]
/// x ^ n % m
pub fn mod_pow(x: usize, n: usize, m: usize) -> usize {
    let mut res = 1;
    let mut x = x % m;
    let mut n = n;
    while n > 0 {
        if n & 1 == 1 {
            res = (res * x) % m;
        }
        x = (x * x) % m;
        n >>= 1;
    }
    res
}

pub struct Combination {
    m: usize,
    f_table: Vec<usize>,
    moi: Vec<usize>,
}

impl Combination {
    // 0 <= size <= 10^8 is constrained.
    pub fn new(mod_num: usize, table_size: usize) -> Self {
        Self {
            m: mod_num,
            f_table: vec![0; table_size],
            moi: vec![0; 0],
        }
    }
    pub fn build(&mut self) {
        let size = self.f_table.len();
        self.f_table = self.fact_table(size, self.m);
        self.moi = self.fact_inv_table(size, self.m);
    }
    fn fact_table(&mut self, len: usize, m: usize) -> Vec<usize> {
        let mut res = vec![1; len + 1];
        for i in 2..len + 1 {
            res[i] = (res[i - 1] * i) % m;
        }
        res
    }

    fn fact_inv_table(&mut self, len: usize, m: usize) -> Vec<usize> {
        let mut res = vec![1; len + 1];
        let mut inv = vec![1; len + 1];
        //inv[i] = MOD - inv[MOD%i] * (MOD / i) % MOD
        for i in 2..len {
            inv[i] = (m - inv[m % i] * (m / i) % m) % m;
            res[i] = inv[i] * res[i - 1];
            // res[i] = self.mod_inverse(i, m) * res[i - 1];
            // res[i] = 1;
            res[i] %= m;
        }
        res
    }

    pub fn p(&mut self, n: usize, k: usize) -> i64 {
        let p = MOD as usize;
        if k == 0 {
            return 1;
        }
        if n < k {
            0
        } else {
            let (a1, e1) = self.mod_fact(n, p);
            let (a2, e2) = self.mod_fact(k, p);
            let (a3, e3) = self.mod_fact(n - k, p);
            if e1 > e2 + e3 {
                0
            } else {
                let moi = self.mod_inverse(a3 % p, p);
                (a1 * self.mod_inverse(a3 % p, p) % p) as i64
            }
        }
    }
    pub fn c(&mut self, n: usize, k: usize) -> i64 {
        let p = MOD as usize;
        if n == 0 && k == 0 {
            return 1;
        }
        if n == 0 {
            return 0;
        }
        if k == 0 {
            return 1;
        }
        if n < k {
            0
        } else {
            let (a1, e1) = self.mod_fact(n, p);
            let (a2, e2) = self.mod_fact(k, p);
            let (a3, e3) = self.mod_fact(n - k, p);
            if e1 > e2 + e3 {
                0
            } else {
                (((a1 * &self.moi[k]) % p * &self.moi[n - k]) % p) as i64
            }
        }
    }
    pub fn h(&mut self, n: usize, k: usize) -> i64 {
        return self.c(n + k - 1, k);
    }

    pub fn factorial(&mut self, n: usize) -> i64 {
        return self.p(n, n);
    }

    fn extgcd(&mut self, a: i64, b: i64) -> (i64, i64, i64) {
        if b == 0 {
            (a, 1, 0)
        } else {
            let (gcd, x, y) = extgcd(b, a % b);
            (gcd, y, x - (a / b) * y)
        }
    }
    fn mod_inverse(&mut self, a: usize, m: usize) -> usize {
        let (_, x, _) = self.extgcd(a as i64, m as i64);
        ((m as i64 + x) as usize % m) % m
    }
    fn mod_fact(&mut self, n: usize, p: usize) -> (usize, usize) {
        if n == 0 {
            (1, 0)
        } else {
            let (a, b) = self.mod_fact(n / p, p);
            let pow = b + n / p;
            if n / p % 2 != 0 {
                (a * (p - self.f_table[(n % p) as usize]) % p, pow)
            } else {
                (a * self.f_table[(n % p) as usize] % p, pow)
            }
        }
    }
}

fn dfs(v: usize, p: usize, graph: &Vec<Vec<(usize, usize)>>, dp: &mut Vec<Vec<M>>, a: &Vec<usize>) {
    let mut mul = M(1);
    let mut sum = M(0);
    let mut sum_pow2 = M(0);
    let mut have_child = false;
    for &(u, w) in &graph[v] {
        if u == p {
            continue;
        }
        dfs(u, v, graph, dp, a);
        mul *= dp[u][0];
        sum += dp[u][0];
        sum_pow2 += dp[u][0] * dp[u][0];
        have_child = true;
    }
    if !have_child {
        mul = M(0);
    }
    dp[v][0] = M(a[v] as i64) * (M(1) + sum);
    dp[v][1] = M(a[v] as i64) * (sum + (sum * sum - sum_pow2) / M(2));
}
fn main() {
    let n: usize = read();
    let mut a: Vec<usize> = read_vec();
    let mut graph = vec![vec![(0 as usize, 0 as usize); (0) as usize]; (n) as usize];
    for i in 0..n - 1 {
        let (mut a, mut b) = readuu();
        a -= 1;
        b -= 1;
        graph[a].push((b, 1));
        graph[b].push((a, 1));
    }
    let mut dp = vec![vec![M(0), M(0)]; n];
    dfs(0, n, &graph, &mut dp, &a);
    let mut res = M(0);
    for i in 0..n {
        res += dp[i][1];
    }
    // d!(dp);
    println!("{}", res);
}