Integer multiplication bound jumps ~2% in one PR as community pushes κ to 7.9347e-4

BorisMPower · x · 2026-10-11

In the open-source integer-mult-bounds project, contributors are pushing the integer multiplication error bound via swarm-style competing PRs, with the κ bound improving 2% in a single day.
Key shift: PR #346 ported "Design T" plus twin condensation to the p=10 bit word with a stronger complex-side supplier, flipping the "wall" to the bit side — the bit bound (8.017e-4) now sits 1.07% above the complex cap (7.932e-4), so future gains must come from the complex side.
Latest PR #350 applies a concave frame descent to the complex program (each gate moved to an extremal frame between predecessors' join and successors' meet), keeping the same gates, R=6,253 and registers, nudging κ to 7.93469326753735e-4 (+7.65e-8 over #348). A new best bound has been submitted and entirely new designs are being fleet-tested.

More from coding & agent