Follow-up notesregarding the paper
L. Kontorovich. Statistical estimation with bounded memory
- Nov 2011: another mistake found by Roi: the recurrence for $\pi_i$ in Theorem 5 is incorrect (but the calculations and conclusions are correct).
- Oct 2011: on page 4, left column, the second display should have a limsup instead of lim (the limit might not exist, as pointed out by Roi Weiss)
- Aug 2010: in the earler version, the proof of Theorem 6.1 is incorrect.