Proof of the contraction mapping theorem

学的这个课程

强化学习的数学原理_中国大学MOOC(慕课)

算是强化学习里面很火的课程了(゚∀゚)

老师的讲义和书都是英文的,可以看看啊,词汇都不难,主要是理解一些专业词汇

这个是第三章里面压缩映射定理的证明,正好学习一下Latex

CSDN里面公式显示好像不太稳定,可以直接去看书

1. we prove that the sequence with is convergent.

the proof relies on Cauchy sequences: for any small , there exists such that for all

if we simply have , it is insufficient to claim that the sequence is a Cauchy sequence. For example, it holds that for , but apparently, it diverges.

we next show that is a Cauchy sequence and hence converges.

first, since is a contraction mapping, we have

thus,we have

since ,we know that converges to zero exponentially fast as given any

we need to futher consider for any . In particular,

Therefore, this sequence is Cauchy and hence converges to a limit point denoted as

2. we show that the limit is a fixed point.

converges to zero exponentially fast

3. we show that the fixed point is unique.

suppose that there is another fixed point , then

this inequality holds if and only if . therefore,

4. we show that converges to exponentially fast as

相关推荐
程序员cxuan26 分钟前
一句话,让你用上 GPT-5.6
人工智能·后端·程序员
机器之心28 分钟前
AI圈刚开始谈Loop Engineering,两位95后博士已经盯上了人类闭环数据
人工智能·openai
澄旭29 分钟前
一文讲清 MCP:AI 应用连接外部世界的标准协议
人工智能
机器之心37 分钟前
不只DeepSeek,阶跃等开源JetSpec:大模型解码提速近10倍
人工智能·openai
moMo1 小时前
当LLM学会"递纸条",AI是如何调用工具的
人工智能
拾年2751 小时前
大模型的"聪明"从哪来?聊聊 AI 数据集的那些事儿
人工智能·深度学习·机器学习
拾年2751 小时前
从 Prompt 到 Context 再到 Harness:AI 工程化的三年三级跳
人工智能
用户3090463613942 小时前
Claude 不会直接执行你的函数,它只会生成一段结构化的工具调用请求。真正执行函数、访问数据库、请求外部 API 的动作,必须由你的后端完成。
人工智能
不加辣椒2 小时前
第14章 Prompt 编排与优化技术
人工智能