README¶
Sustainable Factory¶
Modeling sustainable factory processes with Linked Data and MyST Markdown.
Overview¶
This project parses industrial process descriptions (stored as MyST Markdown and JSON chat exports) and converts them into a structured graph linked data representation using the Industrial Ontologies Foundry (IOF) core schema.
Research and Evidence Workflow¶
Chat-derived physics and engineering work follows an explicit evidence
boundary. Use the research evidence triage skill
to search and cite source chats, then use the Signals evidence modeling skill
to decide whether a result belongs in verified Lean Signals, Signals.Pending,
the process/RDF catalog, or an exclusions list. The chat physics catalog
shows the current applications, processes, products, evidence status, and
recommended model updates. See AGENTS.md for project-wide
boundaries and validation commands.
Features¶
MyST Markdown & JSON Parser: Extracts process steps, properties, equipment, materials, cost figures, metrics, and source citations.
Linked Data
RDFS
ValueFlows (VF), Industrial Ontologies Foundry (IOF), Elementary Multiperspective Material Ontology (EMMO)
schematool(src/schematool) generates a directory of precedent RDFS schema inschema/
Mermaid Integration: Visualizes process flows directly from documentation.
Sphinx Documentation: High-quality HTML output with integrated diagrams.
Installation¶
Prerequisites: Python 3.11+.
# Install the package and its dependencies in editable mode
pip install -e .
CLI Reference (sfcli)¶
The project provides the sfcli tool line for data extraction and conversion:
Parse a file: Read a Markdown or JSON file to verify parsed processes to the console.
sfcli parse <input_file>
Export to RDF: Parse a file and export the output as a
.ttlscript.sfcli export_rdf <input_file> --output process_data.ttl
Batch Export: Batch process an entire directory of MyST and JSON files into a consolidated graph representation.
sfcli batch_export data/chats/ --output all_processes.ttl
Info: Display package version information.
sfcli info
Search Indexing (docindex)¶
The project includes an indexer and search library (docindex) under src/ to index sphinx documentation and chat logs:
Default backend:
oxirs(RDF/SPARQL usingpyoxigraphstore).Supported backends:
oxirs,milli(Meilisearch).Multi-backend: Index to both backends simultaneously using
--backend oxirs,milli.
Usage examples:
# Index chats using default OxiRS backend
docindex index-chats --source data/chats/
# Index Sphinx HTML atomically
docindex index-html --source docs/_build/html --backend oxirs,milli
Directory Structure¶
Here is a brief overview of the top-level directories in this workspace:
data/: Raw input data, including interactive chat exports (Markdown and JSON) describing sustainable processes.docs/: Sphinx documentation source files, configuration, and generated HTML builds.schema/: Foundational Industrial Ontologies Foundry (IOF) RDF schemas, ontologies, and the coresustainable-factory.ttlschema.src/: Supplementary source code utilities likeschematoolandtransform_md.sustainablefactory/: The core Python package providing parsing logic, knowledge graph generation, and RDF extraction features.tests/: Unit testing suite usingpytest.tools/: Standalone helper scripts, such as data aggregation utilities and RDF-to-Mermaid conversions.
Development Tasks¶
A Makefile is included to streamline common tasks:
make test/make pytest: Run unit tests and generate a coverage report.make build: Perform full pipeline generation (aggregates data, transforms markdown to docs, and builds HTML pages).make docs: Builds HTML documentation using Sphinx and MyST-parser.make aggregate_data: Runs data aggregation tools to process technical features.make serve: Serves the documentation using a local Python HTTP server.make signals_build: Builds the Lean 4Signals,SignalsTests,SignalsPending, andSignalsPendingTeststargets insrc/signals/.
Formal Verification & Simulation (src/signals/)¶
The src/signals/ package and its companion Rust/WASM/WebGL2 crate src/signals/numerical_adapter/ formalize and test our unified Complex Grassmannian \(Gr_{\mathbb{C}}(2,4)\) / \(Gr_{\mathbb{C}}(2,N+2)\) and Dilatant Dark Fluid (DDF) mechanics (both without splats and with Normed 2D/3D Anisotropic Tensor Gaussian Splats) across forty-three companion papers in docs/. Reported error scores are listed below; “not reported” means the roadmap gives no numerical error score for that paper:
-
Reported error score: 0.00% relative error for the complex phase-locked models; positive-real baseline: 50% on-axis and 100% at dark fringes.
-
Reported error score: Zero residual for valid 3-point Casorati-Wronskian solutions; no empirical prediction-error metric is reported.
-
Reported error score: 0% for the complex Grassmannian crossing model; positive-real baseline: 50% on-axis and 100% at dark fringes.
-
Reported error score: 0.00% for revised Turner 1PN and 2.5PN models in the stated solar-system and sub-critical regimes; strong-field corrections are conditional.
-
Reported error score: No aggregate numerical error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No aggregate numerical error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No aggregate numerical error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: 0.00% for revised Model 6 and Models 14a/14b in the stated phase-locked, solar-system, and sub-critical regimes; conditional corrections can be nonzero.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No aggregate numerical error score is reported in the roadmap.
-
Reported error score: WTMAD-2 = 1.643 kcal/mol on GMTKN55 (an energy-error metric, not a percentage).
Tuning Materials with Light (Discovered Materials Grand Challenge 04): Time-Linear Grassmannian Generalized Kadanoff–Baym Ansatz (
G-GKBA) and Coupled Electron–Exciton–Phonon–Bogoliubov Dynamics Across Hubbard Chains/Ladders, \(\mathrm{WSe}_2\), Graphite, and \(\mathrm{MgB}_2\) in Lean 4, Python, and Rust (docs/tuning-materials-with-light-plan.md)Reported error score: No aggregate prediction-error score is reported in the roadmap; 1.44 × 10^8 is a speedup, not an error score.
Predicting Heat Transport across Solid Interfaces (Discovered Materials Grand Challenge 05): Grassmannian Thermal Boundary Conductance (
G-TBC), Bond-Stiffness Plücker Transmission, Anharmonic \(\mathrm{Gr}_{\mathbb{C}}(2,5)\) Up-Conversion, and Transferable Semiconductor/Carbide/Polymer Benchmarks in Lean 4, Python, and Rust (docs/interfacial-heat-transport-plan.md)Reported error score: 1.17% overall MAPE across 7 datasets.
A Predictive Theory of Cuprate Superconductivity (Discovered Materials Grand Challenge 01): Grassmannian Zhang–Rice Plücker Pairing (
G-CSC), \(d_{x^2-y^2}\) Dome, Pseudogap Fermi Arcs, Planckian Strange Metal, and 12-Target Benchmarks Across \(\mathrm{LSCO}\), \(\mathrm{Bi2212}\), \(\mathrm{YBCO}\), \(\mathrm{Tl2201}\), and \(\mathrm{Hg1201}\) in Lean 4, Python, and Rust (docs/cuprate-superconductivity-plan.md)Reported error score: 0.79% overall MAPE across 12 targets.
Interfacial Electrical Properties from Processing History (Discovered Materials Grand Challenge 02): Grassmannian Process-to-Electrical Modeling (
G-IEP), Plabic 3-Flux Deal–Grove Kinetics, McWeeny Dangling-Bond Passivation, and 5-Interface Benchmarks Across Planar \(\mathrm{Si/SiO_2}\), Curved \(\mathrm{Si}\) Nanowires, \(\mathrm{SiC/SiO_2}\), \(\mathrm{HfO_2/SiO_2/Si}\), and \(\mathrm{Ge/SiGe}\) in Lean 4, Python, and Rust (docs/interfacial-electrical-properties-plan.md)Reported error score: 1.20% overall MAPE across 5 interfaces.
-
Reported error score: 0.64% overall MAPE across 8 target families.
-
Reported error score: 0.50% relative error for the ZrSiS B^(2/3) Landau-level scaling benchmark (Dirac baseline: 28.94%; Schrödinger baseline: 101.01%).
-
Reported error score: 0% Mercury perihelion error for Cachazo-Skinner and Turner at solar-system scales; Witten conformal-twistor baseline: 100%.
-
Reported error score: No numerical prediction-error score is reported in the roadmap; DRC reports 0 cross-analysis findings.
-
Reported error score: No numerical prediction-error score is reported in the roadmap; DRC reports 0 findings.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: No numerical prediction-error score is reported in the roadmap.
-
Reported error score: 99.99% exact token reconstruction on OpenWebText and 100.00% on Sudoku for the NCM multiscale VQ-VAE tokenizer (
arXiv:2610.14162); DRC reports 0 cross-analysis findings.
-
Reported error score: 0.00% Pauli-blocked interband absorption loss (\(\alpha = 0\) for \(2E_F > \hbar\omega\)), 0.1% speckle contrast (\(C^2 = 10^{-6}\) at \(100\text{ MHz}\)), and 99.8% mask fidelity; DRC reports 0 cross-analysis findings.
-
Reported error score: 0.00 V PEM electrolyzer voltage mismatch, < 1% McWeeny-passivated self-discharge leakage factor (\(3402/625000 \approx 0.54\%\)), and \(10^{-13} < 10^{-12}\text{ cm}^2/\text{s}\) polyiodide shuttle diffusivity; DRC reports 0 cross-analysis findings.
-
Reported error score: 0.00 charge-neutrality residual in
PECcoacervation, exact physiological \(\mathrm{pH} = 7.4\) bimodal \(\mathrm{Mg}\) buffering, and \(92\%\) fused-quartz optical transmittance; DRC reports 0 cross-analysis findings.
-
Reported error score: 0.00 residual on 4-stage Plabic/LGV-to-Complex-VRC relations (\(V K^\top = 0\)) across the \(x_X = -1\) resonance, \(< 10^{-12}\) unitary and idempotency error in PySCF-free Cayley–McWeeny orbital optimization (\(9\times\text{–}144\times\) speedup), and \(10{,}000\times\text{–}810{,}000\times\) 4-index \(g\)-tensor compression in
MADNESSMRA-PNO.
-
Reported error score: \(< 10^{-14}\) expectation and wavefunction discrepancy vs. Tequila’s reference symbolic backend, 0.00 imaginary drift on Hermitian Pauli expectations, \(2.2\times\text{–}61.7\times\) speedup over
qulacson dense \(2^n\) circuits up to \(n = 28\) (\(4.00\text{ GiB}\)), and \(68{,}253\times\) speedup on particle-conservingHCB-SPAcircuits scaling up to \(n = 126\) qubits (\(1.82\text{ ms}\)).