Estimations of ε
#89
Estimations of ε
#89
Conversation
|
Try this example usage: using DynamicalSystems, PyPlot, RecurrenceAnalysis
R = RecurrenceMatrix(tr, 0.1; scale="powerlaw")
Rg = grayscale(R)
imshow(Rg, cmap = "binary_r", extent = (1, size(R)[1], 1, size(R)[2]))Change
|
| @@ -1,3 +1,5 @@ | |||
| using Statistics | |||
| using ChaosTools:linear_region | |||
There was a problem hiding this comment.
You can't add ChaosTools as a dependency of RecurrenceAnalysis. You can just copy the linear_region code, but I think it is probably better to move linear_region to DelayEmbeddings.jl, as it is a more basic package and the method is generally useful.
I'll do this ASAP.
|
Norbert added some comments in #67 that should be incorporated here. |
| s = resolve_scale(x, m, ε; kwargs...) | ||
| if !haskey(kwargs, :scale) | ||
| s = resolve_scale(x, x, m, ε; kwargs...) | ||
| elseif kwargs[:scale] == "powerlaw" |
There was a problem hiding this comment.
Should this be replaced with a type-based dispatch system? I find that it's immensely helpful to specify options, and clarify dispatches...
There was a problem hiding this comment.
I am not sure how helpful this will be in the end of the road, because it would make the code much larger. How do you think it would be done to keep the code the same size?
In addition, from the user perspective method = "linear" or method = "cubic" is simpler to understand than method = Linear(), method = Cubic(). Everyone knows strings, but not everyone knows singleton types and their role in multiple dispatch.
But in general I also prefer using multiple dispatch instead of unending if statements.
|
btw, I have lost interest in this issue. It feels too experimental and theoretically shallow. |
#67