Surprisal of an event is an event

... and the probability distribution of those surprisal events can be used as another model under which to calculate surprisals of events. Chain enough of these together and a fixed point is always found.

Discrete
Continuous
Chained
Distributions
Discrete · Chained Distributions

Build a chain of histograms, each node's events are the surprisals of the node above it. Explore bin edges, priors, and KL/W1 between neighbors.

Open →
Continuous · Chained Distributions

Same chaining idea, but each node fits a smooth density (KDE, adaptive KDE, Gaussian mixture, or B-spline) instead of a histogram.

Open →
Find
Fixed Point
Discrete · Find Fixed Point

Automatically iterate the surprisal-of-surprisal chain to convergence, and see where repeated runs tend to land across bins (simplex, radial, scatter-matrix & parallel-coordinates views).

Open →
Continuous · Find Fixed Point

Continuous-density version of the fixed-point search, with a convergence progression chart and an overlay of every converged density found.

Open →