Sunday, 13 October 2013

general topology - Does a continuous function preserve Bolzano-Weierstrass property?





Let X,Y be arbitrary topological spaces with X satisfying the Bolzano-Weierstrass property (i.e. every infinite subset has some accumulation point) and let f:XY be a continuous function, does f(X) satisfy the B-W property?




I tried to prove it does by contradiction as follows:



Let's assume f(X) doesn't have the BWP, so there must be some infinite subset A wich has not any accumulation point, hence it's is closed. Because f is continuous, f1(A) is also closed and infinite and because X has the BWP there is some accumulation point of f1(A) which is contained in it, let's call x0 one of such points. Now, f(x0) is an isolated point of A, so {f(x0)} is an open neighborhood of f(x0) (with the relative topology of f(X) as a subspace of Y).
This means that f1(f(x0)) is an open neighborhood of x0 which contains other points different than x0. But this doesn't lead me to a contradiction since in principle f could be locally constant at x0. Is there some reason why this cannot happen? Or maybe continuous function doesn't actually preserve BWP, if that's the case could someone provide a counterexample? Constant functions doesn't seem to work because they have finite image, so I thought maybe a locally constant function with infinite image, but I couldn't build up an explicit example.



EDIT to avoid future missunderstandings: What I called 'accumulation point' above is what usually is refered in english as 'limit point'. Explicitly a is an accumulation (limit) point of a set A of a topological space X if every open set containing a intersected with A contains some point different than a


Answer



This is actually false if "accumulation point" = "limit point" where x is a limit point of A iff every neighbourhood of x intersects A{x}.

Your B-W property is then called "limit point compact"



A counterexample: Let Y=N as a discrete space and let X=Y×{0,1} where {0,1} has the indiscrete (trivial) topology with no non-trivial open sets. Let f=πY be the projection onto Y, which is continuous.
X is limit point compact: if AX, and (x,i)A, then (x,i) is a limit point of A, where i{0,1}{i}, because any product open set that contains (x,i) also contains (x,i) and vice versa. But Y is infinite discrete so not limit point compact.



If X is T1 then (here our example is not even T0) limit point compact is equivalent to "strong limit point compact": every infinite set A has an (ω-)accumulation point, i.e. an xX such that every neighbourhood of x contains infinitely many points from A. This property is preserved by continuous functions:



Suppose X is strong limit point compact, and f:XY is continuous.
Let Af[X] be infinite and let {yn:nω} be an enumeration (ynym for nm) of a countably infinite subset of A which means that we have for all n we have xnX such that f(xn)=yn. Then clearly, nm implies xnxm as well, so {xn:nω} is an infinite subset of X. So this has a strong limit point x. Now let y=f(x)f[X].
If O is any open set containing y then f1[O] is open by continuity and contains x, so there are infinitely many n with xnf1[O]. All those n also obey yn=f(xn)O. So y is a strong limit point of A, and f[X] is strong limit point compact.




So your original statement is true for T1 spaces X. We then get even a (slightly) stronger property than B-W in Y (without assuming anything extra on Y), namely strong limit compact.


No comments:

Post a Comment

real analysis - How to find limhrightarrow0fracsin(ha)h

How to find lim without lhopital rule? I know when I use lhopital I easy get $$ \lim_{h\rightarrow 0}...