A Proof That P Is Not Equal To NP?
来源:互联网 发布:2017美工电脑配置要求 编辑:程序博客网 时间:2024/04/27 19:44
From
http://rjlipton.wordpress.com/2010/08/08/a-proof-that-p-is-not-equal-to-np/
A serious proof that claims to have resolved the P=NP question.
Vinay Deolalikar is a Principal Research Scientist at HP Labs who has done important research in various areas of networks. He also has worked on complexity theory, including previous work on infinite versions of the P=NP question. He has just claimed that he has a proof that P is not equal to NP. That’s right: . No infinite version. The real deal.
Today I will talk about his paper. So far I have only had a chance to glance at the paper; I will look at it more carefully in the future. I do not know what to think right now, but I am certainly hopeful.
The Paper
Deolalikar’s draft paper is here—he was kind enough to give me the pointer to his paper. For now please understand that this is a “preliminary version.” See the discussion by Greg Baker, who first posted on his paper.
At first glance it is a long, well written paper, by a serious researcher. He clearly knows a great deal of complexity theory and mathematics. His Ph.D. thesis at USC is titled:
On splitting places of degree one in extensions of algebraic function fields, towers of function fields meeting asymptotic bounds, and basis constructions for algebraic-geometric codes.
My first thought is who knows—perhaps this is the solution we have all have been waiting for. If it is correct, the Clay list will drop down to five. I assume he would take the prize.
But first there is the small issue of correctness. Is his paper correct?
I suggest you look at his paper to see his own summary of his approach, and of course the details of his proof. At the highest level he is using the characterization of polynomial time via finite model theory. His proof uses the beautiful result of Moshe Vardi (1982) and Neil Immerman (1986):
Theorem: On ordered structures, a relation is defined by a first order formula plus the Least Fixed Point (LFP) operator if and only if it is computable in polynomial time.
Then, he attacks SAT directly. He creates an ordered structure that encodes SAT. He then argues if P=NP, then by the above theorem it must follow that SAT has certain structural properties. These properties have to do with the structure of random SAT. This connection between finite model theory and random SAT models seems new to me.
The one thing that strikes me immediately is his use of finite model theory. This is one area of logic that has already led to at least one breakthrough before in complexity theory. I believe that Neil used insights from this area to discover his famous proof that is closed under complement. It is interesting that the “final” proof does not directly use the machinery of finite model theory. Deolalikar’s connection between model theory and the structure of random SAT is interesting. I hope it works, or at least sheds new light on SAT.
An obvious worry about his proof, just from a quick look, is the issue of relativization. I believe that the LFP characterization, and similar first order arguments do relativize in general. However, it is possible that his use of concretely encoded structures prevents his entire argument from relativizing. We will need to check carefully that his proof strategy evades this limitation. I am on record as not being a fan of oracle results, so if this is the problem for his proof, I will have to re-think my position. Oh well.
Deolalikar does cite both Baker-Gill-Solovay for relativization and Razborov-Rudich for the “Natural Proofs” obstacle. His proof strategy ostensibly evades the latter because it exploits a uniform characterization of P that may not extend to give lower bounds against circuits. In fact the paper does not state a concrete time lower bound for SAT, as the proof is by contradiction. Since the gap in the contradiction is between “” and “,” it is possible that a time lower bound of “ for some ” is implied. More will have to wait until there is time to examine all the threads of this long and complex paper closely. However, the author certainly shows awareness of the relevant obstacles and command of literature supporting his arguments—this is a serious effort.
Open Problems
Is his paper correct? How does he avoid all the “barriers” that have been claimed to surround the P=NP question? Let’s hope it all checks out.
=========
相关链接
关于P!=NP
- A Proof That P Is Not Equal To NP?
- to be or not to be, that is a question...
- To be or not to be that is a question
- fragment Trying to instantiate a class com.example.testhuanxindemo.MyFragment that is not a Fragmen
- An attempt was made to set a report parameter that is not defined in this report
- How to arrange 10 digits so that the product of the some of them is equal to a number represented by the remaining digits
- The proof is in the code. That is all.
- 错误 the type "xxxx" is defined in an assembly that is not refernced.You must add a reference to assem
- [morals is a way to success]How to know a float number is equal to another
- TO BE OR NOT TO BE THAT IS AQUESTION
- To be or Not to be that is the question
- File opened that is not a database file file is encrypted or is not a databa
- How to call a closure that is a class variable?
- *(int*)&p , what is that?
- How to sort by a form that is not displayed in the document in MicroStrategy Desktop 8.x
- YAML safe loading is not available. Please upgrade psych to a version that supports safe loading (>=
- 安装cocoapods遇到YAML safe loading is not available. Please upgrade psych to a version that supports saf
- To Be a Coder or Not, That's a Question!
- InstallShield创建自定义对话框 实例
- 使用oledb访问Oracle,ExecuteNonQuery是否立刻提交呢,是否会造成死锁
- VC6环境下USB编程链接错误解决办法
- Dom4j下载及使用Dom4j读写XML简介
- 【转】C# winform程序发布后找不到数据文件(xml,ini,etc...)的解决
- A Proof That P Is Not Equal To NP?
- Socket通信-web服务器基本原理(静态)
- pku2409Let it Bead
- 破解Red Hat Enterprise 4的root密码
- 配置ASP.NET网站使用AppFabric Caching存储Session数据
- WebView,WebViewClient,WebChromeClient区别
- 链式队列一些操作
- Squid-3.1.6在CentOS上编译出错提示缺少libltd库
- JS创建Table表单切合并单元格