We seek stability results on the Steinhaus filtration. Below is the current strategy:
- Construct a correspondence between two Steinhaus filtrations ,
- Show that and are both -simplicial
- Conclude that and induce an -interleaving on and (see 2024-09-06 proposition 2).
- Show that and are -tame.
- Conclude that the bottleneck distance is bounded above by (see 2024-09-06 theorem 1).
Consider two Steinhaus filtrations and given by
where is the subcomplex of formed by including all simplices whose Steinhaus distance
is bounded above by .
Based on the current -interleaving result, we likely want where
i.e., we want to find a correspondence such that and are both -simplicial whenever . So for any simplex , every finite subset should be a simplex in and for any simplex , every finite subset should be a simplex in .
Borrowing from Persistence stability for geometric complexes, define the distortion of a correspondence to be
i.e., the largest difference in Steinhaus distance for each simplex. We then define the pseudometric (does the triangle inequality hold?)
TODO: Figure out what exactly measures here.
Assume that . Then there is a correspondence such . So for any simplex and simplex ,
By the reverse triangle-inequality:
i.e., . So is -simplicial.
We claim that if . Notice that
So for any simplex and ,
By an identical argument we get
i.e., . So is -simplicial.
Therefore, and are both -simplicial. Hence, we get that for some correspondence the induced maps and form -interleaving between and .
Prp
Let and be Steinhaus filtrations with , then the persistence modules and are -interleaved.
Claim: If and are both finite with the same cardinality such that , then
We want to show that there is a correspondence such that . It is shown in the paper that
for all index sets . If we can construct a correspondence so that for all , then
Therefore,
TODO: Figure out what exactly the relationship between and is in the paper so we can verify that is a correspondence.
Question: Is ?