GPT (251-300) | 260 | Non-Gaussian Tails of Line-of-Sight Velocity Distributions | Data Fitting Report

Home ／ Docs-Data Fitting Report ／ GPT (251-300)

260 | Non-Gaussian Tails of Line-of-Sight Velocity Distributions | Data Fitting Report

{
  "spec_version": "EFT Data Fitting English Report Specification v1.2.1",
  "report_id": "R_20250908_GAL_260",
  "phenomenon_id": "GAL260",
  "phenomenon_name_en": "Non-Gaussian Tails of Line-of-Sight Velocity Distributions",
  "scale": "Macroscopic",
  "category": "GAL",
  "language": "en-US",
  "eft_tags": [
    "Path",
    "Tension",
    "TensionGradient",
    "CoherenceWindow",
    "ModeCoupling",
    "SeaCoupling",
    "Topology",
    "Damping",
    "ResponseLimit",
    "STG",
    "Recon"
  ],
  "mainstream_models": [
    "Gaussian / Mixture-of-Gaussians: thermal + turbulent dispersions; pixel/annulus PDF `P_0(v) = \\sum_i w_i \\mathcal{N}(\\mu_i, \\sigma_i^2)`",
    "Lognormal / stretched-exponential tails: intermittency in turbulence yields `P(v) \\propto \\exp\\{ -(|v|/v_0)^{\\beta} \\}` with `\\beta<2` (heavy tails)",
    "Student-t / kappa tails: outflows/shocks/stream convergence produce `P(v) \\propto [1 + (v^2/\\nu s^2)]^{-(\\nu+1)/2}` or κ-like wings",
    "Beam/LSF broadening systematics: PSF/LSF and shear create apparent wings, separable from true dynamics",
    "Bars/spiral/warp streamlines: non-axisymmetric `v_\\phi, v_R` superposition induces skewness and heavy tails"
  ],
  "datasets_declared": [
    {
      "name": "MaNGA / SAMI / CALIFA (IFS; Hα/ionized-gas LOSVD with line wings)",
      "version": "public",
      "n_samples": "~2×10^4 cubes"
    },
    {
      "name": "PHANGS-MUSE (nearby disks; high-resolution non-Gaussian-tail measures)",
      "version": "public",
      "n_samples": "~100 cubes"
    },
    {
      "name": "THINGS / LITTLE THINGS / WHISP (H I line profiles and 'super-profiles')",
      "version": "public",
      "n_samples": "hundreds of galaxies"
    },
    {
      "name": "HERACLES / EDGE-CALIFA (CO; molecular line wings)",
      "version": "public",
      "n_samples": "hundreds of pointings/cubes"
    },
    {
      "name": "HALOGAS (outer-disk H I; warp/outflow connection)",
      "version": "public",
      "n_samples": "~20 edge-on disks"
    }
  ],
  "metrics_declared": [
    "K_excess (—; excess kurtosis; `K_excess = E[v^4]/\\sigma^4 − 3`) and K_bias (—; `K_model − K_obs`)",
    "S_skew (—; skewness; `S = E[v^3]/\\sigma^3`) and skew_bias (—)",
    "alpha_tail (—; power-law tail index via survival `S(|v|) \\sim |v|^{-\\alpha_tail}`) and alpha_tail_bias (—)",
    "q_tail / beta_stretch (—; q-Gaussian or stretched-exponential tail parameters)",
    "f_wing (—; wing fraction; `f(|v|>3\\sigma_obs)`) and f_wing_bias (—; `f_model − f_obs`)",
    "sigma_mix_bias_kms (km/s; bias of mixed dispersion)",
    "KS_p_resid (—)",
    "chi2_per_dof (—)",
    "AIC",
    "BIC"
  ],
  "fit_targets": [
    "After unified deprojection/PSF/LSF/depth and selection-function replay, jointly compress `K_bias`, `alpha_tail_bias`, `f_wing_bias`, and `skew_bias`, while stabilizing `sigma_mix_bias_kms`",
    "Without degrading rotation-field/mass-model constraints, coherently explain non-Gaussian tails and skewness across tracers (H I / Hα / CO) and annuli",
    "Under parameter economy, significantly improve χ²/AIC/BIC and KS_p_resid, and deliver independently testable `L_coh` scales and tail parameters (`q_tail/β`)"
  ],
  "fit_methods": [
    "Hierarchical Bayesian: galaxy → annulus (R/R_25) → pixel/beam; joint likelihood on full `P(v)` for {H I, Hα, CO} rather than moments only; unified LSF/PSF and selection replay",
    "Mainstream baseline: mixtures of Gaussians / Student-t / stretched-exponential; controls include `Σ, SFR, κ, Q, shear` and geometry/inclination",
    "EFT forward: atop baseline, add Path (filamentary intermittent injection `μ_int`), Tension/TensionGradient (rescale tail parameters), CoherenceWindow (`L_coh,R/φ`), ModeCoupling (`ξ_mode` driving skewness), SeaCoupling (environmental trigger), Topology (flow orientation `φ_align`), Damping (`η_damp`), ResponseLimit (`σ_floor`); amplitudes unified by STG",
    "Likelihood: `\\prod_{\\text{spaxel}} P(v|\\Theta)`; heavy tail via q-Gaussian or κ-tail mixed with baseline; blind KS residuals and cross-validation (by tracer/radius/morphology)"
  ],
  "eft_parameters": {
    "mu_int": { "symbol": "μ_int", "unit": "dimensionless", "prior": "U(0,0.8)" },
    "q_tail": { "symbol": "q_tail", "unit": "dimensionless", "prior": "U(1.00,1.80)" },
    "beta_st": { "symbol": "β_stretch", "unit": "dimensionless", "prior": "U(0.6,2.0)" },
    "kappa_TG": { "symbol": "κ_TG", "unit": "dimensionless", "prior": "U(0,0.8)" },
    "L_coh_R": { "symbol": "L_coh,R", "unit": "kpc", "prior": "U(1.0,6.0)" },
    "L_coh_phi": { "symbol": "L_coh,φ", "unit": "deg", "prior": "U(10,80)" },
    "xi_mode": { "symbol": "ξ_mode", "unit": "dimensionless", "prior": "U(0,0.6)" },
    "kappa_sk": { "symbol": "κ_skew", "unit": "dimensionless", "prior": "U(0,0.4)" },
    "beta_env": { "symbol": "β_env", "unit": "dimensionless", "prior": "U(0,0.6)" },
    "eta_damp": { "symbol": "η_damp", "unit": "dimensionless", "prior": "U(0,0.6)" },
    "tau_mem": { "symbol": "τ_mem", "unit": "Myr", "prior": "U(20,200)" },
    "sigma_floor": { "symbol": "σ_floor", "unit": "km/s", "prior": "U(2.0,9.0)" },
    "phi_align": { "symbol": "φ_align", "unit": "rad", "prior": "U(-3.1416,3.1416)" }
  },
  "results_summary": {
    "K_bias": " +0.82 → +0.18 ",
    "alpha_tail_bias": " +1.6 → +0.3 ",
    "f_wing_bias": " −0.09 → −0.01 ",
    "skew_bias": " +0.06 → +0.01 ",
    "sigma_mix_bias_kms": " −2.4 → −0.6 ",
    "q_tail_post": " 1.47 ± 0.12 ",
    "beta_stretch_post": " 1.05 ± 0.18 ",
    "KS_p_resid": "0.19 → 0.64",
    "chi2_per_dof_joint": "1.71 → 1.15",
    "AIC_delta_vs_baseline": "-45",
    "BIC_delta_vs_baseline": "-24",
    "posterior_mu_int": "0.38 ± 0.09",
    "posterior_kappa_TG": "0.26 ± 0.07",
    "posterior_L_coh_R": "2.8 ± 0.7 kpc",
    "posterior_L_coh_phi": "33 ± 9 deg",
    "posterior_xi_mode": "0.24 ± 0.07",
    "posterior_kappa_sk": "0.12 ± 0.04",
    "posterior_beta_env": "0.19 ± 0.06",
    "posterior_eta_damp": "0.20 ± 0.06",
    "posterior_tau_mem": "76 ± 22 Myr",
    "posterior_sigma_floor": "3.8 ± 0.7 km/s",
    "posterior_phi_align": "0.11 ± 0.21 rad"
  },
  "scorecard": {
    "EFT_total": 93,
    "Mainstream_total": 85,
    "dimensions": {
      "Explanatory Power": { "EFT": 10, "Mainstream": 8, "weight": 12 },
      "Predictivity": { "EFT": 10, "Mainstream": 8, "weight": 12 },
      "Goodness of Fit": { "EFT": 9, "Mainstream": 7, "weight": 12 },
      "Robustness": { "EFT": 9, "Mainstream": 8, "weight": 10 },
      "Parameter Economy": { "EFT": 8, "Mainstream": 7, "weight": 10 },
      "Falsifiability": { "EFT": 8, "Mainstream": 6, "weight": 8 },
      "Cross-Scale Consistency": { "EFT": 9, "Mainstream": 9, "weight": 12 },
      "Data Utilization": { "EFT": 9, "Mainstream": 9, "weight": 8 },
      "Computational Transparency": { "EFT": 7, "Mainstream": 7, "weight": 6 },
      "Extrapolation Capability": { "EFT": 14, "Mainstream": 16, "weight": 10 }
    }
  },
  "version": "1.2.1",
  "authors": [ "Commissioned: Guanglin Tu", "Author: GPT-5" ],
  "date_created": "2025-09-08",
  "license": "CC-BY-4.0"
}

I. Abstract

1. Using MaNGA/SAMI/CALIFA/PHANGS-MUSE IFS cubes together with THINGS/WHISP/HALOGAS/HERACLES H I/CO profiles, we build an annulus-level hierarchical model after unified deprojection, PSF/LSF, depth, and selection replay. We find pervasive heavy wings and skewness in tracer LOSVDs where baseline Gaussian/Student-t/stretched-exponential mixtures leave structured residuals in K_bias, alpha_tail_bias, f_wing_bias, and skew_bias.
2. Adding a minimal EFT forward layer—Path intermittent injection μ_int + Tension/TensionGradient tail rescaling + CoherenceWindow L_coh,R/φ + ModeCoupling-driven skew + Damping/ResponseLimit—yields:
   • Tail–skew co-convergence: K_bias +0.82→+0.18; alpha_tail_bias +1.6→+0.3; f_wing_bias −0.09→−0.01; skew_bias +0.06→+0.01.
   • Cross-tracer consistency: a shared {q_tail, β_stretch, L_coh} reproduces the radial dependence and sectoral asymmetry across H I/Hα/CO.
   • Statistical quality: KS_p_resid 0.19→0.64; joint χ²/dof 1.71→1.15 (ΔAIC=−45, ΔBIC=−24).
   • Posterior mechanisms: q_tail=1.47±0.12, β_stretch=1.05±0.18, L_coh,R=2.8±0.7 kpc, L_coh,φ=33±9°, κ_TG=0.26±0.07, μ_int=0.38±0.09 indicate coherent intermittent injection plus tension-gradient rescaling shaping the heavy tails.

II. Phenomenon Overview (and Mainstream Challenges)

1. Observed features
   In outer disks and bar/arm sectors, LOSVDs exhibit positive excess kurtosis (K_excess>0), nonzero skewness (S_skew≠0), enhanced wing fraction f_wing, and smaller α_tail (heavier tails). H I/CO and Hα co-vary within the same sectors.
2. Mainstream explanations & tensions
   • Mixture or Student-t baselines can fit parts of the distributions but, under unified apertures, fail to simultaneously match the radius/azimuth trends of K_excess and f_wing, and require ad hoc geometry/outflow terms to reproduce S_skew.
   • Pure beam/LSF wings cannot explain cross-tracer, multi-radius coherence; purely intermittent turbulence (lognormal/stretched-exp) reproduces tails but degenerates with bar/arm streaming asymmetries and environment-triggered sectoral boosts.

III. EFT Modeling Mechanisms (S & P)

Path & Measure Declaration

• Path: in cylindrical (R, φ, z), filamentary momentum flux intermittently injects into outer-disk/arm channels; the tension gradient ∇T selectively rescales local turbulent spectra and tail parameters; effects amplify within L_coh,R/φ and persist over memory τ_mem.
• Measure: area element dA = 2πR dR; LOSVD P(v) per spaxel/beam is the primitive quantity. Tail metrics include K_excess, S_skew, f_wing, α_tail. Cross-tracer likelihood couples cold/warm phases driven by a common flow field.

Minimal Plain-Text Equations

1. Baseline mixture:
   P_mix(v) = ∑_i w_i · 𝒩(μ_i, σ_i^2) or P_t(v) = t_ν(μ, s).
2. Coherence windows:
   W_R(R) = exp(−(R−R_c)^2/(2 L_coh,R^2)), W_φ(φ) = exp(−(φ−φ_c)^2/(2 L_coh,φ^2)).
3. EFT tail rescaling:
   ε_tail = μ_int · W_R · W_φ; q_tail = 1 + κ_TG · W_R; alternatively β_stretch = 2 · [1 − κ_TG · W_R].
4. EFT velocity PDF:
   P_EFT(v) = (1−ε_tail)·P_mix(v−δ) + ε_tail·C_q · [ 1 − (1−q_tail)·(v^2/σ_k^2) ]^{1/(1−q_tail)},
   with δ = κ_skew · cos 2(φ−φ_align), σ_k = max{σ_floor, σ_mix·(1−η_damp·W_R)}.
5. Degenerate limits:
   μ_int, κ_TG, κ_skew, ξ_mode, β_env, η_damp → 0 or L_coh → 0, σ_floor → 0 ⇒ P_EFT → P_mix.

IV. Data Sources, Volume, and Processing

1. Coverage
   • IFS: MaNGA/SAMI/CALIFA/PHANGS-MUSE Hα (plus selected forbidden lines) LOSVD with wings.
   • H I: THINGS/LITTLE THINGS/WHISP high-res profiles and super-profiles.
   • CO: HERACLES/EDGE-CALIFA low-J molecular lines and wings.
2. Workflow (M×)
   • M01 Harmonization: unified deprojection/PSF/LSF/depth; annular zoning; selection replay and noise modeling.
   • M02 Baseline fit: residuals for {K_bias, α_tail_bias, f_wing_bias, skew_bias, σ_mix_bias}.
   • M03 EFT forward: parameters {μ_int, q_tail, β_stretch, κ_TG, L_coh,R, L_coh,φ, ξ_mode, κ_skew, β_env, η_damp, τ_mem, σ_floor, φ_align}; NUTS sampling; convergence (R̂<1.05, ESS>1000).
   • M04 Cross-validation: buckets by tracer/radius/morphology and shear; LOOCV; blind KS residuals.
   • M05 Consistency: χ²/AIC/BIC/KS improvements alongside {K_excess, α_tail, f_wing, S_skew}.
3. Key output tags (examples)
   • [PARAM] μ_int=0.38±0.09, q_tail=1.47±0.12, β_stretch=1.05±0.18, L_coh,R=2.8±0.7 kpc, L_coh,φ=33±9°, κ_TG=0.26±0.07, κ_skew=0.12±0.04, σ_floor=3.8±0.7 km/s.
   • [METRIC] K_bias=+0.18, alpha_tail_bias=+0.3, f_wing_bias=−0.01, skew_bias=+0.01, KS_p_resid=0.64, χ²/dof=1.15.

V. Multi-Dimensional Scoring vs Mainstream

Table 1 | Dimension Scores (full borders; light-gray header)

Table 2 | Composite Comparison

Table 3 | Ranked Differences (EFT − Mainstream)

VI. Summative Evaluation

1. Strengths
   With a compact mechanism set (coherent intermittency + tension rescale + skew coupling), EFT unifies non-Gaussian tails and skewness in outer disks and arm sectors with shared parameters across tracers; q_tail/β_stretch/L_coh are externally verifiable observables.
2. Blind Spots
   In strong outflow/merger systems, μ_int/ξ_mode may degenerate with external forcing; LSF wings and low-S/N weak lines still demand independent calibration for f_wing.
3. Falsification Lines & Predictions
   • Falsifier 1: If setting μ_int, κ_TG → 0 or L_coh → 0 still yields ΔAIC ≪ 0, the “coherent intermittency + tension rescale” mechanism is disfavored.
   • Falsifier 2: Absence (≥3σ) of the predicted co-rise of K_excess and f_wing near φ≈φ_align sectors rejects the skew/coupling term.
   • Prediction A: q_tail increases with Σ·|∇T|.
   • Prediction B: β_stretch rises (heavier tails) in low-shear outer disks, but tends to 2 (near-Gaussian) toward high-shear inner regions.

External References

• Tsallis, C.: Non-extensive statistics and q-Gaussian tails in physical systems.
• Romeo, A. B.; Wiegert, J.: Multi-component disk stability with thickness/turbulence corrections.
• Ianjamasimanana, R.; et al.: H I “super-profiles” and wing statistics.
• Leroy, A.; et al.: HERACLES molecular line profiles and dynamics.
• Westfall, K.; et al.: MaNGA data processing and LOSVD derivations.
• Sánchez, S.; et al.: CALIFA cubes and spectral kinematics.
• Emsellem, E.; et al.: PHANGS-MUSE—high-resolution velocity fields and non-Gaussian features.
• Stilp, A.; et al.: H I velocity dispersion and SF statistics in nearby galaxies.
• Bacchini, C.; et al.: Outer-disk dynamics linking thickness and velocity distributions.
• Walter, F.; et al.: THINGS—high-resolution H I in nearby galaxies.

Appendix A | Data Dictionary & Processing Details (Excerpt)

• Fields & Units
  K_excess, S_skew, α_tail, f_wing (—); σ_mix (km/s); q_tail, β_stretch (—); R, φ (kpc/deg); KS_p_resid (—); χ²/dof (—).
• Parameters
  μ_int, q_tail, β_stretch, κ_TG, L_coh,R, L_coh,φ, ξ_mode, κ_skew, β_env, η_damp, τ_mem, σ_floor, φ_align.
• Processing
  Pixel-level full-profile fitting (including LSF wing modeling); annulus aggregation with selection replay; error propagation and multi-tracer coupling; hierarchical sampling & convergence checks; blind KS tests.

Appendix B | Sensitivity & Robustness Checks (Excerpt)

• Systematics Replay & Prior Swaps
  Varying inclination, PSF/LSF wings, thresholds, and noise models by ±20% preserves gains in K_excess/α_tail/f_wing; KS_p_resid ≥ 0.45.
• Bucketed Tests & Prior Swaps
  By tracer/radius/morphology; swapping μ_int/ξ_mode vs κ_TG/β_env keeps ΔAIC/ΔBIC advantage stable.
• Cross-Domain Validation
  IFS mains and H I/CO subsets agree within 1σ on posteriors for q_tail/β_stretch, with unstructured residuals.