Speeding up recurrence quantification
asinghvi17 opened this issue · comments
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.
The first thing to look at is which parts are slower and faster. The RQA computation requires maaaaaany steps:
- calculate recurrence matrix.
- calculate histograms of vertical
- histograms of diagonals
- 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.