Zulip Chat Archive
Stream: Is there code for X?
Topic: Discrete subsets of euclidean spaces
Michael Stoll (Jul 09 2024 at 17:59):
How do I express the statement that a subset of a euclidean space (or, more generally, a topological space) is disrcete?
Riccardo Brasca (Jul 09 2024 at 18:07):
cc @Xavier Roblot
Michael Stoll (Jul 09 2024 at 19:05):
Perhaps like this?
import Mathlib.Topology.Order
def IsDiscrete {X} [TopologicalSpace X] (S : Set X) : Prop :=
DiscreteTopology S
Oliver Nash (Jul 09 2024 at 19:22):
Yes, you probably want docs#DiscreteTopology
However for a subset of a topological space, there are intrinsic and extrinsic notions of discreteness. Copying my own remarks from DiscreteSubset.lean
:
Given a topological space `X` together with a subset `s ⊆ X`, there are two distinct concepts of
"discreteness" which may hold. These are:
(i) Every point of `s` is isolated (i.e., the subset topology induced on `s` is the discrete
topology).
(ii) Every compact subset of `X` meets `s` only finitely often (i.e., the inclusion map `s → X`
tends to the cocompact filter along the cofinite filter on `s`).
We have a proof that these are equivalent for closed subsets (subject to mild hypotheses satisfied by Euclidean space) in docs#IsClosed.tendsto_coe_cofinite_iff
Kevin Buzzard (Jul 09 2024 at 20:01):
For example the set of reciprocals of the positive integers is discrete in the sense that the induced topology is discrete, but any holomorphic function vanishing on this set must be identically zero.
Michael Stoll (Jul 10 2024 at 00:02):
Right, for my intended application I'll also need that the subset is closed.
Xavier Roblot (Jul 10 2024 at 09:02):
Yes, docs#DiscreteTopology is what is used for ℤ
-lattices, see for example docs#IsZlattice or docs#Zspan.instDiscreteTopologySubtypeMemSubmoduleIntSpanRangeCoeBasisRealOfFinite
Last updated: May 02 2025 at 03:31 UTC