Intervention anchors and scientific verification in synthetic vascular predictive representations

Abstract:Complete orthogonal predictive coordinates do not by themselves bind a latent direction to a named intervention. We present a mathematical and synthetic audit motivated by vascular device-vessel suitcordance. Capacity-matched least-squares predictors were exactly equivalent under complete fixed output transforms, whereas an anchor-only observer recovered interpretations only within the span of known perturbation signatures. Six three-dimensional configurations across 64 seeds gave a maximum paired prediction discrepancy of 6.7e-15 but a median untransported edit error of 1.513. Coordinate transport removed that error. Noisy and weak anchors constrained calibration stability, and changing the representation basis required recalibration or verified transport. Across 256 additional fits in dimensions 3-24, prediction equivalence persisted within 4.0e-15. We then evaluated nine deliberate runnable fault classes across 64 seeds. All 576 faulty executions completed, but each violated at least one reconstruction, prediction, delivered-edit or scope contract; all 320 valid control records passed. Repeating a faulty implementation gave exact self-agreement despite error against the separately computed simulator expectation. For one omitted-direction defect, probe coverage followed its analytic law, and rank-aware abstention protected unsupported interpretations. Scalar-noise experiments exposed both missed weak faults and excessive rejection under narrow relative tolerances. These controls provide an executable separation of prediction, semantic support and scientific acceptance. They are synthetic numerical audits, not clinical validation, neural JEPA-Anything replication, agent learning or patient treatment-effect estimation.

Artificial Intelligence (cs.AI)

arXiv:2610.11704 [cs.AI]

(or arXiv:2610.11704v1 [cs.AI] for this version)

https://doi.org/10.48550/arXiv.2610.11704 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

MANIFEST_SHA256.txt

README_reproducibility.txt

additional_controls.py

audit_expanded_independently.py

expanded_audit.py

expanded_independent_audit.json

expanded_protocol.json

figures/Figure1_audit_workflow.pdf

figures/Figure1_audit_workflow.png

figures/Figure2_rotation_controls.pdf

figures/Figure2_rotation_controls.png

figures/Figure3_anchor_span_and_noise.pdf

figures/Figure3_anchor_span_and_noise.png

figures/Figure4_conditioning_and_transfer.pdf

figures/Figure4_conditioning_and_transfer.png

figures/Figure5_dimension_stress.pdf

figures/Figure5_dimension_stress.png

figures/Figure6_verifier_contracts.pdf

figures/Figure6_verifier_contracts.png

figures/Figure7_independent_probes.pdf

figures/Figure7_independent_probes.png

figures/Figure8_tolerance_sensitivity.pdf

figures/Figure8_tolerance_sensitivity.png

independent_audit.json

protocol_addendum.json

results/additional_controls.csv

results/additional_controls_summary.json

results/anchor_conditioning.csv

results/angle_sweep.csv

results/basis_transfer.csv

results/correction_trace.json

results/dimension_stress.csv

results/expanded_summary.json

results/independent_probe_coverage.csv

results/noisy_anchors.csv

results/prediction_equivalence.csv

results/random_basis_edits.csv

results/rank_span.csv

results/summary.json

results/tolerance_sensitivity.csv

results/verifier_mutations.csv

(39 additional files not shown) You must enabled JavaScript to view entire file list.

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.