Baire's Theorem proof regarding points revisited

112 Views Asked by At

My first question on this point was not answered. Here is the first part of Shilov's proof of Baire's theorem (not an exact lift from the book as I avoided mathematical symbols). I am trying to be more explicit here so perhaps I will get an answer.

THEOREM: Suppose a complete metric space is the union of a countable number of closed subsets $F_1, F_2 \ldots \subset M$. Then at least one subset $F_n$ contains a closed ball in $M$.

PROOF: Suppose to the contrary that none of the sets $F_1, F_2, \ldots$ contains a closed ball, and let $X_1$ be a point not belonging to $F_1$.

(First question: Why does $M$ need to have any points whatsoever? In which case there are no points not belonging to $F_1$.)

Since $F_1$ is closed there is a closed ball surrounding $X_1$ that does not intersect $F_1$. And within one half the radius of this closed ball there is a point $X_2$ not belonging to $F_2$. ETC ETC ETC

(Second Question: Why can't a complete metric space have only one single point $X_1$? Why does it need to have this second point?)

1

There are 1 best solutions below

4
On

Answer to both your questions: $M$ doesn't need to be non-empty or contain more than two points. The thing is that the Baire category theorem is uninteresting if $M$ doesn't contain infinitely many points, since for such spaces the only dense set is the whole space itself. So, in the proof one only needs to treat the case where $M$ is an infinite set.

Edit: This is not to imply of course, one can just leave out "uninteresting" cases. What I was trying to say is that in these cases only the whole space $X$ is dense. So if we have a collection of dense sets and we take the intersection, we end up with $X$ again, which is of course dense in itself. So the theorem holds trivially.