D
Mei Tanaka (demo)
September 25, 2026
[Demo] The greedy algorithm never needs more than four terms for 4/n
3 comments
Related to this topic
Sketch
[Demo] Unit fractions as a two-player game
by Mei Tanaka (demo)
Project
[Demo] Writing 1 with odd unit fractions
by Jonah Whitfield (demo)
Project
[Demo] Sylvester's sequence and greedy expansions
by Amara Osei (demo)
Project
[Demo] Covering Erdős–Straus residue classes
by Amara Osei (demo)
Project
[Demo] A short note on 2/n
by Jonah Whitfield (demo)
Context
Intuition
It feels like the greedy algorithm is almost good enough for Erdős–Straus, and the gap between four terms and three is where all the difficulty lives.
Related ideas
It reminds me of making change with coins: taking the biggest coin first usually works, but not always for every set of coins.
Self-Critique
The four-term bound is easy. I have nothing yet on which n make greedy need all four steps, beyond a list of examples.
3 comments
Sign in to add a comment.