Write a program to find the nth super ugly number.

Super ugly numbers are positive numbers whose all prime factors are in the given prime list primes of size k.

Example:

Input: n = 12, primes = [2,7,13,19]
Output: 32
Explanation: [1,2,4,7,8,13,14,16,19,26,28,32] is the sequence of the first 12
super ugly numbers given primes = [2,7,13,19] of size 4.

Note:

  • 1 is a super ugly number for any given primes.
  • The given numbers in primes are in ascending order.
  • 0 < k ≤ 100, 0 < n ≤ 106, 0 < primes[i] < 1000.
  • The nth super ugly number is guaranteed to fit in a 32-bit signed integer.

264. Ugly Number II 的拓展,还是找出第n个丑陋数,但质数集合不在只是2,3,5,而是可以任意给定。难度增加了,但本质上和Ugly Number II 没有什么区别,由于不知道质数的个数,可以用一个idx数组来保存当前的位置,然后从每个子链中取出一个数,找出其中最小值,然后更新idx数组对应位置,注意有可能最小值不止一个,要更新所有最小值的位置。

解题思路:
要使得super ugly number不漏掉,那么需要使用每个因子去乘以其对应的“第一个”丑数。那么何为对应的“第一个”丑数?

首先,利用ugly[]数组来保存所有的超级丑数,ugly[i]表示第i+1个超级丑数;

接着利用pointer[]数组来表示每个因子对应的“第一个”丑数的下标。pointer数组长度当然需要和primes长度一致,且初始化为0,代表着每个因子对应的“第一个”丑数都是ugly[0];

接下来我们以primes[2,7,13,19],pointer[0,0,0,0],ugly[0]=1作为初始条件往下看:

遍历primes数组,用每个因子都乘以其对应的第一个丑数,即ugly[0]=1,可以发现1x2=2是最小值,故ugly[1]=2;但要注意,此时的pointer数组发生了变化:

由于当前产生的丑数2是由2这个因子乘以它的对应“第一个”丑数得到的,因此需要将pointer[0]加一。pointer[0]是2这个因子对应的“第一个”丑数的下标,因为当前已经使用了2x1,如果不更新,则下一轮还是会用2这个因子去乘以第一个丑数(ugly[0]).将其更新后,则意味着2这个因子对应的第一个丑数已经改变了,变成了ugly[1].而其他三个对应的“第一个”丑数还是ugly[0]。

我们接着看下一轮:2x2【即ugly[pointer[1]]x2】,1x7,1x13,1x19,发现还是2这个因子得到的数最小,故更新:ugly[2]=2x2=4,pointer[0]=2;

下一轮:4x2,1x7,1x13,1x19,可以发现当前这一轮最小值是7,且由因子7产生,故更新:ugly[3]=7,pointer[1]=1;

以此类推....
如果更新过程中,出现最小值不止一个的话,则其对应的pointer的值都需要增加1。

Java:

public int nthSuperUglyNumber(int n, int[] primes) {
int[] ugly = new int[n+1];
ugly[0]=1;
int[] pointer = new int[primes.length];
for(int i=1;i<n;i++) {
int min=Integer.MAX_VALUE;
int minIndex = 0;
for(int j=0;j<primes.length;j++) {
if(ugly[pointer[j]]*primes[j]<min) {
min=ugly[pointer[j]]*primes[j];
minIndex = j;
}else if(ugly[pointer[j]]*primes[j]==min) {
pointer[j]++;
}
}
ugly[i]=min;
pointer[minIndex]++;
}
return ugly[n-1];
}

Java:1

public int nthSuperUglyNumberI(int n, int[] primes) {
int[] ugly = new int[n];
int[] idx = new int[primes.length]; ugly[0] = 1;
for (int i = 1; i < n; i++) {
//find next
ugly[i] = Integer.MAX_VALUE;
for (int j = 0; j < primes.length; j++)
ugly[i] = Math.min(ugly[i], primes[j] * ugly[idx[j]]); //slip duplicate
for (int j = 0; j < primes.length; j++) {
while (primes[j] * ugly[idx[j]] <= ugly[i]) idx[j]++;
}
} return ugly[n - 1];
}

Java:2

public int nthSuperUglyNumber(int n, int[] primes) {
int[] ugly = new int[n];
int[] idx = new int[primes.length];
int[] val = new int[primes.length];
Arrays.fill(val, 1); int next = 1;
for (int i = 0; i < n; i++) {
ugly[i] = next; next = Integer.MAX_VALUE;
for (int j = 0; j < primes.length; j++) {
//skip duplicate and avoid extra multiplication
if (val[j] == ugly[i]) val[j] = ugly[idx[j]++] * primes[j];
//find next ugly number
next = Math.min(next, val[j]);
}
} return ugly[n - 1];
}

Java: 3 index heap 

public int nthSuperUglyNumberHeap(int n, int[] primes) {
int[] ugly = new int[n]; PriorityQueue<Num> pq = new PriorityQueue<>();
for (int i = 0; i < primes.length; i++) pq.add(new Num(primes[i], 1, primes[i]));
ugly[0] = 1; for (int i = 1; i < n; i++) {
ugly[i] = pq.peek().val;
while (pq.peek().val == ugly[i]) {
Num nxt = pq.poll();
pq.add(new Num(nxt.p * ugly[nxt.idx], nxt.idx + 1, nxt.p));
}
} return ugly[n - 1];
} private class Num implements Comparable<Num> {
int val;
int idx;
int p; public Num(int val, int idx, int p) {
this.val = val;
this.idx = idx;
this.p = p;
} @Override
public int compareTo(Num that) {
return this.val - that.val;
}
} 

Python:

def nthSuperUglyNumber(self, n, primes):
ugly = [1]
pointers = [0]*len(primes) for i in range(1,n):
minu = float("inf")
minIndex = 0
for j in range(len(primes)):
if primes[j] * ugly[pointers[j]] < minu:
minu = primes[j] * ugly[pointers[j]]
minIndex = j
elif primes[j] * ugly[pointers[j]] == minu:
pointers[j] += 1
ugly.append(minu)
pointers[minIndex] += 1
return ugly[-1]  

Python:

# Heap solution. (620ms)
class Solution(object):
def nthSuperUglyNumber(self, n, primes):
"""
:type n: int
:type primes: List[int]
:rtype: int
"""
heap, uglies, idx, ugly_by_last_prime = [], [0] * n, [0] * len(primes), [0] * n
uglies[0] = 1 for k, p in enumerate(primes):
heapq.heappush(heap, (p, k)) for i in xrange(1, n):
uglies[i], k = heapq.heappop(heap)
ugly_by_last_prime[i] = k
idx[k] += 1
while ugly_by_last_prime[idx[k]] > k:
idx[k] += 1
heapq.heappush(heap, (primes[k] * uglies[idx[k]], k)) return uglies[-1]

Python:

# Time:  O(n * k)
# Space: O(n + k)
# Hash solution. (932ms)
class Solution2(object):
def nthSuperUglyNumber(self, n, primes):
"""
:type n: int
:type primes: List[int]
:rtype: int
"""
uglies, idx, heap, ugly_set = [0] * n, [0] * len(primes), [], set([1])
uglies[0] = 1 for k, p in enumerate(primes):
heapq.heappush(heap, (p, k))
ugly_set.add(p) for i in xrange(1, n):
uglies[i], k = heapq.heappop(heap)
while (primes[k] * uglies[idx[k]]) in ugly_set:
idx[k] += 1
heapq.heappush(heap, (primes[k] * uglies[idx[k]], k))
ugly_set.add(primes[k] * uglies[idx[k]]) return uglies[-1]

Python:  

# Time:  O(n * logk) ~ O(n * klogk)
# Space: O(n + k)
class Solution3(object):
def nthSuperUglyNumber(self, n, primes):
"""
:type n: int
:type primes: List[int]
:rtype: int
"""
uglies, idx, heap = [1], [0] * len(primes), []
for k, p in enumerate(primes):
heapq.heappush(heap, (p, k)) for i in xrange(1, n):
min_val, k = heap[0]
uglies += [min_val] while heap[0][0] == min_val: # worst time: O(klogk)
min_val, k = heapq.heappop(heap)
idx[k] += 1
heapq.heappush(heap, (primes[k] * uglies[idx[k]], k)) return uglies[-1]    

C++:

class Solution {
public:
int nthSuperUglyNumber(int n, vector<int>& primes) {
vector<int> res(1, 1), idx(primes.size(), 0);
while (res.size() < n) {
vector<int> tmp;
int mn = INT_MAX;
for (int i = 0; i < primes.size(); ++i) {
tmp.push_back(res[idx[i]] * primes[i]);
}
for (int i = 0; i < primes.size(); ++i) {
mn = min(mn, tmp[i]);
}
for (int i = 0; i < primes.size(); ++i) {
if (mn == tmp[i]) ++idx[i];
}
res.push_back(mn);
}
return res.back();
}
};

C++:  

class Solution {
public:
int nthSuperUglyNumber(int n, vector<int>& primes) {
vector<int> dp(n, 1), idx(primes.size(), 0);
for (int i = 1; i < n; ++i) {
dp[i] = INT_MAX;
for (int j = 0; j < primes.size(); ++j) {
dp[i] = min(dp[i], dp[idx[j]] * primes[j]);
}
for (int j = 0; j < primes.size(); ++j) {
if (dp[i] == dp[idx[j]] * primes[j]) {
++idx[j];
}
}
}
return dp.back();
}
};

  

类似题目:

[LeetCode] 263. Ugly Number 丑陋数

[LeetCode] 264. Ugly Number II 丑陋数 II

All LeetCode Questions List 题目汇总

[LeetCode] 313. Super Ugly Number 超级丑陋数的更多相关文章

  1. [LeetCode]313. Super Ugly Number超级丑数,丑数系列看这一道就行了

    丑数系列的题看这一道就可以了 /* 和ugly number2差不多,不过这次的质因子多了,所以用数组来表示质因子的target坐标 target坐标指的是这个质因子此次要乘的前任丑数是谁 */ pu ...

  2. [LeetCode] Super Ugly Number 超级丑陋数

    Write a program to find the nth super ugly number. Super ugly numbers are positive numbers whose all ...

  3. [LintCode] Super Ugly Number 超级丑陋数

    Write a program to find the nth super ugly number. Super ugly numbers are positive numbers whose all ...

  4. 313 Super Ugly Number 超级丑数

    编写一段程序来寻找第 n 个超级丑数.超级丑数是指其所有质因数都在长度为k的质数列表primes中的正整数.例如,[1, 2, 4, 7, 8, 13, 14, 16, 19, 26, 28, 32] ...

  5. Leetcode 313. super ugly number

    Write a program to find the nth super ugly number. Super ugly numbers are positive numbers whose all ...

  6. [LeetCode] 264. Ugly Number II 丑陋数 II

    Write a program to find the n-th ugly number. Ugly numbers are positive numbers whose prime factors ...

  7. leetcode 263. Ugly Number 、264. Ugly Number II 、313. Super Ugly Number 、204. Count Primes

    263. Ugly Number 注意:1.小于等于0都不属于丑数 2.while循环的判断不是num >= 0, 而是能被2 .3.5整除,即能被整除才去除这些数 class Solution ...

  8. [LeetCode] Ugly Number II 丑陋数之二

    Write a program to find the n-th ugly number. Ugly numbers are positive numbers whose prime factors ...

  9. [LeetCode] 264. Ugly Number II 丑陋数之二

    Write a program to find the n-th ugly number. Ugly numbers are positive numbers whose prime factors ...

随机推荐

  1. ARTS-week4

    Algorithm 将两个有序链表合并为一个新的有序链表并返回.新链表是通过拼接给定的两个链表的所有节点组成的.示例:输入:1->2->4, 1->3->4输出:1->1 ...

  2. qt事件机制(转)

    学习了一段时间的Qt之后,发现Qt的事件机制和其他语言的机制有些不同.Qt除了能够通过信号和槽机制来实现一些Action动作之外,还可以用对象所带的事件,或者用户自定义的事件来实现对象的一些行为处理. ...

  3. Jupyter Notebook(持续更新)

    1.引用Pandas import pandas as pd 2.创建DataFrame bb=pd.DataFrame(enron_data) 3.查看列 & 行 dataFrame.sha ...

  4. 启用Microsoft loopback Adapte

    开始▶控制面板▶系统   系统▶设备管理器   此时,点击操作的菜单是没有有用子菜单的,需要点击一下网络适配器.   再点击操作▶添加过时硬件   添加硬件向导▶下一步   安装我手动从列表选择的硬件 ...

  5. 51nod1463 找朋友

    [传送门] 写的时候一直没有想到离线解法,反而想到两个比较有趣的解法.一是分块,$f[i][j]$表示第$i$块块首元素到第$j$个元素之间满足条件的最大值(即对$B_l + B_r \in K$的$ ...

  6. WinDbg常用命令系列---检查符号X

    x (Examine Symbols) x命令在所有与指定模式匹配的上下文中显示符号. x [Options] Module!Symbol x [Options] * 参数: Options特定符号搜 ...

  7. Huawei Honorcup Marathon 2 垫底记

    先放链接: Huawei Honorcup Marathon 2 在 CF 上面看到这个比赛,就去玩了一下. 一开始的做法时先选一个块,然后不断看其它的每一个块拼在哪里的误差最小,然后拼上去.误差函数 ...

  8. codevs 2780 ZZWYYQWZHZ

    2780 ZZWYYQWZHZ  时间限制: 1 s  空间限制: 32000 KB  题目等级: 青铜 Bronze       题目描述 Description 可爱的小管在玩吹泡泡.忽然,他想到 ...

  9. 原创:Spark中GraphX图运算pregel详解

    由于本人文字表达能力不足,还是多多以代码形式表述,首先展示测试代码,然后解释: package com.txq.spark.test import org.apache.spark.graphx.ut ...

  10. 图解CRM(客户关系管理)全流程

    https://blog.csdn.net/lylmwt/article/details/84921432