LeetCode-Notes
  • Introduction
  • Records of Practice
  • 关于Github 不支持密码问题
  • 面试题
    • 搜索广告
    • 多模态大模型
    • 刷题记录
  • 算法代码实现
  • Python
    • Python 笔记
  • Spark
    • PySpark
    • Spark Issues
    • Spark调优笔记
  • FeatureEngineering
    • Feature Cleaning
    • Feature Selection
    • Feature Transformation
    • Feature Crossing
  • Recommendation Algorithm
    • Recall-and-PreRank
      • Non-Negative Matrix Fatorization(NMF)
      • Fatorization Machine(FM)
      • User-base/Item-base实现
      • 多路召回实现
    • Ranking
      • NeuralFM
      • DeepFM
      • Deep&Cross network (DCN)
    • DeepLearning-Basic
      • Attention
      • Dropout
      • Batch Norm
  • Machine Learning
    • XGBoost
    • Cross Entropy Loss
    • Other models
  • Graph Neural Network
    • GNN-1-Basic
  • Big Data
    • Reservoir Sampling
  • SQL
    • SQL and PySpark functions
    • Query Film Infomation
    • Create, Insert and Alter Actor Table
    • Manage Employment Data
    • Manage Employment Data -2
  • DataStructure
    • Searching
      • Find power
      • 2 Sum All Pair II
      • Two Sum
      • Search in Rotate Array
      • Search In Shifted Sorted Array II
      • Search in 2D array
      • Three Sum with duplicated values
      • Median of Two Sorted Arrays
    • Array
      • Longest Consecutive Subarray
      • Merge Two Array in-place
      • Trapping water
      • Rotate matrix
    • Sorting
      • Merge intervals
      • 排序
      • 最小的k个数
      • Find TopK largest- QuickSelect快速选择 method
      • MergeSort Linkedlist
      • 第K大元素
    • LinkedList
      • Reverse LinkedList I
      • Reverse K-group linked list
      • Detect Start of Cycle
      • HasCycle
      • DetectCycle II
      • 链表的共同节点
      • 链表中倒数第k个节点
      • 删除链表倒数第k个节点
      • 合并两个链表
      • 在排序数组中查找元素的第一个和最后一个位置
      • 删除链表里面重复的元素-1
    • Tree
      • Find Tree height (general iteration method)
      • Check BST and Check CompleteTree
      • ZigZag Order traversal
      • Binary Tree diameter I
      • Maximum Path Sum Binary Tree
      • Maximum Path Sum Binary Tree II
      • Binary Tree Path Sum To Target III
      • Tree diameter 树的直径II
      • Tree ReConstruction
      • Check if B is Subtree of A
      • The Kth smallest in Binary Search Tree
      • 打印Tree的右视图
      • 二叉搜索树的后序遍历序列
      • 重建二叉树
      • 判断二叉树是否对称
      • Path Sum to Target in Binary Tree
      • Tree-PreOrder-InOrder-PostOrder
    • Heap&Queue
      • Top-K smallest
      • 滑动窗口最大值
      • Find the K-Largest
    • 合并k个已排序的链表
    • String
      • Reverse String
      • 最长不含重复字符的子字符串
      • 最长回文串
      • 最长回文子序列-DP
    • DFS/BFS
      • Number of island
      • Number of Provinces
      • All Permutations of Subsets without duplication
      • All Permutations of Subsets with duplication
      • Combinations Of Coins
      • All Subset I (without fixing size of subset, without order, without duplication)
      • All Subset of K size without duplication II
      • All Subset of K size III (with duplication without considering order)
      • All Permutation II (with duplication and consider order)
      • Factor Combination-质数分解
    • DynamicProgramming
      • DP-解题过程
      • Find Continuous Sequence Sum to Target
      • 1800. Maximum Ascending Subarray Sum
      • NC91 最长上升子序列
      • 查找string的编码方式个数
      • Maximum Product
      • Longest Common Substring
      • Longest Common Substring-II
      • minEditCost
      • Backpack I
      • Array Hopper I
      • Minimum distance between strings
      • 最大正方形
  • Big Data Algorithms
    • Big Data Processing Algorithms
      • Reservior Sampling
      • Shuffle
      • MapReduce
      • Bloom Filter
      • BitMap
      • Heap For Big Data
Powered by GitBook
On this page
  • 1. Link
  • 2. 题目描述
  • 输入
  • 返回值

Was this helpful?

  1. DataStructure
  2. DynamicProgramming

Maximum Product

DP; Medium

Previous查找string的编码方式个数NextLongest Common Substring

Last updated 3 years ago

Was this helpful?

1. Link

2. 题目描述

给定一个double类型的数组arr,其中的元素可正可负可0,返回子数组累乘的最大乘积。示例1

输入

复制

[-2.5,4,0,3,0.5,8,-1]

返回值

复制

12.00000

3. 思路

  1. DP

  2. state: max_dp[i] ,min_dp[i] 以当前第i个位置为结尾的连续的乘积的的最大和最小值, 这个i一般是指上一次的结果不包含这次的arr[i]

  3. 因为有可能负负得正得到最大值,所以要保留负数最小值

  4. 每次从max_dp arr[i], min_dparr[i], arr[i] 里面取最大和最小更新

  5. 当前最大值和最小值, 那么其中最大值或最小值肯定有一个是乘上arr[i]或者reset成为arr[i], 这样下次乘上下一个arr[i+1]时乘积始终是连续的

  6. case 1: 前面累乘的值是 和当前值是同号

  7. case 2: 前面累乘的值和当前的值是异号

  8. 我们希望正值和负值的绝对值越乘越大,使得下一次再遇到负值的时候可以把当前累乘最小的负值变成正值,所以需要 max(max_valarr[i], min_val arr[i], arr[i]),

  9. 和arr[i]对比是为了考虑到前面累乘的数字是小数,结果绝对值越乘越小 比 arr[i]的绝对值还小的情况。

class Solution:
    def maxProduct(self , arr ):
        # state: max_dp[i] ,min_dp[i]上次状态的最大和最小值
        # transition function: 
        #max_dp_new = max(max_dp* arr[i] , min_dp *arr[i], arr[i])
        #min_dp_new = min(max_dp* arr[i] , min_dp *arr[i], arr[i])
        # 1. 因为有可能负负得正得到最大值,所以要保留负数最小值
        #2. 每次从max_dp *arr[i], min_dp*arr[i], arr[i] 里面取最大和最小更新
        #当前最大值和最小值, 那么其中最大值或最小值肯定有一个是乘上arr[i]或者reset成为
        # arr[i], 这样下次乘上下一个arr[i+1]时乘积始终是连续的
        #
        #
        if not arr:
            return None
        res = max_dp = min_dp = arr[0]
        for i in range(1, len(arr)):
            max_dp_new = max(max_dp* arr[i] , min_dp *arr[i], arr[i])
            min_dp_new = min(max_dp* arr[i] , min_dp *arr[i], arr[i])
            max_dp = max_dp_new
            min_dp = min_dp_new
            res = max(res, max_dp)
        return res

子数组最大乘积_牛客题霸_牛客网
Logo