Gravitational Manipulation Theory v3.1: Closure of the Synthetic Mechanism-Verification Workflow and Final Claim Boundary

Abstract

This preprint provides the final closure report for the Gravitational Manipulation Theory v3.1 H3–H6 program. It consolidates the externally fixed pre-registration, operational-freeze, execution, replication, audit, and status-closure records into a self-contained account of the completed synthetic mechanism-verification workflow. The report preserves the complete evidence sequence, including the H3 and H5 fail-closed execution attempts, the unexecuted H4 pre-registration, the Generator Rev.H Operational Freeze, the H6 blinded pre-registration, the post-approval H6 single execution, the E7 cross-environment replication request and Windows self-replication result, and the final four-layer claim-boundary record. The registered H6 execution completed all 54 cells and six equal-power null controls, produced 54 sealed checkpoints and a terminal run seal, and yielded mechanism_verdict = PASS, scientific_run_status = VALID_RUN, and scientific_terminal_state = TRANSFER_FAILURE. The frozen transfer-success conjunction was not satisfied because uncertainty_ok and multiplicity_ok were false. TRANSFER_FAILURE is therefore the valid-run result of the frozen decision policy; it is not INVALID_RUN and was not replaced by a post-hoc threshold change or success-seeking rerun. The E7 Windows 11 / CPython 3.14.3 self-replication reproduced all 54 ordered observables bit-for-bit, with maximum absolute difference 0.0 and ordered observable vector digest 22b81f77eb0a22ad7694521b49a4c0b8ce3aad41397435d1dba3402f468bae87. This supports cross-environment equivalence for the frozen synthetic representative path. Because the replication was performed by the same researcher and did not use an independent implementation, it is explicitly classified as self-replication rather than independent replication. The final claim boundary is divided into four layers. The H6 synthetic computational/mechanism workflow is closed. Independent replication, an independent implementation, and formal third-party audit have not been completed. A physical kernel and measured-data validation have not been completed. Promotion to v4.x is not supported by the H6 evidence alone. The report introduces no new execution, scientific result, threshold, or interpretation. All numerical results, record roles, SHA-256 commitments, and DOI relationships are drawn from the externally fixed evidence records identified in the Authoritative Evidence Records section. The consolidated evidence index is available at https://doi.org/10.5281/zenodo.22161994.