Resolution Horizon – Finding the mathematical limit where AI overfits to noise

bjoern_janson1 pts0 comments

GitHub - bjoern-janson/resolution-horizon: A framework for measuring the limits of invariant discovery under bounded observation. · GitHub

/" data-turbo-transient="true" />

Skip to content

Search or jump to...

Search code, repositories, users, issues, pull requests...

-->

Search

Clear

Search syntax tips

Provide feedback

--><br>We read every piece of feedback, and take your input very seriously.

Include my email address so I can be contacted

Cancel

Submit feedback

Saved searches

Use saved searches to filter your results more quickly

-->

Name

Query

To see all available qualifiers, see our documentation.

Cancel

Create saved search

Sign in

/;ref_cta:Sign up;ref_loc:header logged out"}"<br>Sign up

Appearance settings

Resetting focus

You signed in with another tab or window. Reload to refresh your session.<br>You signed out in another tab or window. Reload to refresh your session.<br>You switched accounts on another tab or window. Reload to refresh your session.

Dismiss alert

{{ message }}

bjoern-janson

resolution-horizon

Public

Notifications<br>You must be signed in to change notification settings

Fork

Star

main

BranchesTags

Go to file

CodeOpen more actions menu

Folders and files<br>NameNameLast commit message<br>Last commit date<br>Latest commit

History<br>17 Commits<br>17 Commits

docs

docs

.gitignore

.gitignore

LICENSE

LICENSE

README.md

README.md

demo.ipynb

demo.ipynb

experiment.py

experiment.py

instrument.py

instrument.py

test_suite.py

test_suite.py

View all files

Repository files navigation

Resolution Horizon

Measuring the limits of invariant discovery under bounded observation.

Overview

Resolution Horizon is an experimental research framework for studying how much mathematical structure can be recovered from finite, noisy observations of dynamical systems.

Rather than treating discoverable structure as an intrinsic property of a system, the framework proposes that structural recovery is jointly determined by the dynamics, the observer, and the available computational resources.

The central hypothesis is that every observation process possesses a measurable resolution horizon ($k^*$): a finite level of structural complexity beyond which additional analysis no longer reveals genuine invariants and instead begins to model observational uncertainty.

The project combines differential geometry, dynamical systems, statistical estimation, and information theory into a unified empirical framework for studying this boundary.

Central Hypothesis

Given finite observations of a dynamical system, increasing structural depth initially improves invariant discovery, but beyond a critical depth, estimation uncertainty dominates, producing an experimentally measurable resolution horizon.

The primary empirical quantity is the information efficiency exchange rate $\eta(k)$, balancing compression value against structural ambiguity:

$$\eta(k) = \frac{\Delta L(k)}{C_G(k)} = \frac{L(\mathcal{D}_N \mid \text{Baseline}) - L\left(\mathcal{D}_N \ \Big\vert\ \frac{\text{Lie}_k(\hat{F})}{\hat{G}^{(k)}}\right)}{L(\theta_G) + \log N\left(\delta, \ \hat{G}^{(k)}(\epsilon_k) \cap B_R, \ d_{C^{k+1}}\right)}$$

The optimal discovery horizon is defined as:<br>$$k^* = \arg\max_k \eta(k)$$

The Resource Response Surface

The horizon is an emergent response surface shaped by data and environmental constraints:<br>$$k^* = f\left(N, \sigma, E_{\mathrm{cov}}, \lambda_{\min}(\Gamma)\right)$$

Symbol<br>Meaning<br>Predicted Sign

$N$<br>Number of observations<br>$\frac{\partial k^*}{\partial N} &gt; 0$

$\sigma$<br>Observation noise<br>$\frac{\partial k^*}{\partial \sigma} &lt; 0$

$E_{\mathrm{cov}}$<br>State-space coverage deficit<br>$\frac{\partial k^*}{\partial E_{\mathrm{cov}}} &lt; 0$

$\lambda_{\min}(\Gamma)$<br>Excitation/intervention diversity<br>$\frac{\partial k^*}{\partial \lambda_{\min}(\Gamma)} &gt; 0$

These directional predictions form the primary falsifiable claims of the project.

The Three-Regime Phase Footprint

The interaction between the twin informational curves produces a distinctive coupled geometric signature across algebraic depth $k$:

Regime I (Discovery): $\Delta L(k) \uparrow$ and $C_G(k) \downarrow$. Structural compression rises efficiently while coordinate ambiguity contracts.

Regime II (Horizon): $\eta(k)$ achieves its global maximum. This is the optimal structural resolution boundary ($k = k^*$).

Regime III (Misresolution): $\Delta L(k) \downarrow$ and $C_G(k) \uparrow$. The instrument overfits the equivalence relation itself, parsing high-order derivative noise as false background symmetry.

Repository Structure

├── instrument.py # Quantitative measurement core & metric calculators<br>├── experiment.py # Regularization Bypass protocol & causal interventions<br>├── test_suite.py # Automated topological verification harness<br>└── README.md # Project specification and deployment guide

Philosophy<br>Resolution Horizon treats scientific discovery itself as a measurable computational process. Instead of asking:

What...

horizon resolution partial frac discovery structural

Related Articles