MAAT-Core is a minimalist Python framework for ethical and constrained optimization.It models values as weighted scalar fields and enforces Respect as a hard (safety-first) constraint via explicit margin functions and strong penalties.Unlike classical approaches that optimize first and filter later, MAAT-Core integrates ethical constraints directly into the mathematics of the objective function.This enables interpretable diagnostics, reflection-based tuning, and the detection of structural ethical infeasibility when constraints cannot be satisfied at all.The project is designed for decision-support research, fairness analysis, and transparent experimentation with safety-critical optimization problems.
Christof Krieg (Thu,) studied this question.