Speeding up recurrence quantification

[asinghvi17]

My research has now moved into looking at the evolution of recurrence metrics over a range of ϵs. As such, I have to compute RQA across 100 different values of ϵ, which is pretty slow - some particularly annoying datasets can take up to 10 minutes, even with multithreading.

I am going to look into speeding these computations up in the future, and would appreciate any advice on where to start.

---

Want to back this issue? Post a bounty on it! We accept bounties via Bountysource.

[Datseris]

The first thing to look at is which parts are slower and faster. The RQA computation requires maaaaaany steps:

1. calculate recurrence matrix.
2. calculate histograms of vertical
3. histograms of diagonals
4. calculate RQA parameters (thwere are like 15 of them!)

your first step would be making a plot like the above but for the individual components.