577

Source document

[2026-03-28 15:20] xenonfun: ``` [GPU] ⟨O(0)⟩ = 2.622187e5, ⟨O(0)⟩² = 6.875864e10, ⟨O²⟩ = 6.875864e10, var = 2.254227e0 [GPU] Measurement complete. 500 configs. [lattice] Done: 500 configs in 122.3s [lattice] Submitted. Batches: 19, Workers: 7 [lattice] Got job: 64³, 500 configs, seed=27, op=glueball [GPU] Thermalization: 1000 steps on 64³ = 262144 points... [GPU] Thermalization done. [GPU] Config 0/500 [GPU] Config 10/500 ``` got a few server local MacOS clients, a remote WGPU client, and 1 CUDA remote client. working out kinks in distrubuted version. this converages much better single instance but finally got it going correct directions, also that grid is too small for this sim (trying to measure ratio difference between proton and neutron here, not enough space for the quarks to do their things it works at 256^3 for sure). ``` --- Lattice QCD experiment --- POST /api/lattice/start - start lattice measurement job POST /api/lattice/poll - worker gets config (unique seed) POST /api/lattice/submit - worker submits correlator result GET /api/lattice/status - combined results + κ estimate GET /api/lattice/stream - SSE live stats stream GET /commutator/lattice - lattice experiment dashboard TLS: enabled (certs/cert.pem) ``` also neared you need TLS in order to use WebGPU on a non-local site. (files: Screenshot_2026-03-28_at_11.15.29_AM.png, Screenshot_2026-03-28_at_11.15.37_AM.png) [2026-03-28 23:05] ajaxdavis: Added lean to my genealogy project works great so far [2026-03-28 23:58] ajaxdavis: (files: Screenshot_from_2026-03-29_09-58-11.png) [2026-03-28 23:59] ajaxdavis: (files: Screenshot_from_2026-03-29_09-59-13.png) [2026-03-29 00:10] xenonfun: that's cool I hadn't even thought of that.