TY  - RPRT
TI  - On physical implementability of reinforcement-learning-driven reasoning
AU  - Eiji Konishi
PY  - 2026
UR  - https://arxiv.org/abs/2610.14131
ID  - 2610.14131
ER  -