Methods & Reproducibility

Research Status Taxonomy

• Concept: an initial research direction or engineering proposal that has not yet been formalized.
• Theory: a defined mathematical or conceptual model with stated assumptions.
• Simulation: numerical evaluation of a model under specified parameters and boundary conditions.
• Prototype: a software or hardware implementation used to test a limited function.
• Preprint: a publicly available manuscript that has not completed formal publication review.
• Under review: a manuscript or chapter submitted to a publication process.
• Published: work released through a publisher, journal, or edited volume.
• Experimentally validated: a result supported by reproducible physical measurement. This label is used only when direct evidence exists.

Assumptions and Boundary Conditions

Every model depends on assumptions. The lab aims to identify them before presenting results, including the mathematical domain, initial conditions, parameter ranges, simplifications, omitted effects, and the point at which a model should no longer be trusted.

• State the governing equations or control rules.
• List initial conditions and parameter ranges.
• Identify approximations, idealizations, and excluded physics.
• Define stability, success, failure, and invalid-operation criteria.
• Distinguish model behavior from claims about nature.

Computational Method

Computational work should be reproducible from a recorded environment rather than only from a narrative description. When a project reaches public-release readiness, its record should include the model version, software dependencies, solver settings, random seeds where relevant, numerical tolerances, hardware assumptions, and the code or pseudocode necessary to reproduce the core result.

Minimum Computational Record

• Project and model version
• Source code or complete pseudocode
• Dependency and environment information
• Input data and preprocessing steps
• Parameter files and configuration
• Solver, convergence, and tolerance settings
• Expected outputs and representative checks
• Known failure modes and unresolved discrepancies

Validation and Falsification

A useful research claim must be vulnerable to failure. Each project should identify observations, simulations, or measurements that would weaken or reject the current model.

• Internal checks: dimensional consistency, numerical convergence, conservation tests, and limiting-case behavior.
• Comparative checks: agreement or disagreement with established models, published datasets, and independent calculations.
• Sensitivity checks: determine whether results depend excessively on a narrow parameter choice.
• Failure criteria: define the observation or result that would show the present formulation is inadequate.
• Independent replication: encourage separate implementation and report discrepancies rather than hiding them.

Data and Code Availability

Data, code, parameter files, and supporting materials are released when they are technically ready, legally shareable, and safe to publish. A project page should clearly state whether each artifact is available, planned, restricted, or not yet prepared.

When full code cannot be released, the lab should provide enough methodological detail, equations, pseudocode, and parameter information for independent evaluation of the central claim.

Revision and Corrections Policy

• Substantive changes should be dated and described.
• Preprints may be revised in response to review, replication, or newly identified errors.
• Published-work pages should distinguish the original publication from later commentary or correction.
• Withdrawn or unsupported directions should be marked archived rather than silently presented as active.
• Material errors should be corrected openly and linked to the affected record.

Conflicts of Interest and Sponsorship

Funding does not determine scientific conclusions. Sponsored work should disclose the sponsor, the funded scope, and any relevant restrictions. Supporters do not receive authority to suppress unfavorable findings or alter the reported result outside a separate lawful agreement governing proprietary research.

Replication and Technical Feedback

Researchers submitting feedback should identify the relevant publication, equation, dataset, software version, parameter choice, or experimental condition. Replication reports are especially useful when they include enough information to reproduce both agreement and disagreement.

Submit technical feedback or a replication inquiry

Current Method Records

• Plasma Resonator Phase Stabilisation research overview
• Nudimmud Physics program
• AI-Assisted Research Systems
• Publications & Peer Review

---

For current programs and evidence status, visit Active Research Programs and Publications & Peer Review.