非常详细的Splay笔记

本文深入探讨了Splay树的原理与操作,包括旋转、调整、构建等核心方法,并通过具体实例展示了其在反转子序列问题上的应用。Splay树是一种自调整的二叉搜索树,能够高效地进行查找、插入和删除操作。

Splay笔记

0.
第一次接触是去年暑假一位巨佬学长介绍了这么一种神奇的平衡树,不过当时感觉没怎么完全弄懂。后来仔细想了一下个人感觉理解大有加深,这个数据结构确实非常优秀~~~传到优快云上来啦!
1. rotate()
To maintain the features of BST, we can get these basic rotation rules:

If A is the x son of B,

  • B will become the !x son of A after the rotation.
  • The !x son of B will still be the !x son of B after the rotation.
  • The !x son of A will become the x son of B after the rotation.
  • The x son of A will still be the x son of A after the rotation.

Before & After one rotation:
在这里插入图片描述
CODE:

bool dir(int x)
{
    return x==tree[tree[x].fa].ch[1];
}

void pushup(int x)
{
    tree[x].tot=tree[tree[x].ch[0]].tot+tree[tree[x].ch[1]].tot+1;
    return;
}

void connect(int x,int fa,int son)
{
    tree[x].fa=fa;
    tree[fa].ch[son]=x;
    return;
}

void rotate(int x)
{
    int y=tree[x].fa;
    if(y==root)
        root=x; //Remember to change the root. 
    int r=tree[y].fa;
    int yson=dir(x),rson=dir(y);
    int build=tree[x].ch[yson^1];
    connect(build,y,yson);
    connect(y,x,yson^1);
    connect(x,r,rson);
    pushup(y);
    pushup(x);
    return;
}

More explanations to the rotate function:
在这里插入图片描述
The left is one of the original relationship among the grandfather, the father, and the son. After rotate(x), the relationship among them becomes the right part of the picture.
Other situations can be gotten similarly by reflecting or reversing.
2. splay()

splay(x,to) is to rotate x continuously until it becomes to’s son.

There are two conditions of x, x’s father, and x’s grandfather: whether they are in a straight line or not.

  • If dir(x)≠\not==dir(y), we can just simply rotate x twice.
    在这里插入图片描述
  • If dir(x)===dir(y), if we rotate x’s father then rotate x, we will get this:
    在这里插入图片描述
    Instead, if we rotate x twice, we will get this.
    在这里插入图片描述
    The first one can keep the tree more balanced, especially when there is a long chain in the tree.

But actually, both methods meet some situations that will make the situation more or less balance.

I choose the first in the following code.
CODE:

void splay(int x,int to)
{ 
    while(tree[x].fa!=to)
    {
        int y=tree[x].fa;
        if(tree[y].fa==to) //Sometimes x only needs to be rotated once. 
            rotate(x);
        else if(dir(x)==dir(y))
            rotate(y),rotate(x);
        else
            rotate(x),rotate(x);
    }
    pushup(x);
    return;
}

3. build()
I prefer to use recursion to build the tree…

int build(int l,int r)
{
    if(l>r)
        return 0;
    int mid=(l+r)>>1;
    connect(build(l,mid-1),mid,0);
    connect(build(mid+1,r),mid,1);
    tree[mid].rev=0;
    pushup(mid);
    return mid; //The return value is the root. 
}

4. Two Basic Problems
Important:Whenever\color{red}{Important:Whenever}Important:Whenever an\color{red}{an}an operation\color{red}{operation}operation about\color{red}{about}about nodes(which\color{red}{nodes(which}nodes(which means\color{red}{means}means asking\color{red}{asking}asking rank\color{red}{rank}rank is\color{red}{is}is excluded)\color{red}{excluded)}excluded) is\color{red}{is}is done,\color{red}{done,}done, splay\color{red}{splay}splay the\color{red}{the}the node.\color{red}{node.}node.

  • P3391 (Luogu)

Overview: Several commands (l,r) which mean that the subsequence from l to r in the array is reversed are given. Print out the array you get after m commands.

/*
This function is for updating the reverse tag of the node. 
The main idea is that there is no need to reverse a section twice, so you can save lots of time by marking. (Similar to the lazy tag of the segment tree.)
*/
void pushdown(int x)
{
    if(tree[x].rev)
    {
        swap(tree[x].ch[0],tree[x].ch[1]); 
        tree[tree[x].ch[0]].rev^=1;
        tree[tree[x].ch[1]].rev^=1;
        tree[x].rev=0;
    }
    return;
}

int find(int x)
{
    int now=root;
    x--;
    pushdown(now);
    while(x!=tree[tree[now].ch[0]].tot)
    {
        if(tree[tree[now].ch[0]].tot<x)
            x-=tree[tree[now].ch[0]].tot+1,now=tree[now].ch[1];
        else
            now=tree[now].ch[0];
        pushdown(now);
        }
    return now;
}

PRINT:

void print(int now)
{
    if(!now)
        return;
    pushdown(now);
    print(tree[now].ch[0]);
    if(now!=1 && now!=n+2)
      cout<<now-1<<' ';
    print(tree[now].ch[1]);
    return; 
}

MAIN:

PosL=find(l);
splay(PosL,0);
PosR=find(r+2);
splay(PosR,root);
tree[tree[PosR].ch[0]].rev^=1;

(待填坑)有空我把splay的普通平衡树也实现一下放上来。有空再更一下splay的时间复杂度~~~

内容概要:本文详细介绍了“秒杀商城”微服务架构的设计与实战全过程,涵盖系统从需求分析、服务拆分、技术选型到核心功能开发、分布式事务处理、容器化部署及监控链路追踪的完整流程。重点解决了高并发场景下的超卖问题,采用Redis预减库存、消息队列削峰、数据库乐观锁等手段保障数据一致性,并通过Nacos实现服务注册发现与配置管理,利用Seata处理跨服务分布式事务,结合RabbitMQ实现异步下单,提升系统吞吐能力。同时,项目支持Docker Compose快速部署和Kubernetes生产级编排,集成Sleuth+Zipkin链路追踪与Prometheus+Grafana监控体系,构建可观测性强的微服务系统。; 适合人群:具备Java基础和Spring Boot开发经验,熟悉微服务基本概念的中高级研发人员,尤其是希望深入理解高并发系统设计、分布式事务、服务治理等核心技术的开发者;适合工作2-5年、有志于转型微服务或提升架构能力的工程师; 使用场景及目标:①学习如何基于Spring Cloud Alibaba构建完整的微服务项目;②掌握秒杀场景下高并发、超卖控制、异步化、削峰填谷等关键技术方案;③实践分布式事务(Seata)、服务熔断降级、链路追踪、统一配置中心等企业级中间件的应用;④完成从本地开发到容器化部署的全流程落地; 阅读建议:建议按照文档提供的七个阶段循序渐进地动手实践,重点关注秒杀流程设计、服务间通信机制、分布式事务实现和系统性能优化部分,结合代码调试与监控工具深入理解各组件协作原理,真正掌握高并发微服务系统的构建能力。
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值