This theory considers the relations between topology and systems of neighborhood filters.
The standard way of defining a topological space is by specifying a collection of sets that we consider "open" (see the Topology_ZF theory). An alternative of this approach is to define a collection of neighborhoods for each point of the space.
We define a neighborhood system as a function that takes each point \(x\in X\) and assigns it a collection of subsets of \(X\) which is called the neighborhoods of \(x\). The neighborhoods of a point \(x\) form a filter that satisfies an additional axiom that for every neighborhood \(N\) of \(x\) we can find another one \(U\) such that \(N\) is a neighborhood of every point of \(U\).
definition
\( \mathcal{M} \text{ is a neighborhood system on } X \equiv (\mathcal{M} : X\rightarrow Pow(Pow(X))) \wedge \) \( (\forall x\in X.\ (\mathcal{M} (x) \text{ is a filter on } X) \wedge (\forall N\in \mathcal{M} (x).\ x\in N \wedge (\exists U\in \mathcal{M} (x).\ \forall y\in U.\ (N\in \mathcal{M} (y)) ) )) \)
A neighborhood system on \(X\) consists of collections of subsets of \(X\).
lemma neighborhood_subset:
assumes \( \mathcal{M} \text{ is a neighborhood system on } X \) and \( x\in X \) and \( N\in \mathcal{M} (x) \)
shows \( N\subseteq X \) and \( x\in N \)proofSome sources (like Wikipedia) use a bit different definition of neighborhood systems where the \(U\) is required to be contained in \(N\). The next lemma shows that this stronger version can be recovered from our definition.
lemma neigh_def_stronger:
assumes \( \mathcal{M} \text{ is a neighborhood system on } X \) and \( x\in X \) and \( N\in \mathcal{M} (x) \)
shows \( \exists U\in \mathcal{M} (x).\ U\subseteq N \wedge (\forall y\in U.\ (N\in \mathcal{M} (y))) \)proofGiven a neighborhood system \(\{\mathcal{M}_x\}_{x\in X}\) we can define a topology on \(X\). Namely, we consider a subset of \(X\) open if \(U \in \mathcal{M}_x\) for every element \(x\) of \(U\).
The collection of sets defined as above is indeed a topology.
theorem topology_from_neighs:
assumes \( \mathcal{M} \text{ is a neighborhood system on } X \)
defines Tdef: \( T \equiv \{U\in Pow(X).\ \forall x\in U.\ U \in \mathcal{M} (x)\} \)
shows \( T \text{ is a topology } \) and \( \bigcup T = X \)proofSome sources (like Wikipedia) define the open sets generated by a neighborhood system "as those sets containing a neighborhood of each of their points". The next lemma shows that this definition is equivalent to the one we are using.
lemma topology_from_neighs1:
assumes \( \mathcal{M} \text{ is a neighborhood system on } X \)
shows \( \{U\in Pow(X).\ \forall x\in U.\ U \in \mathcal{M} (x)\} = \{U\in Pow(X).\ \forall x\in U.\ \exists V \in \mathcal{M} (x).\ V\subseteq U\} \)proofOnce we have a topology \(T\) we can define a natural neighborhood system on \(X=\bigcup T\). In this section we define such neighborhood system and prove its basic properties.
For a topology \(T\) we define a neighborhood system of \(T\) as a function that takes an \(x\in X=\bigcup T\) and assigns it the collection of supersets of open sets containing \(x\). We call that the "neighborhood system of \(T\)"
definition
\( \text{ neighborhood system of } T \equiv \{ \langle x,\{N\in Pow(\bigcup T).\ \exists U\in T.\ (x\in U \wedge U\subseteq N)\}\rangle .\ x \in \bigcup T \} \)
The way we defined the neighborhood system of \(T\) means that it is a function on \(\bigcup T\).
lemma neigh_fun:
shows \( (\text{ neighborhood system of } T): \bigcup T \rightarrow Pow(Pow(\bigcup T)) \)proofThe value of the neighborhood system of \(T\) at \(x\in \bigcup T\) is the collection of supersets of open sets containing \(x\).
lemma neigh_val:
assumes \( x\in \bigcup T \)
shows \( (\text{ neighborhood system of } T)(x) = \{N\in Pow(\bigcup T).\ \exists U\in T.\ (x\in U \wedge U\subseteq N)\} \) using assms, ZF_fun_from_tot_val1 unfolding NeighSystem_defThe next lemma shows that open sets are members of (what we will prove later to be) the natural neighborhood system on \(X=\bigcup T\).
lemma open_are_neighs:
assumes \( U\in T \), \( x\in U \)
shows \( x \in \bigcup T \) and \( U \in \{V\in Pow(\bigcup T).\ \exists U\in T.\ (x\in U \wedge U\subseteq V)\} \) using assmsAnother fact we will need is that for every \(x\in X=\bigcup T\) the neighborhoods of \(x\) form a filter
lemma neighs_is_filter:
assumes \( T \text{ is a topology } \) and \( x \in \bigcup T \)
defines Mdef: \( \mathcal{M} \equiv \text{ neighborhood system of } T \)
shows \( \mathcal{M} (x) \text{ is a filter on } (\bigcup T) \)proofThe next theorem states that the the natural neighborhood system on \(X=\bigcup T\) indeed is a neighborhood system.
theorem neigh_from_topology:
assumes \( T \text{ is a topology } \)
shows \( (\text{ neighborhood system of } T) \text{ is a neighborhood system on } (\bigcup T) \)proofAny neighborhood of an element of the closure of a subset intersects the subset.
lemma neigh_inter_nempty:
assumes \( T \text{ is a topology } \), \( A\subseteq \bigcup T \), \( x \in \text{Closure}(A,T) \) and \( N \in (\text{ neighborhood system of } T)(x) \)
shows \( N\cap A \neq 0 \)proofWe can create a topology from a neighborhood system and neighborhood system from topology. The question is then if we start from a neighborhood system, create a topology from it then create the topology's natural neighborhood system, do we get back the neighborhood system we started from? Similarly, if we start from a topology, create its neighborhood system and then create a topology from that, do we get the original topology? This section provides the affirmative answer (for now only for the first question). This means that there is a one-to-one correspondence between the set of topologies on a set and the set of abstract neighborhood systems on the set.
Each abstract neighborhood of \(x\) contains an open neighborhood of \(x\).
lemma open_nei_in_nei:
assumes \( \mathcal{M} \text{ is a neighborhood system on } X \), \( x\in X \), \( N\in \mathcal{M} (x) \)
defines Tdef: \( T \equiv \{U\in Pow(X).\ \forall x\in U.\ U \in \mathcal{M} (x)\} \)
shows \( N\in Pow(X) \) and \( \exists U\in T.\ (x\in U \wedge U\subseteq N) \)proofIn the the next theorem we show that if we start from a neighborhood system, create a topology from it, then create it's natural neighborhood system, we get back the original neighborhood system.
theorem nei_top_nei_round_trip:
assumes \( \mathcal{M} \text{ is a neighborhood system on } X \)
defines Tdef: \( T \equiv \{U\in Pow(X).\ \forall x\in U.\ U \in \mathcal{M} (x)\} \)
shows \( (\text{ neighborhood system of } T) = \mathcal{M} \)proofSome sources (like Metamath) take a somewhat different approach where instead of defining the collection of neighborhoods of a point \(x\in X\) they define a collection of neighborhoods of a subset of \(X\) (where \(X\) is the carrier of a topology \(T\) (i.e. \(X=\bigcup T\)). In this approach a neighborhood system is a function whose domain is the powerset of \(X\), i.e. the set of subsets of \(X\). The two approaches are equivalent in a sense as having a neighborhood system we can create a set neighborhood system and vice versa.
We define a set neighborhood system as a function that takes a subset \(A\) of the carrier of the topology and assigns it the collection of supersets of all open sets that contain \(A\).
definition
\( \text{ set neighborhood system of } T \) \( \equiv \{\langle A,\{N\in Pow(\bigcup T).\ \exists U\in T.\ (A\subseteq U \wedge U\subseteq N)\}\rangle .\ A\in Pow(\bigcup T)\} \)
Given a set neighborhood system we can recover the (standard) neighborhood system by taking the values of the set neighborhood system at singletons \({x}\) where \(x\in X=\bigcup T\).
lemma neigh_from_nei:
assumes \( x\in \bigcup T \)
shows \( (\text{ neighborhood system of } T)(x) = (\text{ set neighborhood system of } T)\{x\} \) using assms, ZF_fun_from_tot_val1 unfolding NeighSystem_def, SetNeighSystem_defThe set neighborhood system of \(T\) is a function mapping subsets of \(\bigcup T\) to collections of subsets of \(\bigcup T\).
lemma nei_fun:
shows \( (\text{ set neighborhood system of } T):Pow(\bigcup T) \rightarrow Pow(Pow(\bigcup T)) \)proofThe value of the set neighborhood system of \(T\) at subset \(A\) of \(\bigcup T\) is the collection of subsets \(N\) of \(\bigcup T\) for which exists an open subset \(U\subseteq N\) that contains \(A\).
lemma nei_val:
assumes \( A \subseteq \bigcup T \)
shows \( (\text{ set neighborhood system of } T)(A) = \{N\in Pow(\bigcup T).\ \exists U\in T.\ (A\subseteq U \wedge U\subseteq N)\} \) using assms, ZF_fun_from_tot_val1 unfolding SetNeighSystem_defA member of the value of the set neighborhood system of \(T\) at \(A\) is a subset of \(\bigcup T\). The interesting part is that we can show it without any assumption on \(A\).
lemma nei_val_subset:
assumes \( N \in (\text{ set neighborhood system of } T)(A) \)
shows \( A \subseteq \bigcup T \) and \( N \subseteq \bigcup T \)proofIf \(T\) is a topology, then every subset of its carrier (i.e. \(\bigcup T\)) is a (set) neighborhood of the empty set.
lemma nei_empty:
assumes \( T \text{ is a topology } \), \( N \subseteq \bigcup T \)
shows \( N \in (\text{ set neighborhood system of } T)(0) \) using assms, empty_open, nei_valIf \(T\) is a topology, then the (set) neighborhoods of a nonempty subset of \(\bigcup T\) form a filter on \(X=\bigcup T\).
theorem nei_filter:
assumes \( T \text{ is a topology } \), \( D \subseteq (\bigcup T) \), \( D\neq 0 \)
shows \( (\text{ set neighborhood system of } T)(D) \text{ is a filter on } (\bigcup T) \)proofIf \(N\) is a (set) neighborhood of \(A\) in \(T\), then exist an open set \(U\) such that \(N\) contains \(U\) which contains \(A\). This is similar to the Metamath's theorem with the same name, except that here we do not need assume that \(T\) is a topology (which is a bit worrying).
lemma neii2:
assumes \( N \in (\text{ set neighborhood system of } T)(A) \)
shows \( \exists U\in T.\ (A\subseteq U \wedge U\subseteq N) \)proofAn open set \(U\) covering a set \(A\) is a set neighborhood of \(A\).
lemma open_superset_nei:
assumes \( V\in T \), \( A\subseteq V \)
shows \( V \in (\text{ set neighborhood system of } T)(A) \)proofAn open set is a set neighborhood of itself.
corollary open_is_nei:
assumes \( V\in T \)
shows \( V \in (\text{ set neighborhood system of } T)(V) \) using assms, open_superset_neiAn open neighborhood of \(x\) is a set neighborhood of \(\{ x\}\).
corollary open_nei_singl:
assumes \( V\in T \), \( x\in V \)
shows \( V \in (\text{ set neighborhood system of } T)\{x\} \) using assms, open_superset_neiThe Cartesian product of two neighborhoods is a neighborhood in the product topology. Similar to the Metamath's theorem with the same name.
lemma neitx:
assumes \( T \text{ is a topology } \), \( S \text{ is a topology } \) and \( A \in (\text{ set neighborhood system of } T)(C) \) and \( B \in (\text{ set neighborhood system of } S)(D) \)
shows \( A\times B \in (\text{ set neighborhood system of } (T\times _tS))(C\times D) \)proofAny neighborhood of an element of the closure of a subset intersects the subset. This is practically the same as neigh_inter_nempty, just formulated in terms of set neighborhoods of singletons. Compare with Metamath's theorem with the same name.
lemma neindisj:
assumes \( T \text{ is a topology } \), \( A\subseteq \bigcup T \), \( x \in \text{Closure}(A,T) \) and \( N \in (\text{ set neighborhood system of } T)\{x\} \)
shows \( N\cap A \neq 0 \)proofassumes \( \forall x\in X.\ b(x) \in Y \)
shows \( \{\langle x,b(x)\rangle .\ x\in X\} : X\rightarrow Y \)assumes \( x\in X \)
shows \( \{\langle x,b(x)\rangle .\ x\in X\}(x)=b(x) \)assumes \( T \text{ is a topology } \)
shows \( (\bigcup T) \in T \)assumes \( T \text{ is a topology } \) and \( x \in \bigcup T \)
defines \( \mathcal{M} \equiv \text{ neighborhood system of } T \)
shows \( \mathcal{M} (x) \text{ is a filter on } (\bigcup T) \)assumes \( U\in T \), \( x\in U \)
shows \( x \in \bigcup T \) and \( U \in \{V\in Pow(\bigcup T).\ \exists U\in T.\ (x\in U \wedge U\subseteq V)\} \)assumes \( A \subseteq \bigcup T \)
shows \( cl(A) \subseteq \bigcup T \), \( cl(\bigcup T - A) = \bigcup T - int(A) \)assumes \( x\in \bigcup T \)
shows \( (\text{ neighborhood system of } T)(x) = \{N\in Pow(\bigcup T).\ \exists U\in T.\ (x\in U \wedge U\subseteq N)\} \)assumes \( A \subseteq \bigcup T \) and \( U\in T \) and \( x \in cl(A) \cap U \)
shows \( A\cap U \neq 0 \)assumes \( \mathcal{M} \text{ is a neighborhood system on } X \) and \( x\in X \) and \( N\in \mathcal{M} (x) \)
shows \( N\subseteq X \) and \( x\in N \)assumes \( \mathfrak{F} \text{ is a filter on } X \)
shows \( 0\notin \mathfrak{F} \), \( X\in \mathfrak{F} \), \( \mathfrak{F} \subseteq Pow(X) \), \( \forall A\in \mathfrak{F} .\ \forall B\in \mathfrak{F} .\ A\cap B\in \mathfrak{F} \) and \( \forall B\in \mathfrak{F} .\ \forall C\in Pow(X).\ B\subseteq C \longrightarrow C\in \mathfrak{F} \)assumes \( \mathcal{M} \text{ is a neighborhood system on } X \) and \( x\in X \) and \( N\in \mathcal{M} (x) \)
shows \( N\subseteq X \) and \( x\in N \)assumes \( x\in X \) and \( \phi (x) \)
shows \( \exists x\in X.\ \phi (x) \)assumes \( \mathcal{M} \text{ is a neighborhood system on } X \)
defines \( T \equiv \{U\in Pow(X).\ \forall x\in U.\ U \in \mathcal{M} (x)\} \)
shows \( T \text{ is a topology } \) and \( \bigcup T = X \)assumes \( T \text{ is a topology } \)
shows \( (\text{ neighborhood system of } T) \text{ is a neighborhood system on } (\bigcup T) \)assumes \( \mathcal{M} \text{ is a neighborhood system on } X \), \( x\in X \), \( N\in \mathcal{M} (x) \)
defines \( T \equiv \{U\in Pow(X).\ \forall x\in U.\ U \in \mathcal{M} (x)\} \)
shows \( N\in Pow(X) \) and \( \exists U\in T.\ (x\in U \wedge U\subseteq N) \)assumes \( f: X\rightarrow Y \), \( g: X\rightarrow Z \) and \( \forall x\in X.\ f(x) = g(x) \)
shows \( f = g \)assumes \( f:X\rightarrow Y \), \( y\in f(x) \)
shows \( x\in X \)assumes \( A \subseteq \bigcup T \)
shows \( (\text{ set neighborhood system of } T)(A) = \{N\in Pow(\bigcup T).\ \exists U\in T.\ (A\subseteq U \wedge U\subseteq N)\} \)assumes \( T \text{ is a topology } \)
shows \( 0 \in T \)assumes \( V\in T \), \( A\subseteq V \)
shows \( V \in (\text{ set neighborhood system of } T)(A) \)assumes \( N \in (\text{ set neighborhood system of } T)(A) \)
shows \( A \subseteq \bigcup T \) and \( N \subseteq \bigcup T \)assumes \( T \text{ is a topology } \), \( S \text{ is a topology } \)
shows \( (T\times _tS) \text{ is a topology } \), \( \text{ProductCollection}(T,S) \text{ is a base for } (T\times _tS) \), \( \bigcup (T\times _tS) = \bigcup T \times \bigcup S \)assumes \( T \text{ is a topology } \), \( S \text{ is a topology } \)
shows \( (T\times _tS) \text{ is a topology } \), \( \text{ProductCollection}(T,S) \text{ is a base for } (T\times _tS) \), \( \bigcup (T\times _tS) = \bigcup T \times \bigcup S \)assumes \( T \text{ is a topology } \), \( N \subseteq \bigcup T \)
shows \( N \in (\text{ set neighborhood system of } T)(0) \)assumes \( N \in (\text{ set neighborhood system of } T)(A) \)
shows \( \exists U\in T.\ (A\subseteq U \wedge U\subseteq N) \)assumes \( T \text{ is a topology } \), \( S \text{ is a topology } \) and \( U\in T \), \( V\in S \)
shows \( U\times V \in \text{ProductTopology}(T,S) \)assumes \( N \in (\text{ set neighborhood system of } T)(A) \)
shows \( A \subseteq \bigcup T \) and \( N \subseteq \bigcup T \)assumes \( T \text{ is a topology } \), \( S \text{ is a topology } \)
shows \( (T\times _tS) \text{ is a topology } \), \( \text{ProductCollection}(T,S) \text{ is a base for } (T\times _tS) \), \( \bigcup (T\times _tS) = \bigcup T \times \bigcup S \)assumes \( T \text{ is a topology } \), \( D \subseteq (\bigcup T) \), \( D\neq 0 \)
shows \( (\text{ set neighborhood system of } T)(D) \text{ is a filter on } (\bigcup T) \)assumes \( x\in \bigcup T \)
shows \( (\text{ neighborhood system of } T)(x) = (\text{ set neighborhood system of } T)\{x\} \)assumes \( T \text{ is a topology } \), \( A\subseteq \bigcup T \), \( x \in \text{Closure}(A,T) \) and \( N \in (\text{ neighborhood system of } T)(x) \)
shows \( N\cap A \neq 0 \)