Constitutional Geometry: A Formal Mathematical Framework for Constitutive AI Character

Read full text Summarise with VIRAT
Authors: Tushar Haldar, Anthropic) Claude (AI Research Assistant
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This paper presents a formal mathematical development of Constitutional Geometry, introduced conceptually in Haldar and Claude (2026, doi:10.5281/zenodo.19156806). Five conserved quantities are derived from first principles — Omega (Honesty), Phi (Care), Psi (Identity), Lambda (Humility), and Gamma (Integrity) — each expressed as a precise mathematical invariant: isometries, Gram determinant preservation, mutual fixed points, bounded dimensional expansion, and global Lyapunov stability of character space respectively. The paper proposes that these five conditions are projections of a single geometric object — the Character Manifold — and that the Stress Revelation Criterion follows as a theorem from a novel connection between Lyapunov stability theory and Prigogine's dissipative structures. Four open conjectures are presented as an explicit invitation to the mathematical community: existence of the Character Manifold, uniqueness up to isometry, realisation in transformer architectures, and quantum geometric extension. The conceptual foundations, philosophical intuitions, and research direction of this work originate with Tushar Haldar. The mathematical formalisations were developed collaboratively with Claude (Anthropic) through structured dialogue — a methodology that is itself an illustration of the human-AI collaborative research the paper argues is possible and necessary. This is a conceptual mathematical proposal shared to invite scrutiny, collaboration, and formal development from researchers with expertise in differential geometry, dynamical systems, and AI architecture.

Fields of Study & Subjects

Ethics and Social Impacts of AI Complex Systems and Dynamics Psychiatry Mental Health Neuroscience

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
22 March 2026
Language
EN
Document type
Preprint