E47 Research WorkBench

CARRIER DIMENSION

125 states

V₂ ⊗ V₂ ⊗ V₂

SELECTED SUBSPACE

47 states

J = 2 ⊕ J = 5

EXACT RANK FRACTION

0.376

47 / 125 · dimension ratio

COMPLEMENT RADIUS

0.882353

15 / 17 · optimal step

Experiment setup

E2 MODEL

Evolution Initial state distribution Iterations / sample count 220

Filter step ε

Inspect / edit sector weights

Weights are normalized on each run. Editing selects Custom.

K = (C − 6I)(C − 30I)
G = I − εK²
P = projection onto ker K

Convergence to the selected subspace

Complement norm and its rigorous decay bound

DISCRETE

● Observed error┄ Decay boundLOG SCALE

FINAL ERROR—

CONDITIONAL OCCUPANCY—

FILTER SUCCESS PROBABILITY—

Normalized selected occupancy

Ω(t)

47/125 is a rank fraction. A particular state's occupancy depends on its initial weights.

Casimir sector census

7 SECTORS

Selected sectors at C = 6 and C = 30 account for 25 + 22 states.

Inspectable samples

↑ Import run

Sample	Scaled time τ	Error	Bound	Occupancy	Probability / norm²

Canonical 47-face crystal

Facet edges Auto rotate

Drag to rotate · click a facet to inspect · arrow keys rotate · selected face highlighted.

Geometry inspector

Facet

The halfspace construction is geometrical. Its spin labels are an assigned correspondence.

Face degree distribution

Face adjacency spectrum

A single near-zero eigenvalue certifies graph connectivity within numerical tolerance.

Reconstruction ledger

Generated by Python from src/e47 canonical operators. This browser displays the saved checks.

Check	Residual	Tolerance	Result

Comparison CSV: parameter provenance open

The original rows are retained without alteration. Bismuth and woven-system entries are reduced models, not material measurements.

Source provenance

Archived inputs are hash identified. Repairs live in the workbench adapter; original input bytes remain inspectable.

Source	SHA-256

PYTHONPATH=src python -m e47.workbench regenerates the bundle and matrix validations.

RESEARCH CONTINUITY

One experiment. Inspectable records.

Keep code in GitHub, research context in Notion, documents in Google Workspace, and next actions in Linear. These are verified navigation destinations; live account synchronization is not enabled in this app.

Portable experiment handoff

Export JSON for exact replay, CSV for spreadsheet analysis, and Markdown for research notes. Imported receipts are validated and recomputed locally. Imported files cannot execute code.

Experiments stay in browser memory until exported. No workspace credentials or private document contents are bundled.

JSON receipt → reproduce
CSV samples → inspect
Research note → document
Linear issue → next action