MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Do paracompact non-Hausdorff spaces admit partions of unity? I'm just curious.

share|cite|improve this question
up vote 7 down vote accepted

The answer is no. Take the "classical" example of the line with two origins. This space is non-Hausdorff, paracompact and doesn't admit partitions of unity.

EDTI: I think the question is a kind of "duplicate" . Ok, but if you have an example for a non-Hausdorff manifold, which doesn't admit partitions of unity, you have an example for a non-Hausdorff paracompact space with the same property.

First the definition:
The line with two origins is the quotient space of two copies of the real line $\mathbb{R} \times {a}$ and $\mathbb{R} \times {b}$.
with equivalence relation given by $(x,a) \sim (x,b)\text{ if }x \neq 0$.
Since all neighbourhoods of $0_a$ intersect all neighbourhoods of $0_b$, it is non-Hausdorff.
However, this space is paracompact, since $\mathbb{R}$ is paracompact.

For the non-existence of a partition of unitiy: take the open covering $ U = (-\infty,0) \cup \{ 0_a \} \cup (0,\infty)$ and $\tilde{U} = (-\infty,0) \cup \{ 0_b \} \cup (0,\infty)$. Assume, there is a partition of unity subordinate to this cover. Then the value of each origin would have to be $1$ which cannot be true. (Edit: villemoes was a little faster :-) )

share|cite|improve this answer
It was in fact the question to which you linked which made me wonder this. This is not a duplicate question, since Aston insisted on a Hausdorff manifold. How can I see that the example you give is paracompact, and how can I see there is no partition of unity? – David Carchedi Apr 1 '10 at 8:23
Take any open cover of the "real line" for which there are exactly one open set U_1 containing one copy of 0 (call this 0_1), and one open set U_2 containing the other copy of 0 (0_2). A partition of unity subordinate to this open covering must contain a function f_1 supported in U_1, such that f_1(0_1) = 1, since 0_1 is only contained in this open set. Similarly for 0_2. It follows by continuity that for some small epsilon, we have f_1(epsilon) + f_2(epsilon) > 1, so it is not a partition of unity. – villemoes Apr 1 '10 at 10:52

It's worth noting that any $T_1$ space which admits partitions of unity for finite (two element even) covers is Hausdorff:

Proof: Let $x, y \in X$. Let $U = X \ \{x\}, V = X \ \{y\}$. Then let $\{f, g\}$ form a partition of unity with $f$ subordinate to $U$ and $g$ subordinate to $V$. Then $A = \{ t : f(t) > \frac{1}{2} \}$ and $B = \{ t : g(t) > \frac{1}{2} \}$ A and B are disjoint open sets with $y \in A$ and $x \in B$.

Edit: On closer inspection, this if of course just the standard proofs that the existence of partitions of unity for finite covers implies normality + the fact that $T_1$ normal spaces are hausdorff

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.