# Neighbourhoods

Let \( X \) be a set with topology \( \mathcal{T} \). A
*neighbourhood* of \( x \) is an open subset of \( X \)
containing \( x \).

This is the definition according to Munkres. Tom Leinster gives a different definition that distinguishes between a 'neighbourhood' and an 'open neighbourhood'.

*Neighbourhood* is the closest we get to the metric space idea of open balls around
a point.

The term *neighbourhood* packs a noun-verb pair into a noun, which
I think is part of why using it makes it easier to conceptualize compared to
'open set containing \( x \)'.

The next lemma is of interest because it is a *lemma* expressed in the
terms of topology that is in some sense an isomorphism of a *definition*
that appeared when formulating open sets in metric spaces. When defining open sets of a topology, for our definition we took the metric space's derived result of open sets being closed under arbitrary union and finite
intersection. Now we have come to a derived result of topological spaces that
matches the founding definition of open sets in metric spaces. Isn't that an
interesting loop.

### Lemma. Open iff all elements have a neighbourhood subset

Let \( X \) be a topological space and \( U \subseteq X \). Then \( U \) is open in \( X \) iff for all \( x \in U \), there is a neighbourhood of \( x \) contained in \( U \).

For comparison, the definition of open sets for a metric space is copied below:

#### Metric space: Open subsets, definition

Let \( X \) be a metric space and \( U \subseteq X \). Then \( U \) is open in \( X \) iff for all \( u \in U \), there is a \( \varepsilon > 0 \) such that \( B(u, \varepsilon) \subseteq U \).