Zulip Chat Archive
Stream: Formal conjectures
Topic: Erdős Problem #348
Wouter van Doorn (Sep 25 2025 at 10:58):
As I mentioned on the website (see here: https://www.erdosproblems.com/forum/thread/348), it looks like the Lean formalization uses a definition of 'complete' where ALL positive integers need to be the sum of distinct elements. This version I can solve. However, Erdös and Graham called a sequence complete if all large enough integers can be written as a sum of distinct elements.
Paul Lezeau (Sep 25 2025 at 11:12):
Indeed! This should be written using IsAddComplete (from here)
Paul Lezeau (Sep 25 2025 at 11:26):
@Wouter van Doorn If you fancy taking a look:): google-deepmind/formal-conjectures#685
Wouter van Doorn (Sep 25 2025 at 11:44):
Well, I'm sorry to say, but I'm not quite the right person to ask. I've tried learning Lean, but haven't gotten very far yet unfortunately. So I'm assuming your fix is correct, but I can't confidently verify that! Hopefully someone else can :)
Moritz Firsching (Sep 25 2025 at 11:45):
No worries, we will review it ourselves. Many thanks in any case for pointing out the inaccuracy!
Last updated: Dec 20 2025 at 21:32 UTC