Network-gyroid mesh convergence
Three OpenFOAM meshes, pressure-drop sequence, apparent order and GCI with the under-resolution boundary retained.
Open record →Research index · Current programme
The next major goal is empirical closure.
Featured technical result
Under-resolution is information, not an embarrassment.
The gyroid study identifies exactly why one more refinement campaign is more valuable than a stronger claim.
564,291 → 1,960,469 cells · ΔP 10.661 → 11.399 Pa
Fine-grid GCI 19.945%; under-resolved, standalone and not a measurement.Open evidence →
Research agenda
Six evidence-closing programmes.
The agenda favours the next experiment that removes the most consequential uncertainty.
Seal exact scaffold truth
Re-run the connected gyroid through current governed OpenFOAM, retain post-mesh geometry authority, close conservative transport and add canonical FEniCSx mechanics for the same candidate.
Benchmark the right surrogate family
Compare scalar baselines, POD-ROM, Gaussian processes, true 3D FNO/PINO, geometry-informed operators and mesh-graph models on a leakage-safe real-truth corpus.
Close Nano’s sim-to-real gap
Acquire blinded rig trajectories and fields, retain hardware configuration and estimate residuals across carrier, fluid and actuation regimes.
Shadow Quantum Energy live
Observe forecast error, workload availability, scheduler actions and settlement outcomes without granting dispatch authority.
Admit one measured discovery loop
Bind candidates, controls, assay protocol, replicates, raw observations and model update in one reviewed Quantum Materials record.
Build Cell’s lifecycle protein catalogue
Define stage-specific proteins with abundance, timing, media context, cell-line scope and controls; then calibrate observation operators and reconcile Cell-F1/F2 predictions with laboratory traces.
Technical record index · 2026
Results, manifests and traces.
A publication-style index replaces a generic blog grid.
MicroCarrierBench v0.1
Four-task internal in-silico comparison of magnetic, acoustic and passive reference carriers.
Open record →12,000-job energy scale study
Twelve deterministic public-trace slices with exclusions, confidence interval and non-causal hindsight bound.
Open record →Evidence-aware autonomous loop
Origin-main fixture proving evidence-linked batch change and physical-claim blocking.
Open record →Cell-F1 seven-day reference
Full public request and run with 174 snapshots, population state and absolute-unit environment fields.
Open JSON →Public figure provenance
Source repository revisions, upstream file blobs, parameters, output hashes and evidence boundaries.
Open JSON →