Pragma Research

Physics and intelligence.

We study how bounded, self-reading systems can settle into shared states, and what that idea makes possible in physics and AI.

The preprints.

Read the technical accounts on PhilPapers.

Physics

Observer Patch Holography.

Local patches hold state and records, compare what they share through their boundaries, and repair disagreements. The physics preprint develops the mathematical structures that follow under stated assumptions.

AI

Cadence.

Cadence applies the patch-net idea to learning machines. It is a novel real-time AI modeled after animal and human brains, and it learns from live experience. The preprint describes the architecture and the computational hypothesis behind it.

For the formal definitions and executable models, explore the physics and proofs repository or the Cadence source.

The blog

Ideas in plain language.

Explanations and perspectives on the physics, mathematics and learning systems behind the work.