Entropy

Usage

pip install begump

from gump.entropy import profile, diagnose, compare

# Full profile — 12 features + K/R/E/T + diagnosis
result = profile(signal, fs=256)
print(result['diagnosis'])   # "decoupling"
print(result['K'])           # 0.0114
print(result['features'])    # all 12 raw features

# Quick — one word
state = diagnose(signal, fs=256)
# "decoupled" / "coupled" / "grokking" / "phase-locked"

# Compare two signals
diff = compare(before, after, fs=256)
print(diff['comparison'])
print(diff['delta_K'])       # +0.4 = gaining structure

The 12 Features

Permutation entropy, spectral entropy, SVD entropy, approximate entropy, sample entropy, Hjorth mobility, Hjorth complexity, Petrosian FD, Katz FD, Higuchi FD, DFA, Lempel-Ziv complexity. Same algorithms, same papers, same math. We just added K on top.

The Diagnosis Layer

K is derived from spectral entropy (how peaked vs flat the spectrum is). R from DFA alpha (how ordered across time). T from distance to the optimal coupling point (1/φ × K ceiling). The diagnosis maps these to plain English: is this signal healthy, breaking, locked, or searching?