Worked Examples
Sustainable Catalyst Research Lab: End-to-End Sample Workflow
Complete an all-feature retrofit demonstration in Sustainable Catalyst Research Lab.
Sustainable Catalyst Research Lab · How-to Guide
Governed scientific calculators, computation, datasets, runs, provenance, methods, experiments, calibration, queues, instruments, and publication packages.
Purpose
This end-to-end workflow demonstrates every documented Sustainable Catalyst Research Lab capability through one fictional resilient-community-retrofit case. The objective is not to prove a real-world conclusion; it is to verify the complete product workflow with controlled data.
Prerequisites
- The current product release
- A new synthetic workspace
- The all-feature CSV and JSON
- No production or private information
- Time to review every checkpoint
Sample data
This article uses a fictional Resilient Community Retrofit Program. The files contain one demonstration row for every documented feature, so each capability can be tested without private or production data.
Download CSV feature examples · Download JSON workspace example
Preview
| # | Feature | Sample action | Expected result | Classification |
|---|---|---|---|---|
| 1 | Scientific Calculators | calculate a heat-transfer coefficient with units | a method-labeled result | synthetic-demo |
| 2 | Python Compute Core | run summary statistics through the supported backend | a typed compute response | synthetic-demo |
| 3 | Numerical Solvers | fit a simple energy model and inspect convergence | solver output and diagnostics | synthetic-demo |
| 4 | Visualization | plot measured versus predicted energy use | a figure tied to dataset and run IDs | synthetic-demo |
| 5 | Projects and Workspaces | create LAB-RETROFIT-001 | a persistent research workspace | synthetic-demo |
The downloadable files include all 20 feature rows.
Procedure
Scenario
A community center is assessing energy, comfort, financing, implementation, evidence, and monitoring needs. Use only the supplied fictional records.
- 1. Scientific Calculators: calculate a heat-transfer coefficient with units. Confirm a method-labeled result.
- 2. Python Compute Core: run summary statistics through the supported backend. Confirm a typed compute response.
- 3. Numerical Solvers: fit a simple energy model and inspect convergence. Confirm solver output and diagnostics.
- 4. Visualization: plot measured versus predicted energy use. Confirm a figure tied to dataset and run IDs.
- 5. Projects and Workspaces: create LAB-RETROFIT-001. Confirm a persistent research workspace.
- 6. Dataset Registry: register monthly meter data with schema, source, license, and units. Confirm a versioned dataset record.
- 7. Reproducible Runs: execute the baseline model twice with fixed inputs and seed. Confirm a rerunnable run record and matching fingerprint.
- 8. Evidence and Provenance: trace a savings estimate to source files, method, and code. Confirm an end-to-end provenance graph.
- 9. Method Review: document assumptions, diagnostics, limits, and reviewer decision. Confirm a governed method record.
- 10. External Discovery: locate candidate weather-normalization data. Confirm discovery records awaiting review.
- 11. Experiment Framework: define hypothesis, factors, controls, outcomes, and stopping rule. Confirm an experiment specification.
- 12. Parameter Studies: sweep occupancy and tariff assumptions. Confirm a parameter-result table.
- 13. Design of Experiments: create a factorial temperature-by-occupancy design. Confirm a design matrix and analysis plan.
- 14. Calibration and Validation: calibrate a thermal model and test held-out data. Confirm fit and validation metrics.
- 15. Distributed Compute: dispatch 500 approved scenario combinations. Confirm worker receipts and consolidated results.
- 16. Persistent Queue: queue a long calibration run and survive restart. Confirm a durable job record.
- 17. Workflow Orchestration: connect dataset, method, run, review, and export. Confirm a typed execution history.
- 18. Instrumentation: register a temperature logger, channels, calibration, and custody. Confirm instrument and observation records.
- 19. Offline Field Work: collect temperatures without connectivity and synchronize later. Confirm queued observations with conflict handling.
- 20. Publication Packages: assemble methods, data, code, figures, results, and provenance. Confirm a portable reproduction bundle.
Save and review
- Save the final workspace under the sample workspace ID.
- Close and reopen it in a clean session.
- Compare every feature result with the sample expected_result field.
- Document unresolved differences and do not hide warnings.
- Export one complete artifact bundle.
- Send one supported handoff to an integration and verify its receipt.
Supported integrations
- Workbench
- Catalyst Data
- Decision Studio
- Knowledge Library
- Site Intelligence
- Platform Core
Expected output
- a method-labeled result
- a typed compute response
- solver output and diagnostics
- a figure tied to dataset and run IDs
- a persistent research workspace
- a versioned dataset record
- a rerunnable run record and matching fingerprint
- an end-to-end provenance graph
- a governed method record
- discovery records awaiting review
- an experiment specification
- a parameter-result table
- a design matrix and analysis plan
- fit and validation metrics
- worker receipts and consolidated results
- a durable job record
- a typed execution history
- instrument and observation records
- queued observations with conflict handling
- a portable reproduction bundle
- A complete saved workspace
- A portable export
- A validated cross-product handoff receipt
Verification
- Every CSV feature row has a corresponding saved result.
- The JSON workspace ID and product version are preserved.
- The export can be inspected without temporary browser state.
- The receiving product identifies source, destination, schema, artifact, and review state.
- A human reviewer can reproduce or audit the demonstration.
Responsible use
The Lab supports research and teaching, not unsupported scientific, engineering, clinical, or regulatory claims.
Replace synthetic data only with information you are authorized to use, after reassessing licensing, privacy, uncertainty, and review obligations.
