w***g 发帖数: 5958 | 1 【 以下文字转载自 Mathematics 讨论区 】
发信人: borg (伦理审查协会), 信区: Mathematics
标 题: arXiv又放出神作:NP=P (by 国防科大)
发信站: BBS 未名空间站 (Fri May 31 16:03:53 2013, 美东)
http://arxiv.org/abs/1305.5976 |
S**I 发帖数: 15689 | 2 老张再度掀起了民科攻坚世界难题的热潮。
【在 w***g 的大作中提到】 : 【 以下文字转载自 Mathematics 讨论区 】 : 发信人: borg (伦理审查协会), 信区: Mathematics : 标 题: arXiv又放出神作:NP=P (by 国防科大) : 发信站: BBS 未名空间站 (Fri May 31 16:03:53 2013, 美东) : http://arxiv.org/abs/1305.5976
|
w***g 发帖数: 5958 | 3 这东西是在老张之前做出来的. 作者也不像是民科, 虽然这证明正确的可能性微乎其微
, 我还是希望它是对的.
【在 S**I 的大作中提到】 : 老张再度掀起了民科攻坚世界难题的热潮。
|
S**I 发帖数: 15689 | 4 If he proved P≠NP, there is a slight chance that I might believe he got it;
P=NP sounds too good to be true.
【在 w***g 的大作中提到】 : 这东西是在老张之前做出来的. 作者也不像是民科, 虽然这证明正确的可能性微乎其微 : , 我还是希望它是对的.
|
w***g 发帖数: 5958 | 5 这文章确实不靠谱, 浪费感情了.
it;
【在 S**I 的大作中提到】 : If he proved P≠NP, there is a slight chance that I might believe he got it; : P=NP sounds too good to be true.
|
g*****g 发帖数: 34805 | 6 If he can prove NP=P, it's easy to verify, just give a P solution to an NP
problem. |
b********e 发帖数: 58 | 7 Thanks for the link.
Rule of thumb, if a paper of such topic was written with tools other than
LaTex, most likely it is flawed. :-) |
s*******e 发帖数: 42 | 8 The math only appears to confuse readers. He can get A for a homework for
doing this. But I doubt he can actually discover something in this way. He
should do some simulations to show his point. If he hasn't done, it is
probably wrong. |
r****y 发帖数: 26819 | 9 google了一下,发现先驱还有这位:
http://www.iaeng.org/publication/IMECS2010/IMECS2010_pp292-294.
【在 w***g 的大作中提到】 : 【 以下文字转载自 Mathematics 讨论区 】 : 发信人: borg (伦理审查协会), 信区: Mathematics : 标 题: arXiv又放出神作:NP=P (by 国防科大) : 发信站: BBS 未名空间站 (Fri May 31 16:03:53 2013, 美东) : http://arxiv.org/abs/1305.5976
|
s*******e 发帖数: 42 | 10 It's talking about probabilistic algorithm (and hints general algorithm may
work too).
IMHO, NP problems may be solved in P with probability > 1 - epison (epison
can be arbitrarily small), if the space used by the algorithm can be
arbitrarily big.
Think Go playing as an example, if you solve it by searching, it is
exponential time. But if you solve it by pattern matching (plus some
algorithm), it may be polynomial time, if your pattern book is extremely big
to cover 1 - epison cases. |
s*******e 发帖数: 42 | 11 Did AlphaGo steal the algorithm here? |