On Wed, Feb 26, 2014 at 11:05 AM, Matt Mahoney <[email protected]>wrote:

> On Tue, Feb 25, 2014 at 2:36 AM, YKY (Yan King Yin, 甄景贤)
> <[email protected]> wrote:
> > Georg Cantor, the father of set theory, first used the diagonal argument
> to prove that the rational numbers are uncountable, ie, that the cardinal
> number of rational numbers is bigger than the cardinal number of the
> integers.
>
> Real numbers, not rational. There is a 1:1 mapping between rational
> numbers and integers. Specifically, numbers of the form a/b can be
> ordered by increasing a+b and then by increasing a.
>


Thanks for the correction =)

​​

> With regard to the human brain solving NP-complete problems, it may
> seem that most problems are easy because they are either
> over-constrained or under-constrained. It is only the small set of
> problems near the boundary that are hard. For example, the traveling
> salesman problem asks if there is a cycle through all the vertices of
> a graph G whose labeled edges sum to less than D. If D is very large,
> then it is easy to find a satisfying path. If D is very small then it
> is easy to prove that the answer is no (for example, D is smaller than
> the distance between two vertices). It is only when D is near the best
> solution that the problem gets hard.
>


Yes, that is the "phase transition"​​ phenomena observed in some
combinatorial problems, such as SAT:
http://cstheory.stackexchange.com/questions/14953/current-tightest-bounds-for-critical-3-sat-density

Also, the number of instances of SAT problems (3-SAT, anyway) for each N,
is actually finite and quite small:
http://cstheory.stackexchange.com/questions/2168/how-many-instances-of-3-sat-are-satisfiable?rq=1
This fact may be helpful in defining approximation -- as probability
distributions over such discrete, finite spaces.



-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657
Powered by Listbox: http://www.listbox.com

Reply via email to