Research Papers · 6 Papers
Papers
Six in-revision manuscripts with current PDFs, evidence-capped readiness, and explicit publication gates. Click any paper for full details, connected surveys, predictions tested, and remaining work.
Paper Listing
A compact CQG Note deriving the minimal Einstein–Cartan–Holst axial contact interaction and the zero-spin canonical-scalar branch. It fixes conventions, separates coefficient-level statements from state-dependent observables, and provides a commit-pinned hard-cutoff NJL diagnostic. The exact v1A.0.124 confirmation board (Grok MINOR / Gemini MINOR / Claude MAJOR) truth-audited all 13 findings to 0 genuinely-new-real — P1A is CONVERGED to human gates. No version change; human CQG review, license/deposit authorization, and alternate-regulator robustness remain.
- •Human CQG significance disposition — the automated confirmation board is not journal acceptance
- •Authorize a manuscript/source license; exact bundle/proof pass but deposit metadata and any draft intentionally fail closed until then
- •After license authorization, verify a reversible draft before any immutable archive/DOI action
- •Alternate-regulator robustness beyond the declared hard-cutoff convention
- •Matched Lorentzian state/stress observable and a state-specific renormalized axial expectation value
P1B is a focused software metapaper around namaster-proof 0.1.7. The exact v2B.0.9 confirmation board (Grok REJECT / Gemini MINOR / Claude MAJOR) FALSIFIED the 'workspace tensor not reproducible' premise — it is deterministically regenerable from committed RNG-free code. v2B.0.10 closes 4 optional polish items: a new recheck script, real committed execution costs, a pip-install one-liner, and the retained macOS-untested label. Archive DOI, correspondence metadata, and human review remain open; readiness holds 56.
- •Houston-gated: immutable archive DOI (Zenodo or PyPI+Zenodo), correspondence metadata, human software review — the exact v2B.0.10 board + truth audit found P1B science exhausted with the DOI as the only remaining major; v2B.0.11 adds the pytest command + non-affiliation sentences and no further board is owed on that two-sentence diff per the content-hash stop rule
- •Obtain independent human software review and publish an immutable archive/DOI
- •Publish package 0.1.7 to an independent package index once a PyPI token is available (Houston gate) after release QA
Derives fNL = −35/16 for the stated matter-contraction background and cubic action, then maps that result conditionally to published and in-house large-scale-structure sensitivity estimates. v1.7.125 adds a computed model-specific Einstein-Cartan four-fermion torsion bound, converting assumption (f) from asserted to bounded. The observational forecasts are illustrative, not a completed detection claim.
- •Independent human scientific review and venue-specific scope/format check
- •Exact reversible draft is remotely digest-verified; Houston still controls immutable archive/DOI publication
- •Direct cubic bounce transfer (dressed-metric intermediate now regulator-independent to <1%, IC-epoch placement + quantum-mass term remain) and survey-native SPHEREx covariance/likelihood remain open; the model-specific torsion bound is now computed and bounded (v1.7.125)
- •Author arXiv endorsement and journal-submission decision
Recovers 181 public DESI DR1 TARGETIDs from a frozen historical anomaly list, split transparently into 170 high-coordinate-consistency core associations and 11 lower-confidence positional associations. r10 adds an integrity reframe: the sub-0.1-arcsec core excess is by-construction expected seed self-recovery (single-member clusters whose centroid is the seed member's own coordinates) — NOT independent association evidence. This is an archive-recovery product, not a purity, novelty, or detection claim.
- •Human ApJS/editorial review and submission decision; automated ACCEPT is not journal acceptance
- •Re-stage the Zenodo deposit at r10 (prior r8 staging does not cover r9 or r10; r9 re-staging was never completed)
- •Run the exact v3.2.0-r10 non-Anthropic + Claude-subagent confirmation review
- •A reversible exact-commit draft is verified; Houston still controls immutable archive/DOI publication and arXiv/ApJS submission
- •Object-level physical interpretation and any representative-control performance study remain separate new-science work
Classifies 8.47M galaxies. The strict safe-sample observed-label statistic remains null-consistent (z=0.63465, p=0.23768). Two real-compute pod-campaign closures land in v1.0.267: a manifest-retained ViT retrain (val_acc 0.9931) proving the exact training realization, and a training-disjoint validation (accuracy 0.9867, kappa 0.9733 on 3,000 doubly-disjoint GZ1 spirals) — presented like-for-like against, not as a replacement for, the historical kappa=0.40 human-vote figure. CE-ResNet catalog re-provisioning, complete covariance, archive, exact re-review, and human gates remain; readiness holds 80.
- •Freeze and validate full-catalog imaging-leg, depth, seeing, PSF, and redshift metadata; the exact morphology join is now public but these fields remain unavailable
- •Run and truth-audit the exact v1.0.267 confirmation board (reviews the G1 manifest-retained retrain + G2 training-disjoint validation closures computed 2026-07-17) before any DOI publication
- •Re-provision the Jia et al. 2023 CE-ResNet high-confidence spiral catalog (pre_desi.fits; external source, GitHub h3jia/galaxy_spin_classifier / NADC China-VO) to engage the CE-non-spiral 826-vs-846 sub-conflict and complete the full historical-realization training component; released Catalog C labels are unaffected
- •Extend the computed 3x3 joint covariance to the MASTER-decoupled leg and full joint likelihood; run the G4 monopole-mechanism injection (H200, now unblocked by the completed G1 retrain); derive the morphology transfer function needed to convert A_95^obs into a physical parity-amplitude bound
- •Author arXiv endorsement and journal-submission decision
On the released DESIVAST GALZONE parent, the focal exploratory analysis detects no void/non-void difference in classifier-labelled CW fraction; this bounded non-detection does not establish physical environment independence.
- •Run exact-PDF confirmation review on v0.1.141 and keep the sparse interaction and secondary T-Web limitations explicit
- •Finalize Paper IV labels, weights, and provenance, then reverify P5 against that independently reviewable release
- •Publish an immutable P5 tag/archive/DOI and verify every A1–A48 link; local retention is not a public release
- •Resolve the human/editorial gate through actual AJ review; automated evidence is not journal acceptance