Quote:
Originally Posted by thylacine
Exactly! You got it completely! The limit (as n goes to infinity) of the probabilities (for a permutation of n elements to be a derangement) exists and is 1/e, but it seems that the limit of the corresponding probability distributions is not well defined.
I wonder if this has some connection to the topic of `amenable groups'.
http://en.wikipedia.org/wiki/Amenable_group
Anyway it's fair to say that the "probability" of a "random" permutation of and infinite set being a derangement "should" be 1/e. It's just so tempting to say that. So I wonder to what extent this can be set up in a well defined way in the infinite case.
Sorry if this thread seems out of place amongst the "what is the probability of getting two heads in a row' threads.
Sure, that would be "nice". It seems like there could be a connection with
discrete amenable groups, but you would have to consult a researcher in
this field to get an idea of what type of "nice theorems" work with what kind
of discrete groups ( and it seems that the countable case could have been
analyzed already ). I wonder if it is true that for all p in (0,1) ( forget 0 and
1 for now ), there exists some "way" to choose a random bijection ( that
resembles the "equally weighted" means for finite permutations ) of the set
N of natural numbers onto itself such that the "probability" of no fixed point
is p. Of course, there "should be" many "ways" where this "probability" is 1/e
and it could be that for "almost every reasonable way" this "probability" is 1/e.
Also, it could even be wrong to think that "infinite permutations" should
behave like finite permutations. [ I put probability in quotes, because there
may not be a nice way to define such a measure on the set of permutations. ]
It's refreshing to have "open problems" rather than the usual common
questions that could be answered by "googling".