W7

IFU ExGal configuration parameters

Describes version 2.0

Table 19, Table 20 and Table 21 list the binning, stellar-fitting and emission-line/index controls. W7.1 records the segmentation preparation resources.

Each IFU mode-resolution-mode-detector-binning JSON file (configs/ExGal_configs/*.json) carries one flat parameter set that configures segmentation and patching (described in the accompanying paper), extragalactic spatial binning, pPXF stellar analysis, PyAPS’s EMI implementation of pPXF for WEAVE emission-line analysis, and LS together. This page lists every key used by PyAPS, grouped by the pipeline stage it controls, with a brief description of its role. Values shown are the current production configuration unless a row is marked as run-specific; the exact JSON content for a given mode, resolution mode and detector-binning combination is recorded in the run manifest.

Table 19. Settings for pre-filtering and spatially binning extragalactic IFU patches.
Key Role Notes
VORONOI Enable adaptive Voronoi/PowerBin tessellation for ExGal rows 0/1 switch; unbinned analysis uses one bin per good spaxel instead
VORONOI_GAL Same switch for the Galactic (RVSpecFit (RVS)/FERRE) path Independent of VORONOI; Gal binning is typically left off
TARGET_SNR Target per-bin signal-to-noise for ExGal tessellation Drives how many spaxels each bin combines
TARGET_SNR_GAL Target per-bin signal-to-noise for the Galactic path Only used when VORONOI_GAL is enabled
COVAR_VOR Voronoi covariance correction factor Corrects the naive
N

noise estimate for spatially correlated noise between combined spaxels
SPBIN_SIZE_EXGAL Fixed-box spatial pre-binning factor before ExGal tessellation ≤0

disables prebinning
SPBIN_SIZE_GAL Fixed-box spatial pre-binning factor before Galactic preparation Independent of the ExGal value
PIXELSIZE Spaxel angular size (arcsec) recorded with the cube Carried through to bin-area and aperture calculations
MIN_SNR Effective per-spaxel S/N floor used before binning Derived automatically unless overridden; see the next three rows
MIN_SNR_METHOD Method used to derive MIN_SNR Default percentile: a percentile of the patch’s own spaxel S/N distribution
MIN_SNR_PERCENTILE Percentile used by MIN_SNR_METHOD=percentile Default 2nd percentile
MIN_SNR_FLOOR Lower bound on the derived MIN_SNR Prevents an unreasonably low threshold on a low-S/N field
MIN_SNR_CEILING Upper bound on the derived MIN_SNR Prevents an unreasonably high threshold on a high-S/N field
MIN_SNR_GAL Equivalent S/N floor for the Galactic path Independent of the ExGal MIN_SNR family
SB_FILTER Enable the surface-brightness/flux pre-filter 0/1 switch, applied before spatial prebinning
SB_FILTER_MODE Thresholding mode for the flux filter Default percentile
SB_FILTER_PERCENT Percentile used to set the flux-filter threshold Default 10th percentile
SB_FILTER_MIN_FRAC Minimum fraction of spaxels the filter must keep Safety bound against over-trimming a faint patch
SB_FILTER_SNR_MIN Minimum S/N a spaxel must have to count toward the filter statistics
SB_FILTER_DELTA Tolerance added to the flux-filter threshold
SB_FILTER_MIN_SPAXELS Minimum absolute spaxel count the filter must keep Second safety bound, independent of SB_FILTER_MIN_FRAC
Table 20. IFU ExGal stellar-continuum (pPXF) parameters.
Key Role Notes
SIGMA Initial guess for the stellar velocity dispersion (km s
−1

)

 Starting point for the pPXF fit
VELSCALE Log-rebinned pixel velocity scale (km s
−1

per pixel)

 None lets pPXF derive it from the data
SSP_LIB Stellar-population template library identifier E.g. popstar_total; selects the template file on disk
PCA Use the PCA-compressed template library variant Trades template completeness for fitting speed
NORM_TEMP Template normalisation convention LIGHT or MASS-weighted
LMIN_PPXF, LMAX_PPXF Rest-frame wavelength window fitted by pPXF (Å)
MOM Number of kinematic (LOSVD) moments fitted E.g. 2 (V,
σ

) or 4 (adds
h3,h4

)
ADEG Degree of the additive Legendre polynomial −1

disables the additive polynomial
MDEG Degree of the multiplicative Legendre polynomial
MC_PPXF Number of Monte Carlo realisations for pPXF uncertainties
FOR_ERRORS Whether Monte Carlo uncertainty estimation is run at all Gate on the MC_PPXF loop
REDDENING Initial guess/bounds for a fitted intrinsic reddening term Distinct from the fixed foreground correction below
EBmV Fixed foreground Galactic-extinction
E(B−V)

applied before fitting

 None/0.0 disables foreground de-reddening
SPAXEL_WEIGHTED_LSF Use the per-spaxel geometric/flux-weighted LSF for this cube Off by default; falls back to one flat global FWHM/LSF curve
LSF_N_BUCKETS Maximum number of resolution buckets for spaxel-weighted LSF Groups near-identical per-bin LSF curves to bound template preparation cost
Table 21. Settings for the pPXF emission-line path (EMI) and line-strength analysis of extragalactic IFU data.
Key Role Notes
EMI_FILE Emission-line list/configuration (YAML) Defines the fitted line groups
LMIN_EMI, LMAX_EMI Rest-frame wavelength window fitted by EMI (Å)
EMI_MOM Number of kinematic moments fitted per emission-line group
USE_STELLAR_CONTINUUM Include the pPXF stellar-continuum fit in the EMI model
FIX_STELLAR_CONTINUUM Hold that stellar continuum fixed while fitting emission lines Off lets the continuum vary jointly with the lines
MC_EMI Number of Monte Carlo realisations for EMI uncertainties
EMIPPXF_LEVEL Analysis level for the EMI step BIN or per-spaxel; also tags the output filename
LS_FILE Line-strength index-band definition file
CONV_COR Target index-measurement resolution (Å FWHM) Templates/data are convolved to this common resolution before index measurement
MC_LS Number of Monte Carlo realisations for line-strength uncertainties
NWALKER, NCHAIN MCMC walker count and chain length Used by the stellar-population (age/metallicity) fit in the LS module
LS_RESOLUTION_MODE Policy for a negative or invalid target/data resolution difference E.g. SKIP_NEGATIVE, ADAPTIVE_TARGET
LS_MODE Numeric mode flag paired with LS_RESOLUTION_MODE
LS_RES Resolution-mode label recorded in output headers/filenames E.g. ADAPTED
Table 22. Settings for continuum-based seg2d detection, optional emission-line seg3d detection, and construction of IFU patches.
Key Role Notes
seg2d_white_src Source of the white-light image for 2D segmentation E.g. weave (built from the L1 cube itself)
seg2d_search_gaia Query Gaia for stellar sources to mask
seg2d_extract Run source extraction at all False keeps only the central whole-field target
seg2d_exclude_ctarg Drop the central target once at least one real source is extracted
seg2d_mask_stars Add Gaia stars as explicit mask rows in the patch table
seg2d_ext_thresh Source-extraction detection threshold (
σ

)
seg2d_minarea Minimum source area (pixels)
seg2d_deblend_nthresh Deblending threshold levels
seg2d_radii_factor Ellipse-radius scale factor applied to detected sources
class_z_rad Aperture radius (arcsec) used to collapse a spectrum for classification Shared by seg2d- and seg3d-merged rows alike
seg3d Run the 3D matched-filter detection pipeline Enabled in all six LIFU configurations; disabled in all six mIFU configurations
seg3d_merge Merge reliable seg3d line-only candidates into the patch table Same LIFU/mIFU enablement as seg3d; requires seg2d=True
seg3d_lmin, seg3d_lmax Wavelength window searched by seg3d (Å) 3700 to 9390 for LR; 4800 to 6740 for HR
seg3d_threshold Nominal detection signal-to-noise threshold 6.0; the merge gate is calibrated separately
seg3d_spatial_bin Spatial block-binning factor applied before detection 4; boosts per-pixel S/N for the search step only
seg3d_continuum_method Continuum-removal method pca (default) or a robust running-median alternative
seg3d_ivar_calibration Inverse-variance noise-scale calibration mode E.g. per-wave
seg3d_min_npix Minimum candidate spatial extent (pixels) kept after vetoes 3
seg3d_fig Write the seg2d/seg3d comparison diagnostic figure
seg3d_merge_min_snr Floor on the merge signal-to-noise threshold 15.0
seg3d_merge_min_purity Target purity used to derive data-driven single-line and group thresholds 0.95; a route is disabled if no sufficiently sampled threshold reaches it
seg3d_merge_group_min_lines Minimum number of exact-pixel candidates in a multi-line group 3
seg3d_merge_group_min_snr Floor on quadrature-combined group S/N 15.0; field calibration can raise, not lower, this value
seg3d_merge_group_min_n_trust Minimum positive groups used to calibrate the group gate 3; lower-statistics trade-off relative to the single-line value 15
seg3d_group_radius_px Spatial friends-of-friends radius on the binned detection grid 0: exact pixel only, independent of wavelength
seg3d_merge_extreme_safety_margin Multiplier above the loudest event in the complete sign-flipped search 1.2; not a calibrated false-alarm probability
seg3d_merge_aperture_arcsec Aperture radius assigned to merged seg3d targets 2.0″

Segmentation example and preparation resources

Section C of the paper records the observing block (OB) 11299 settings, candidate accounting and their relation to the current production configuration. The original detection run used PyAPS commit 21f6d479, with grouping/extreme-S/N implementation 46175f1. The full aps_ifu_prepare run used eight preparation processes, 1917.8 s wall time, 2092.2 s user+system CPU time and 11.8 GiB peak RSS. These values include Redrock classification of the 23 continuum apertures. The combined segmentation work took less than two minutes; separate seg2d and seg3d timings are not available.