1 paper · 1 filter
Brian Ransom
Historically, proofs of BPI in models without choice have relied on a contradiction framework that was introduced by Halpern. We introduce the filter extension property…