TALOS: Basin magnetic contrasts and the limits of Martian dynamo attribution

Abstract

The investigation TALOS is a reproducible investigation of weak magnetic signatures around major Martian impact basins. It addresses a central problem in planetary inference: a persistent pattern in an orbital field map can be measured without uniquely identifying the physical history that produced it. The research connects local magnetic contrasts, geographic prediction controls and conditional thermal experiments within a single executable archive. Its central result is a basin-associated magnetic deficit that survives the specified sensitivity tests; the origin and chronology of that deficit remain open. Source data and measurement The observational analysis starts from the published Langlais et al. (2019) and Morschhauser et al. (2014) magnetic coefficients, with their source bytes and attribution retained. Vector magnitude is integrated over spherical cells before taking a base-ten logarithm. Exact solid angles and Gauss-Legendre quadrature account for latitude-dependent cell area. The primary five-degree grid contains 2,592 cells. Hellas, Utopia, Argyre and Isidis use geometry transcribed from Frey (2008); each cap is compared with an adjacent annulus of equal continuous area. Odyssey GRS thorium and potassium and MOLA topography enter separate predictive controls. The magnetic result At 200 km above the magnetic models' reference radius, the equal-basin joint cap-minus-annulus contrast is -0.3876 dex, corresponding to a geometric-mean field ratio of 0.410. The comparison model gives 0.424. The observed contrast is more negative than all 1,999 sampled rigid rotations of the basin constellation, giving a corrected alignment rank of 0.0005 at the simulation resolution floor. Longitude-only rotations, which retain basin latitudes, give 0.0040. These ranks characterize specified spatial reference distributions; they are not probabilities of impact causation. Individual-basin evidence is uneven: Utopia's Holm-adjusted rank is 0.097. The models share Mars Global Surveyor observations, so their agreement tests model sensitivity rather than independent observational replication. Geographic and physical tests Geographic controls The geographic analysis makes a feature-identifiability problem explicit. For a full-rank set of landmark vectors, their angular-distance cosines recover Cartesian position exactly. This elementary identity motivates direct coordinate controls and spatially buffered evaluation. Basin-distance random-forest RMSE rises from 0.215 dex in random folds to 0.467 dex with a 500 km exclusion buffer and 0.554 dex at 1,000 km. In the 500 km task, adding basin distances to coordinates and surface covariates changes RMSE from 0.436 to 0.480 dex. The study therefore separates interpolation skill from evidence for a special basin mechanism. Full fold membership and held-out predictions are included. Paired thermal scenarios The physical analysis contains 2,048 paired, unfitted thermal scenarios. Forced and unforced members share their interior parameters, while an explicitly prescribed response law changes boundary heat extraction. The median additional duration below a thermal-admissibility threshold is 81 Myr, with 2.5th and 97.5th design quantiles of 0 and 880 Myr. These values summarize the chosen parameter design; they are neither posterior estimates nor inferred Martian dynamo ages. Nine unforced parameter sets with a common 3.7 Ga crossing demonstrate why a chronological anchor alone cannot select a unique interior history. Heat-deposition depth A separate spherical-mean conduction experiment deposits the same initial energy, 10^27 J, at two depths: 200 km below the surface and 200 km above the core-mantle boundary. Exact shell volumes, spherical conductances and a conservative time update close the discrete energy budget. In the deep case, the peak reduction in outward boundary power is 0.01985 TW at 155 Myr, and 20.25% of the initial heat reaches the core boundary by 400 Myr. The experiment isolates deposition depth under declared assumptions. It does not simulate an impact shock, mantle convection or magnetic induction. Contribution and research archive Scientific contribution TALOS contributes an audited chain from source products to scientific claims, continuous field observables, explicit geographic negative controls, archived spatial reference draws, paired counterfactuals and numerical conservation tests. It combines established physical and statistical methods without claiming discovery priority for impact suppression, spatial rotations or the coordinate identities. Synthetic harmonic-field benchmarks test implementation behavior under controlled conditions and are kept separate from geological calibration. Archive and reproduction The archive contains the 18-page manuscript and editable LaTeX, six figures, input data and labels, the attributed basin catalogue, derived field grids, all numerical outputs, Python source, tests, source manifests, environment records and reproduction instructions. The recorded release passed 91 numerical tests. After installing the documented dependencies, run python -m talos verify from the extracted archive root, followed by python -m talos reproduce. After dependency installation, numerical reproduction requires no API token or remote data service. The recorded environment is Python 3.12 on macOS arm64; other platforms have not been tested. Authorship, origin and rights Onur H. Evgin and the Eolisa Space Science Team are credited in the archive metadata, with Eolisa Space LLC affiliation. Software source is licensed under MIT. The new manuscript, scientific figures, derived products and explanatory documentation are licensed under CC BY 4.0. Original scientific products retain their own rights and source attribution, as detailed in LICENSES.md and licenses/DATA_ATTRIBUTION.md. The scientific-content license does not grant unrestricted rights to the Eolisa Space brand asset.