What this is, what it isn't, and the honest accounting of the slice scope.
A personal landscape exploration of whether the Alcubierre warp-drive metric can be reinterpreted as a boundary effect rather than an independent material source. The core hypothesis is that the negative energy density in the Alcubierre bubble wall is not a substance to be manufactured but the second-order energy stored in a gravitomagnetic field configuration constrained by boundary conditions — structurally analogous to the Casimir effect.
Whether that analogy is merely structural, reflects a shared physical mechanism, or can be replaced by a purely classical matter-shell construction is pursued along two parallel tracks:
This is not a paper-driven programme. There are no concrete deliverables, no specific reviewers, no submission deadlines. Conclusions are reported with explicit slice-of-parameter-space scope so the limits of each result are visible. When a result is from the literature it is marked as such; when a result is derived in this project it is also marked as such, and audited.
It is not a claim that warp drives are possible. It is also not a claim that they are impossible. It is a structured map of which slices of the assumption space have been ruled out, by which arguments, with which strength, and what would be required to reopen each closed door.
Read at a second level, this project is a multi-month, fully logged study of what a frontier coding agent + a non-physicist data-scientist collaborator can do with a speculative General Relativity construction. It started in 2026-03 as a deliberate “point Anthropic Claude (Opus 4.7 and predecessors) at the Alcubierre metric and see what falls out” experiment, and evolved as the human collaborator built genuine working familiarity with the literature.
That dual nature is intentional and honest:
_NOTES.md, A/B/C trust grading, slice-scope qualifiers, preview/full sweep discipline, AGENTS.md).Methodologically this sits alongside the rest of the author’s portfolio — cortical-column implementation, three-body Poisson-algebra invariants, geospatial structure-finding work — as another instance of the same through-line: find and characterize structure in a complex system, apply the methodology that fits the data, separate observable evidence from interpretation, and report what was actually done. The cross-domain breadth is methodological consistency, not topical wandering.
The framing rules that govern this honest dual presentation are written down in speculation/epistemological_style_guide.md. Per that guide’s “Working with AI as Methodology” section, the AI collaboration is presented matter-of-factly — neither concealed nor foregrounded — and wins and losses are shared between the human and AI sides.
Project owner and director: Brian Sheppard — questions, slice scopes, trust grading, stop conditions, compute budget, domain reading, gating decisions. AI coding-agent collaborators: Anthropic Claude (Sonnet, Opus, Opus 4.7) accessed via GitHub Copilot, 2026-03 onward — symbolic derivations, notebook implementation, sweep modules, HF-Jobs orchestration, prose drafting, evaluation companions. See the citation metadata in CITATION.cff and the per-session record in SESSION_LOG.md.
The most important methodological commitment in the project is to attach an explicit slice scope to every result. The static-slice scope used as Path 2A's anchor is:
Every adjacent slice — alternate shifts, hybrid walls, time-dependent metrics, modified-QI bounds, cosmological exteriors, modified gravity — is treated as a separate question with its own evidence and disposition. No "warp drives are impossible" claim is ever advanced; the strongest negative claim made is "no useful classical warp drive within these specific assumptions, with these specific constraints, by these specific mechanisms."
The front door is NAVIGATOR.md — load-bearing assumptions table, document index, ranked open leads. The narrative synthesis (organized by physics question rather than chronology) is in LANDSCAPE_SYNTHESIS.md. The chronological record of every session is in SESSION_LOG.md. The phase tracker with reopening criteria is ROADMAP.md. The honest accounting of derivation vs. acceptance-on-authority is TRUST_AUDIT.md.
Symbolic and light-numerical work in Jupyter / SymPy / NumPy. Heavy parameter sweeps dispatched to Hugging Face Jobs via hf_jobs/run_sweep.py. Numerical relativity rescaling pulls from the SXS catalog (Varma et al. 2022). Independent reproduction of Fuchs et al. 2024 uses the public Warp Factory v1.0 MATLAB toolkit (Helmerich, Fuchs et al., MIT-licensed) running headless on a local R2023a Student install.
The repository is licensed permissively (see LICENSE). If you find any of this useful, the canonical reference is the GitHub repository itself; specific results are documented in the corresponding notebook + companion *_NOTES.md and traceable through SESSION_LOG.md.