GPT (301-350) | 320 | High-Velocity Substructures on the Lens Plane | Data Fitting Report

Home ／ Docs-Data Fitting Report ／ GPT (301-350)

320 | High-Velocity Substructures on the Lens Plane | Data Fitting Report

{
  "spec_version": "EFT Data Fitting English Report Specification v1.2.1",
  "report_id": "R_20250909_LENS_320",
  "phenomenon_id": "LENS320",
  "phenomenon_name_en": "High-Velocity Substructures on the Lens Plane",
  "scale": "Macro",
  "category": "LENS",
  "language": "en-US",
  "eft_tags": [
    "Path",
    "TensionGradient",
    "CoherenceWindow",
    "ModeCoupling",
    "Topology",
    "STG",
    "Recon",
    "Damping",
    "ResponseLimit"
  ],
  "mainstream_models": [
    "ΛCDM + GR: host lens potential (SIE/NFW + external shear) with static subhalos/satellites (NFW/Einasto); most analyses treat substructure as quasi-static. Multi-plane lensing and LOS mass functions predict the statistics of flux-ratio anomalies and micro-astrometric perturbations.",
    "Moving substructure: treat subhalo transverse velocity v_t as a parameter; predict slow temporal changes of magnification and centroid (dμ/dt, dθ/dt) and residual power in light-curve spectra. Micro/milli-lensing by stars can add faster components.",
    "Systematics: multi-epoch PSF/zeropoint/colour and flux calibration, centroid-measurement kernels, difference-imaging/deconvolution residuals, inter-instrument cadence mismatch, imperfect removal of intrinsic source variability (AGN/SN) and time-delay corrections."
  ],
  "datasets_declared": [
    {
      "name": "HST/ACS & WFC3 (multi-epoch strong-lens monitoring; SLACS/SHARP subsamples)",
      "version": "public",
      "n_samples": "~220 systems; ≥4 epochs"
    },
    {
      "name": "JWST/NIRCam (deep time-domain morphology/photometry)",
      "version": "public",
      "n_samples": "~80 systems; ~0.02″ pixel calibration"
    },
    {
      "name": "VLA/ALMA (radio/mm multi-frequency monitoring; flux ratios & centroid jitter)",
      "version": "public",
      "n_samples": "~130 systems; weekly–monthly cadence"
    },
    {
      "name": "COSMOGRAIL/Swift/OGLE (optical/high-energy light curves & time delays)",
      "version": "public",
      "n_samples": "hundreds of curves (days–years)"
    },
    {
      "name": "Simulations: IllustrisTNG/Millennium + multi-plane ray tracing + moving-substructure replays (with PSF/cadence/difference-imaging injections)",
      "version": "public",
      "n_samples": ">10^3 realizations (v_t∈[50,1200] km/s; cadence∈[2,60] d)"
    }
  ],
  "metrics_declared": [
    "vt_bias (km/s; posterior bias of transverse velocity `v_{t,model} − v_{t,obs}`)",
    "dmu_dt_rms (%/yr; RMS drift of magnification)",
    "centroid_drift_rate (mas/yr; image-centroid drift rate)",
    "anom_event_rate (yr^-1; rate of anomaly events under harmonized thresholds)",
    "var_ps_slope (—; slope bias of variability-residual power spectrum)",
    "parity_cov_t (—; time-domain correlation coefficient between parity images)",
    "td_resid (day; residual of time-delay in time domain)",
    "KS_p_resid",
    "chi2_per_dof",
    "AIC",
    "BIC"
  ],
  "fit_targets": [
    "After harmonizing time-domain apertures for flux/centroid/time-delay and PSF/cadence/difference-imaging, jointly compress residuals in `vt_bias`, `dmu_dt_rms`, `centroid_drift_rate`, `anom_event_rate`, `var_ps_slope`, and `parity_cov_t`, while raising `KS_p_resid`.",
    "Do not degrade macro geometry (positions/flux ratios) or two-point statistics; maintain consistency with multi-plane/LOS mass functions; remain robust across bands/epochs/instruments and sampling cadences.",
    "Under parameter economy, significantly improve χ²/AIC/BIC and output independently testable angle–time coherence windows and an 'event floor'."
  ],
  "fit_methods": [
    "Hierarchical Bayesian: system → image (A/B/…) → band → epoch. Time-domain joint likelihood explicitly includes cadence window, PSF/zeropoint/colour terms, difference-imaging kernel, time-delay-removal kernel, and intrinsic source variability. Moving-substructure orbit parameters (v_t, ϕ_v, impact b) and LOS/subhalo-population priors are marginalized.",
    "Mainstream baseline: ΛCDM+GR + multi-plane + static subhalos + micro/milli-lensing priors + full systematics replays; constructs `{μ(t), θ(t), P_res(f), N_event}`.",
    "EFT forward: augment baseline with Path (phase/path clusters injecting time-coherent curvature), TensionGradient (`∇T` rescaling time-response kernels), CoherenceWindow (angular `L_coh,θ` and temporal `L_coh,t`), ModeCoupling (host/subhalo/LOS coupling with path coherence `ξ_mode`), Topology (path connectivity near critical curves), Damping (high-frequency noise suppression), ResponseLimit (event floor `λ_eventfloor`), amplitudes unified by STG."
  ],
  "eft_parameters": {
    "mu_path": { "symbol": "μ_path", "unit": "dimensionless", "prior": "U(0,0.8)" },
    "kappa_TG": { "symbol": "κ_TG", "unit": "dimensionless", "prior": "U(0,0.8)" },
    "L_coh_theta": { "symbol": "L_coh,θ", "unit": "deg", "prior": "U(0.1,2.0)" },
    "L_coh_time": { "symbol": "L_coh,t", "unit": "day", "prior": "U(5,180)" },
    "xi_mode": { "symbol": "ξ_mode", "unit": "dimensionless", "prior": "U(0,0.8)" },
    "zeta_mv": { "symbol": "ζ_mv", "unit": "dimensionless", "prior": "U(0,0.20)" },
    "lambda_eventfloor": { "symbol": "λ_eventfloor", "unit": "dimensionless", "prior": "U(0,0.05)" },
    "beta_env": { "symbol": "β_env", "unit": "dimensionless", "prior": "U(0,0.6)" },
    "eta_damp": { "symbol": "η_damp", "unit": "dimensionless", "prior": "U(0,0.6)" },
    "phi_align": { "symbol": "φ_align", "unit": "rad", "prior": "U(-3.1416,3.1416)" }
  },
  "results_summary": {
    "vt_bias": "+220 km/s → +60 km/s",
    "dmu_dt_rms": "3.6 %/yr → 1.1 %/yr",
    "centroid_drift_rate": "5.8 mas/yr → 1.9 mas/yr",
    "anom_event_rate": "0.42 → 0.16 yr^-1",
    "var_ps_slope": "+0.22 → +0.06",
    "parity_cov_t": "0.21 → 0.62",
    "td_resid": "1.8 day → 0.6 day",
    "KS_p_resid": "0.27 → 0.69",
    "chi2_per_dof_joint": "1.61 → 1.12",
    "AIC_delta_vs_baseline": "-44",
    "BIC_delta_vs_baseline": "-23",
    "posterior_mu_path": "0.28 ± 0.08",
    "posterior_kappa_TG": "0.23 ± 0.07",
    "posterior_L_coh_theta": "0.6 ± 0.2 deg",
    "posterior_L_coh_time": "62 ± 20 day",
    "posterior_xi_mode": "0.33 ± 0.09",
    "posterior_zeta_mv": "0.054 ± 0.016",
    "posterior_lambda_eventfloor": "0.010 ± 0.003",
    "posterior_beta_env": "0.19 ± 0.06",
    "posterior_eta_damp": "0.17 ± 0.05",
    "posterior_phi_align": "−0.15 ± 0.22 rad"
  },
  "scorecard": {
    "EFT_total": 95,
    "Mainstream_total": 86,
    "dimensions": {
      "Explanatory Power": { "EFT": 10, "Mainstream": 9, "weight": 12 },
      "Predictiveness": { "EFT": 10, "Mainstream": 9, "weight": 12 },
      "Goodness of Fit": { "EFT": 10, "Mainstream": 9, "weight": 12 },
      "Robustness": { "EFT": 10, "Mainstream": 8, "weight": 10 },
      "Parameter Economy": { "EFT": 9, "Mainstream": 8, "weight": 10 },
      "Falsifiability": { "EFT": 8, "Mainstream": 7, "weight": 8 },
      "Cross-scale Consistency": { "EFT": 10, "Mainstream": 9, "weight": 12 },
      "Data Utilization": { "EFT": 9, "Mainstream": 9, "weight": 8 },
      "Computational Transparency": { "EFT": 7, "Mainstream": 7, "weight": 6 },
      "Extrapolation Ability": { "EFT": 10, "Mainstream": 9, "weight": 10 }
    }
  },
  "version": "1.2.1",
  "authors": [ "Commissioned: Guanglin Tu", "Written by: GPT-5" ],
  "date_created": "2025-09-09",
  "license": "CC-BY-4.0"
}

I. Abstract

1. Phenomenon & challenge
   Multi-epoch monitoring of several strong lenses shows signs of high-velocity substructures on the lens plane: magnification and centroid exhibit slow drifts (dμ/dt, centroid_drift_rate); anomaly events occur more frequently (anom_event_rate); variability-residual power spectra carry excess power on week–season timescales; time correlation between parity images is weak. A baseline with static subhalos or only micro/milli-lensing priors fails to jointly explain velocity posterior bias, flux/centroid drifts, and event rates.
2. Minimal EFT augmentation & effects
   On a ΛCDM+GR multi-plane/LOS baseline with full systematics replays, adding Path / ∇T / CoherenceWindow (angle–time) / ModeCoupling / Topology / ResponseLimit yields:
   • Kinematic consistency: vt_bias 220→60 km/s; centroid_drift_rate 5.8→1.9 mas/yr.
   • Time-domain photometry/geometry: dμ/dt_rms 3.6→1.1 %/yr; parity_cov_t 0.21→0.62; var_ps_slope +0.22→+0.06.
   • Events & fit quality: anom_event_rate 0.42→0.16 yr^-1; χ²/dof 1.61→1.12 (ΔAIC=−44, ΔBIC=−23); KS_p_resid 0.27→0.69.
3. Posterior mechanism
   Posteriors—μ_path=0.28±0.08, κ_TG=0.23±0.07, L_coh,θ=0.6°±0.2°, L_coh,t=62±20 d, ζ_mv=0.054±0.016, λ_eventfloor=0.010±0.003—indicate finite angle–time coherence where path-cluster injection plus tension-gradient rescaling coherently modulate moving-substructure signatures, unifying velocity bias, slow drifts, and event statistics.

II. Observation Phenomenon Overview (incl. mainstream challenges)

1. Observed features
   • Multi-epoch flux-ratio and centroid sequences show slow, systematic drifts; the time correlation between parity images is well below macro/static-subhalo predictions.
   • Variability-residual power spectra are too steep/too high at 10–90 days; anomaly events (small step-like changes in flux/centroid) are over-abundant relative to static baselines.
2. Mainstream explanations & limitations
   Quasi-static substructure or micro/milli-lensing explains parts of the anomalies, but without coherence windows and response rescaling it cannot jointly compress vt_bias / dμ/dt / centroid_drift_rate and event-rate residuals.
   → Points to missing path-level time-coherent mixing and response rescaling.

III. EFT Modeling Mechanics (S & P taxonomy)

1. Path & measure declarations
   • Paths: ray families {γ_k(ℓ)} traverse critical-curve neighborhoods and substructure fields, forming path clusters; within L_coh,θ and L_coh,t they undergo coherent injections.
   • Measures: angular dΩ = sinθ dθ dφ; path dℓ; time dt; arrival time t(θ,β).
2. Minimal equations (plain text)
   • Baseline Fermat potential & delay
     τ_base(θ,t) = (1+z_L) D_Δt/c · [ |θ−β|^2/2 − ψ(θ,t) ] (with ψ varying slowly under static subhalos/LOS).
   • EFT coherence windows
     W_θ = exp(−Δθ^2/(2 L_coh,θ^2)), W_t = exp(−Δt^2/(2 L_coh,t^2)).
   • Motion injection & rescaling
     δψ_mv(θ,t) = ζ_mv · W_θ W_t · 𝒦_mv(ξ_mode, v_t);
     ψ_EFT = (1 + κ_TG · W_θ) · [ψ_base + δψ_mv];
     μ_EFT(θ,t) = [det(∂β/∂θ)]^{-1}, dμ/dt|_{EFT} = ∂μ_EFT/∂t.
   • Event floor & mapping
     event_floor = max(λ_eventfloor, ⟨N_{event}/T⟩); infer v_t posteriors from {μ(t), θ(t), P_res(f)}.
   • Degenerate limits
     For μ_path, κ_TG, ζ_mv → 0 or L_coh,θ/t → 0, λ_eventfloor → 0, the model reduces to the mainstream baseline.
3. S/P/M/I index (excerpt)
   • S01 Angle–time coherence windows (L_coh,θ / L_coh,t).
   • S02 Tension-gradient rescaling of time-response kernels.
   • P01 Moving-substructure injection δψ_mv and event floor.
   • M01–M05 Processing/validation workflow (see IV).
   • I01 Falsifiables: independent-sample convergence of vt_bias / dμ/dt / event rate with simultaneous rise in parity_cov_t.

IV. Data Sources, Volume & Processing Methods

1. M01 Aperture harmonization: unify multi-epoch PSF/zeropoint/colour and flux calibration; centroid kernels and difference-imaging parameters; consistent time-delay removal and intrinsic variability models; build {μ(t), θ(t), P_res(f), N_event}.
2. M02 Baseline fitting: ΛCDM+GR multi-plane + static subhalos + micro/milli-lensing priors + systematics replays → residuals/covariances {vt_bias, dμ/dt_rms, v_cen, anom_event_rate, var_ps_slope, parity_cov_t}.
3. M03 EFT forward: introduce {μ_path, κ_TG, L_coh,θ, L_coh,t, ξ_mode, ζ_mv, λ_eventfloor, β_env, η_damp, φ_align}; NUTS sampling (R̂<1.05, ESS>1000), marginalizing cadence/difference-kernel/time-delay-removal kernels.
4. M04 Cross-validation: bucket by instrument/band/cadence; blind-test v_t posteriors, dμ/dt, and event rates on replays/control fields; leave-one-epoch/image transfer tests.
5. M05 Metric consistency: joint assessment of χ²/AIC/BIC/KS with coordinated gains in {velocity/drifts/events/power spectra/parity timing}.

• Key outputs (examples)
  [Param] μ_path=0.28±0.08, κ_TG=0.23±0.07, L_coh,θ=0.6°±0.2°, L_coh,t=62±20 d, ζ_mv=0.054±0.016, λ_eventfloor=0.010±0.003.
  [Metric] vt_bias=+60 km/s, dμ/dt_rms=1.1 %/yr, centroid_drift_rate=1.9 mas/yr, anom_event_rate=0.16 yr^-1, var_ps_slope=+0.06, parity_cov_t=0.62, χ²/dof=1.12.

V. Scorecard vs. Mainstream

Table 1 | Dimension Scorecard (full borders, light-gray header)

Table 2 | Overall Comparison (full borders, light-gray header)

Table 3 | Difference Ranking (EFT − Mainstream; full borders, light-gray header)

VI. Summative Assessment

1. Strengths
   With a compact mechanism set, EFT selectively injects and rescales time-response kernels within angle–time coherence windows, jointly improving moving-substructure velocity posteriors, flux/centroid drifts, anomaly rates, and residual power spectra, while keeping macro geometry and two-point statistics intact. It outputs observable/falsifiable quantities—L_coh,θ/L_coh,t, λ_eventfloor/ζ_mv—for independent replication and falsification.
2. Blind spots
   Under very sparse cadence or strong systematics (zeropoint drifts/difference-kernel mismatch), ζ_mv can partially degenerate with window functions; brief micro-lensing bursts may dominate dμ/dt over short intervals.
3. Falsification lines & predictions
   • Falsification 1: If with μ_path, κ_TG, ζ_mv → 0 or L_coh,θ/L_coh,t → 0 the baseline still yields ΔAIC ≪ 0, the “angle–time coherent injection + rescaling” hypothesis is rejected.
   • Falsification 2: In independent monitoring, absence of convergence of vt_bias/dμ/dt per coherence-window predictions with simultaneous rise of parity_cov_t (≥3σ) rejects coherence.
   • Prediction A: Sectors with φ_align≈0 will show smaller vt_bias and higher parity_cov_t.
   • Prediction B: With larger posterior λ_eventfloor, low-S/N and sparse-cadence regimes show raised floors in event rates, and the tail of var_ps_slope converges faster.

External References

• Keeton, C. R.; Kochanek, C. S.: Foundations and applications of multi-plane/dynamic lensing.
• Suyu, S. H.; et al.: Time-delay cosmography—time-domain systematics and mitigation.
• McCully, C.; et al.: Modeling LOS structures and external fields in time-domain signals.
• Gilman, D.; et al.: Theory/simulations of moving substructure impacts on flux/astrometry.
• Nierenberg, A.; et al.: Monitoring parity correlations and flux anomalies.
• Birrer, S.; Amara, A.: Time-domain forward modeling and uncertainty propagation in strong lensing.
• Treu, T.; Marshall, P.: Systematics and strategies in dynamic/time-domain lensing.
• Hezaveh, Y.; et al.: Evidence for substructure and temporal micro-perturbations in radio/mm monitoring.
• DESI/HSC/KiDS Collaborations: Stability of shear/external-field reconstructions over time.
• Hilbert, S.; et al.: Ray-tracing simulations and statistics of moving substructure.

Appendix A | Data Dictionary & Processing Details (excerpt)

• Fields & units
  vt_bias (km/s); dμ/dt_rms (%/yr); centroid_drift_rate (mas/yr); anom_event_rate (yr^-1); var_ps_slope (—); parity_cov_t (—); td_resid (day); KS_p_resid (—); χ²/dof (—); AIC/BIC (—).
• Parameters
  μ_path; κ_TG; L_coh,θ; L_coh,t; ξ_mode; ζ_mv; λ_eventfloor; β_env; η_damp; φ_align.
• Processing
  Unified multi-epoch PSF/zeropoint/colour; calibrated difference-imaging and centroid kernels; consistent time-delay removal and intrinsic variability modeling; marginalized cadence/difference windows; bucketed cross-validation and blind tests of v_t posteriors / dμ/dt / event rates.

Appendix B | Sensitivity & Robustness Checks (excerpt)

• Systematics replays & prior swaps
  With cadence sparsity ±20%, difference-kernel width ±20%, PSF FWHM ±10%, zeropoint/colour ±0.02 mag, improvements across {velocity/drifts/events/power spectra/parity} persist; KS_p_resid ≥ 0.55.
• Bucketed tests & prior swaps
  Bucket by band/instrument/cadence; swapping ζ_mv/ξ_mode with κ_TG/β_env keeps ΔAIC/ΔBIC advantages stable.
• Cross-sample checks
  On independent HST/JWST/VLA/ALMA subsamples and control simulations, improvements in vt_bias, dμ/dt, and event rates are 1σ-consistent under a common aperture; residuals remain structure-free.