TMBR: May 2026

Prev: April 2026

This Month in Beaver Research

Next: June 2026

This edition of TMBR is in progress and has not yet been released. Please add any notes you think may be relevant (including in the form a of a TODO with a link to any relevant Discord discussion).

BB Adjacent

General Recursive Functions: TODO: Further progress

• Some new cryptids were hand-built:
  • Size 56, by Shawn on 2 May (simulating 5x+1 problem starting at 7).^[1]
  • Size 49, by aparker, star and Shawn on 3 May (simulating Brocard's problem).^[2]
• The first non-trivial divergent GRF was found (size 15). It halts iff there exists some n ≥ 1 such that n+3 divides $Tetr(n)=\frac{n(n+1)(n+2)}{6}$.^[3] aparker^[4] and star^[5] proved that there is no such n.

SKI Calculus:

TODO

TODO: Check for other BB-like functions' progress

Misc

• mxdys made his inductive decider for history-linear machines compilable into an executable file.^[6]

Holdouts

• BB(2,5)
  • Andrew Ducharme found a FAR solution for a machine, which previously had an informal nonhalting argument. On June 1st, mxdys provided the Rocq proof, reducing the Rocq-holdout count to 65.
• BB(4,3)
  • Andrew Ducharme reduced the number of holdouts from 5,641,006 to 5,127,263, a 9.11% reduction, with mxdys's new inductive decider.^[7]
• BB(3,4)
  • XnoobSpeakable completed Stage 2 of Phase 3, reducing to number of holdouts from 12,049,358 to 11,362,197, a 5.70% reduction.
• BB(2,6)
  • Andrew Ducharme reduced the number of holdouts from 536,112 to 527,232 via Enumerate.py and TM-enum, a 1.66% reduction.^[8][9]
  • Using the new inductive decider by mxdys and TM-enum, Andrew Ducharme first reduced the number of holdouts from 527,232 to 501,914^[10][11], then to 439,120 again using the inductive decider.^[12] In total this was a 16.71% reduction.
  • Using FAR, Andrew Ducharme further reduced the number of holdouts from 439,120 to 418,127, a 4.78% reduction.
  • Using some mxdys deciders, Andrew Ducharme continued reducing the number of holdouts through May 17, from 418,127 to 413,513 (a 1.10% reduction).^{[13][14][15][16]}
• BB(2,7)
  • Terry Ligocki enumerated 130K more subtasks, increasing the number of holdouts to 1,087,732,936. A total of 350K subtasks out of the 1 million subtasks (or 35%) have been enumerated.^[17]