Code accompanying our papers on the "Generative Distributional Control" framework

Official code for Fisher information embedding for node and graph learning (ICML 2023)

[ICLR 2026] Official implementation of the paper: Learning a distance measure from the information-estimation geometry of data

Fisher-Bures Adversary Graph Convolutional Networks

Research-grade PyTorch math: differential geometry, spectral graph theory, discrete Ricci flow, simplicial topology, persistent homology, cellular sheaves, SO(3) Lie primitives, information geometry, tensor decompositions, content-addressable provenance. GPU-native, batched-first, audit-clean, cited.

PyTorch implementation of α-geodesical skew divergence

Picture of the Space of Learnable Tasks (ICML 23)

Topological Manifold Control: A multi-paradigm (PyTorch/JAX) geometric morphing engine utilizing differentiable soft-gating, optimized via XLA fused kernels and autograd-isolated non-blocking pipelines.

The Golomb Universe

geometric-ai quantum-machine-learning information-geometry self-supervised-learning reinforcement-learning edge-ai numpy python

Decoder-only Transformer where attention operates on geodesic distances in a learned Riemannian manifold with gravitational curvature and variable dimensionality per token. Based on Directional Relational Manifolds (DRM)

Curriculum learning framework that uses geometrically structured datasets and lets the model self discover learning path through emergent dynamics

Trainer for manifold-valued Hidden Markov Models

Book II of Dark Geometry: the Standard Model gauge group, particle masses, couplings and mixing angles derived from d=3 and the Planck mass alone — zero free parameters. ~170 predictions, ~80 sub-percent (Koide leptons, v_H to 22 ppm, m_p/m_e=6π⁵). Verification code + master table.

Categorical model of subjective observer collapse and cognitive projections.

Universal Self-Generation Theory

Information-geometric early warning signals: KL divergence rate, Fisher-Rao distance, geodesic acceleration

Information Geometry. Monge mapa (monotone rearrangement).

Information Geometry. Geometrija prostora kombinacija — v6 walk-forward follow.

Information Geometry. Fisher na uslovnoj raspodeli (ne globalna frekvencija).