本系列索引
归并排序原理
归并排序,顾名思义,先有向下拆分,再有向上合并 以升序排序为例,首先我们将原数组拆分成两份,然后定义一个新数组和两个指针指向拆分后的两个数组,当左边数组指针指向的值比右边的小的时候就将其放入新数组中,否则就将右边的放入新数组中,如此操作直到双指针遍历完毕
归并排序的作用
这个算法可以用来处理大数据,比如说现有内存2GB,如何对40GB大小的文件进行升序排序?
首先我们可以将这个大文件拆分成20份2GB的小文件,然后进行20路归并。
每次从20个值中选出最大值,这个操作可以用小根堆完成。
由于数组temp 可以放在外部的硬盘中,因此归并排序和快速排序相比,归并排序是一种外部排序。
相关习题
剑指 Offer 51 数组中的逆序对
function counter(nums: number[], l: number, r: number) {
if (l >= r) return 0
const mid = (l + r) >> 1
let res = 0, p1 = l, p2 = mid + 1
res += counter(nums, l, mid)
res += counter(nums, mid + 1, r)
const temp: number[] = []
while (p1 <= mid || p2 <= r) {
if (p2 > r || (p1 <= mid && nums[p1] > nums[p2])) {
temp.push(nums[p1++])
res += r - p2 + 1
} else {
temp.push(nums[p2++])
}
}
for (let i = l; i <= r; i++) nums[i] = temp[i - l]
return res
}
function reversePairs(nums: number[]): number {
return counter(nums, 0, nums.length - 1)
}
leetCode 148 排序链表
这道题解法和数组排序类似,只是由于链表不能直接获取链表长度,所以对归并函数新增一个参数用于表示链表长度,方便我们对链表拆成两份
function mergeSort(head: ListNode | null, len: number) {
if (!head || !head.next) return head
const mid = len >> 1
let p = head
for (let i = 1; i < mid; i++) p = p.next
const temp = p
p = p.next
temp.next = null
const lList = mergeSort(head, mid), rList = mergeSort(p, len - mid)
const vHead = new ListNode()
let p1 = lList, p2 = rList
p = vHead
while (p1 || p2) {
if (!p2 || (p1 && p1.val < p2.val)) {
p.next = p1
p1 = p1.next
} else {
p.next = p2
p2 = p2.next
}
p = p.next
}
p.next = null
return vHead.next
}
function sortList(head: ListNode | null): ListNode | null {
if (!(head && head.next)) return head
let len = 0, p = head
while (p) {
len++
p = p.next
}
return mergeSort(head, len)
}
leetCode 1305 两颗二叉搜索树中的所有元素
在二叉树一章提过,二叉搜索树的中序遍历就是一个升序数组,所以这题我们先对两颗二叉搜索树进行中序遍历,再进行合并
function median(root: TreeNode | null, ans: number[]) {
if (!root) return root
median(root.left, ans)
ans.push(root.val)
median(root.right, ans)
}
function getAllElements(root1: TreeNode | null, root2: TreeNode | null): number[] {
const arr1: number[] = [], arr2: number[] = [], ans: number[] = []
median(root1, arr1)
median(root2, arr2)
let p1 = 0, p2 = 0
while (p1 < arr1.length || p2 < arr2.length) {
if (p2 >= arr2.length || (p1 < arr1.length && arr1[p1] < arr2[p2])) {
ans.push(arr1[p1++])
} else {
ans.push(arr2[p2++])
}
}
return ans
}
leetCode 347 区间和的个数
这个地方用到了前面文章提到的前缀和数组,要计算区间和我们只要计算前缀和数组的差值就行了
根据题意我们的前缀和数组要符合的条件是
l
o
w
e
r
≤
s
u
m
[
j
]
?
s
u
m
[
i
]
≤
u
p
p
e
r
(
i
<
j
)
lower \leq sum[j] - sum[i] \leq upper (i < j)
lower≤sum[j]?sum[i]≤upper(i<j)
let _lower: number, _upper: number, temp: number[] = []
function countTwoParts(nums: number[], l1: number, r1: number, l2: number, r2: number) {
let ans = 0
for (let a = l1, b = l1, i = l2; i <= r2; i++) {
while (a <= r1 && nums[i] - nums[a] > _upper) a++
while (b <= r1 && nums[i] - nums[b] >= _lower) b++
ans += b - a
}
return ans
}
function mergeSort(arr: number[], l: number, r: number) {
if (l >= r) return 0
let ans = 0
const mid = (l + r) >> 1
ans += mergeSort(arr, l, mid)
ans += mergeSort(arr, mid + 1, r)
ans += countTwoParts(arr, l, mid, mid + 1, r)
let p1 = l, p2 = mid + 1, k = l
while (p1 <= mid || p2 <= r) {
if (p2 > r || (p1 <= mid && arr[p1] < arr[p2])) {
temp[k++] = arr[p1++]
} else {
temp[k++] = arr[p2++]
}
}
for (let i = l; i <= r; i++) arr[i] = temp[i]
return ans
}
function countRangeSum(nums: number[], lower: number, upper: number): number {
const sum = [0]
nums.forEach((num, i) => sum.push(num + sum[i]))
_lower = lower, _upper = upper, temp = new Array(sum.length).fill(0)
return mergeSort(sum, 0, sum.length - 1)
}
leetCode 315 计算右侧小于当前元素的个数
首先定义一种数据结构 [val: number, ind: number, cnt: number],用于存储原数组的值、下标和题目所求数量(初始为0),归并的时候根据原数组的值来排,在合并的时候对 cnt 进行操作,最后根据下标还原元素在原数组的位置并输出 cnt
type dataType = [val: number, ind: number, cnt: number]
function mergeSort(data: dataType[], l: number, r: number) {
if (l >= r) return
const mid = (l + r) >> 1, temp: dataType[] = []
mergeSort(data, l, mid)
mergeSort(data, mid + 1, r)
let p1 = l, p2 = mid + 1
while (p1 <= mid || p2 <= r) {
if (p2 > r || (p1 <= mid && data[p1][0] > data[p2][0])) {
data[p1][2] += r - p2 + 1
temp.push(data[p1++])
} else {
temp.push(data[p2++])
}
}
for (let i = l; i <= r; i++) data[i] = temp[i - l]
}
function countSmaller(nums: number[]): number[] {
const data: dataType[] = [], ans: number[] = []
nums.forEach((num, i) => data.push([num, i, 0]))
mergeSort(data, 0, data.length - 1)
data.forEach(d => ans[d[1]] = d[2])
return ans
}
53 最大子序和
到现在为止,说到区间和,我们应该就要想到前序和了,这题也是用前序和数组来解,在遍历前序和的时候我们只要维护前序和数组的最小值即可
function maxSubArray(nums: number[]): number {
const sums = [0]
nums.forEach((num, i) => sums.push(num + sums[i]))
let ans = sums[1]
for (let i = 1, pre = 0; i < sums.length; i++) {
ans = Math.max(sums[i] - pre, ans)
pre = Math.min(pre, sums[i])
}
return ans
}
复习
leetCode 23 合并K个升序链表
这题使用堆来解
首先将所有链表压入小顶堆,定义虚拟头节点和移动指针P,然后当栈不为空时一直弹出堆顶元素并连接在新链表后,如果弹出的节点有下一位则将下一位节点继续压入堆
type dataType = ListNode | null
class SmallHeap {
data: dataType[]
constructor(data: dataType[]) {
this.data = data
this.init()
}
init() {
for (let i = 0; i < this.data.length; i++) {
if (!this.data[i]) {
this.data.splice(i, 1)
i--
}
this.sortUp(i)
}
}
push(val: dataType) {
this.data.push(val)
this.sortUp(this.data.length - 1)
}
pop() {
if (!this.data.length) return null
let ret: dataType
if (this.data.length === 1) {
ret = this.data.pop()
} else {
ret = this.data[0]
this.data[0] = this.data.pop()
this.sortDown(0)
}
if (ret.next) this.push(ret.next)
return ret
}
sortUp(index: number) {
let parentIndex: number
while (index > 0) {
parentIndex = (index - 1) >> 1
if (this.data[index].val < this.data[parentIndex].val) {
this.swap(index, parentIndex)
} else {
return
}
index = parentIndex
}
}
sortDown(index: number) {
let target: number, lIndex: number, rIndex: number
while (index < this.data.length) {
target = index, lIndex = (index << 1) + 1, rIndex = (index << 1) + 2
if (lIndex < this.data.length && this.data[lIndex].val < this.data[target].val) {
target = lIndex
}
if (rIndex < this.data.length && this.data[rIndex].val < this.data[target].val) {
target = rIndex
}
if (target === index) return
this.swap(index, target)
index = target
}
}
swap(a: number, b: number) {
[this.data[a], this.data[b]] = [this.data[b], this.data[a]]
}
size() {
return this.data.length
}
}
function mergeKLists(lists: Array<ListNode | null>): ListNode | null {
const ret = new ListNode(), q = new SmallHeap(lists)
let p = ret
while (q.size()) {
p.next = q.pop()
p = p.next
p.next = null
}
return ret.next
}
leetCode 1508 子数组和排序后的区间和
这题也是用堆来解,重点是定义好节点的结构 type dataType = [val: number, pos: number]
第一个元素是原数组的每一项,第二项是原数组的累加位置
比如示例1的 [1, 2, 3, 4] 中的 1 的子数组有 [1] [1, 2] [1, 2, 3] [1, 2, 3, 4]
type dataType = [val: number, pos: number]
class Priority {
private readonly data: dataType[]
constructor() {
this.data = []
}
size() {
return this.data.length
}
push(val: dataType) {
this.data.push(val)
this.sortUp(this.data.length - 1)
}
pop() {
if (!this.data.length) return null
if (this.data.length === 1) return this.data.pop()
const ret = this.data[0]
this.data[0] = this.data.pop() as dataType
this.sortDown(0)
return ret
}
sortUp(i: number) {
let parentIndex: number
while (i) {
parentIndex = (i - 1) >> 1
if (this.data[i][0] < this.data[parentIndex][0]) this.swap(i, parentIndex)
else return
i = parentIndex
}
}
sortDown(i: number) {
let target: number, lChild: number, rChild: number
while (i < this.data.length) {
target = i, lChild = (i << 1) + 1, rChild = (i << 1) + 2
if (lChild < this.data.length && this.data[lChild][0] < this.data[target][0]) target = lChild
if (rChild < this.data.length && this.data[rChild][0] < this.data[target][0]) target = rChild
if (i === target) return
this.swap(i, target)
i = target
}
}
swap(i: number, j: number) {
[this.data[i], this.data[j]] = [this.data[j], this.data[i]]
}
}
function rangeSum(nums: number[], n: number, left: number, right: number): number {
let ans = 0, item: dataType
const q = new Priority(), mod = 1e9 + 7
nums.forEach((num, i) => q.push([num, i]))
for (let i = 1; i <= right && q.size(); i++) {
item = q.pop() as dataType
if (item[1] + 1 < nums.length) q.push([item[0] + nums[item[1] + 1], item[1] + 1])
if (i >= left) ans = (item[0] + ans) % mod
}
return ans
}
面试题 04.08 首个共同祖先
首先定义一个递归函数,赋予它的意义是找到子树中的目标节点
如果在左右子树都找到目标节点,就说明本节点是目标节点的共同祖先; 如果左树是空,则说明首个共同祖先在右树; 否则,首个共同祖先在左树。
var lowestCommonAncestor = function(root, p, q) {
if (!root) return null
if (root === p || root === q) return root
const lTree = lowestCommonAncestor(root.left, p, q), rTree = lowestCommonAncestor(root.right, p, q)
if (lTree && rTree) return root
if (!rTree && lTree) return lTree
return rTree
}
1302 层数最深叶子节点的和
这题使用递归,接收节点、深度和状态记录
状态记录 type stateType = [val: number, depth: number] 第一个参数是累加和,第二个参数记录的是最深层数
type stateType = [val: number, depth: number]
function counter(root: TreeNode | null, depth: number, curState: stateType) {
if (!root) return curState
if (depth === curState[1]) {
curState[0] += root.val
} else if (depth > curState[1]) {
curState[0] = root.val
curState[1] = depth
}
counter(root.left, depth + 1, curState)
counter(root.right, depth + 1, curState)
}
function deepestLeavesSum(root: TreeNode | null): number {
if (!root) return 0
const curState: stateType = [root.val, 0]
counter(root, 0, curState)
return curState[0]
}
发发牢骚
工作之后时间真的少很多,一周加班三四天,晚上回到家锻炼洗漱后就差不多该睡觉了。。有时题目虽然在公司偷偷摸鱼做完,但是不方便写,忙了一天回到家后真的是精疲力竭,挺无语的
本来还在学校的时候想着毕业后自己一个人生活了就不会被干扰、就有更多时间钻研了,但是事实是每个时间段都有那个时间段的烦恼,唉,想办法克服当下困难才是正道。。继续加油吧,总会有上岸的一天的
|