On the power of multitask representation learning in linear MDP

Published in Arxiv-preprint, 2021