Artificial Physics: Evolutionary Emergence of Quantum Structures in Resource-Constrained DSL Competition

We present a computational framework for understanding the emergence of quantum-like structures through evolutionary competition of Domain-Specific Languages (DSLs) under resource constraints. Using dynamic task evaluation---where graph size N ~ U(3,10) and steps k ~ U(2,6) vary randomly---we prevent scalar DSLs from "memorizing" fixed solutions. Our experiments with N=100 population and 10 runs show that Matrix DSLs achieve 100% dominance within 3.8 ± 1.7 generations (95% CI: [2.6, 5.0]), providing evidence for the Substrate Hypothesis. We further investigate spontaneous emergence of matrix operations from scalar-only DSLs across 500 generations (Experiment 2A), finding that matrix operations never emerged (0/5 runs) despite scalar DSLs achieving only 60--80% success on dynamic tasks. When matrix operations are injected (Experiment 2B), success rates jump from 70--80% to 100% (+24.0% ± 5.5%), demonstrating the decisive advantage of matrix structure for generalization.