The Theory of Persistence
Mathematical atlas

PT mathematics / tool

tool

Persistence calculator

Directly manipulating the GFT partition between entropy and persistent information.

Plain

The idea

The calculator makes GFT tactile: choose a distribution and see the total budget split between persistent structure and uncertainty.

When the distribution is uniform, everything is entropy. When it is concentrated, persistence rises.

GFT calculator

P = [0.40, 0.30, 0.20, 0.10] D_KL H log₂(m) = D_KL + H
Standard

Standard reading

The tool computes $H(P)$, $D_{KL}(P\|U_m)$, and $\log_2(m)$ for a normalized discrete distribution.

It visually shows that the sum does not move: shifting probability mass does not change the total budget, only its partition.

Takeaways

  • A simple tool to understand the fundamental principle.
  • Uniform: maximal entropy, zero persistence.
  • Concentrated: strong persistence, low entropy.
Technical

Technical formulation

Computations use base 2: $H=-\sum p_i\log_2 p_i$ and $D_{KL}=\sum p_i\log_2(p_i/(1/m))$. Zeros are ignored in logarithmic sums.

This page illustrates the GFT identity; it depends on no fit and no empirical data.

Monograph: ch04_gft.

Formulas

$H(P)=-\sum_i p_i\log_2 p_i$
$D_{KL}(P\|U_m)=\sum_i p_i\log_2(mp_i)$
$\log_2(m)=D_{KL}+H$
public code

Code and scripts

The links below point to public resources or planned GitHub repositories. No local working path is exposed to the reader.

GFT identity

The same calculation as the widget, but as a reproducible script.

Expected: residual < 1e-12

View script idle
tool

Manipulate GFT

Move the sliders: the sum stays equal to the total budget.

log₂(m)
2.000
D_KL
0.000
H
2.000