Does Policy Mirror Descent Converge in Policy with Non-Decomposable Mirror Maps? A Counterexample
AI use, as the source states it
“In this note, we provide a counterexample with assistance from OpenAI's GPT-5.6 model, showing that PMD with a non-decomposable Legendre mirror map need not converge in policy.”
Model: GPT-5.6, as stated
Named here, or see a mistake? Write to privacy@papex.org.