Confirmed results come first. Everything after them is open research: maps, measurements and partial frameworks, each labelled for what it is.
Confirmed
Results checked and accepted outside this site.
Accepted on OEIS
A397240: ouroboros polyiamonds, eight new terms
A397240 counts the free ouroboros polyiamonds with 2n cells: shapes made of triangles in which every cell has exactly two neighbours. Johan Wennemyr computed eight new terms, a(15) to a(22), up to 630,379, each cross-checked by independent methods. The OEIS editors accepted them on 29 June 2026; the entry reads “a(15)-a(22) from Johan Wennemyr, Jun 26 2026”.
MIRRIM found 35729046·2639963+1323299 on 30 October 2026: 192,656 digits. It is proven prime twice, by two pieces of code that share nothing (a Brillhart–Lehmer–Selfridge proof and PFGW). Earlier the same month: a 41,205-digit Proth prime, also proven twice. These are special-form primes: their shape is what makes a proof possible. The engine’s aim is arbitrary primes.
Unfinished work. Each card says what it claims and what it does not.
Featured Research Map
The current public-facing priority is a careful Goldbach package.
It is relevant enough to show because it isolates a named analytic
obstruction; it is not presented as a solved theorem.
Open problem mapPublic-review candidateNot a proof
A Depth-3 Buchstab Minorant Map for Binary Goldbach
The Goldbach packet isolates a concrete remaining wall in a
depth-3 Buchstab route: a Type-III residual shearing problem,
sharpened to the one-R-match variance obstruction.
The value is the map: what has been reduced, what failed, and
exactly what an expert should check next.
Claim: structural reduction and obstruction map.
Not claimed: proof of Goldbach, almost-proof, or closed Type-III estimate.
Expert question: is the one-R-match variance wall structural, or can a pre-L2 identity avoid it?
Prime work is organized around bounded measurements and public-safe
theorem surfaces. The strongest current pages are the MIRRIM engine
and the Goldbach map; other prime directions remain internal or
conditional unless explicitly surfaced.
Speed measuredLast measured 14 Nov 2026
MIRRIM Prime Engine
The primes MIRRIM finds are proven (see Confirmed above). Last search run, 31 October 2026: a 41,210-digit prime found and proven inside one hour on one machine. Last measurement, 14 November 2026: one squaring of a 100-million-digit number takes 1.59 s, so a proof built from squarings alone would take about 17 years. The work now is a proof that needs fewer full-size steps. No speed run is going.
Derived & numerically verified for one solvable model (self-dual frustrated TFIM); entanglement correction is negative. NOT the quantum PCP conjecture; not peer-reviewed.