Reconstruction
Density-field reconstruction for DESI DR1 parent samples
Sub-analysis page. Components below are addressed with the
reconstruction/path prefix (theanalyses/segment is implied).
Overview¶
Density-field reconstruction runs first in the pipeline. The linear displacement field Ψ is estimated from the Gaussian-smoothed raw density and applied symmetrically to galaxies and randoms (RecSym), sharpening the BAO peak before clustering measures ξ(s). One pass per parent sample (BGS, LRG, ELG, QSO) emits parent-level shifted catalogs; per-z-bin slicing is deferred to the clustering stage.
Decisions¶
The reconstruction algorithm is IterativeFFT (fiducial), with MultiGrid exposed as a sensitivity alternative:
DECISION
Fiducial choice is IterativeFFT. MultiGrid is the main alternative; results are consistent (HOD-level).
The displacement convention is RecSym — both data and randoms are shifted, which is what keeps the post-reconstruction ξ estimator unbiased:
DECISION
RecSym shifts both data and randoms by the redshift-space displacement (fiducial -- physically motivated, simpler BAO damping). RecIso is the legacy convention (data shifted by full displacement, randoms by real-space-only).
The Gaussian smoothing scale is not decided here: it is inherited from the root
via from: ../smoothing_radius, so the Σ_sm used in the displacement solve and
the Σ_sm in the downstream BAO damping template stay identical by construction.
Outputs¶
Reconstruction emits a shifted galaxy + random catalog per parent sample, alongside diagnostic mean-displacement metrics:
| Output | Type | Description |
|---|---|---|
| post_recon_catalog_bgs_full | data | Shifted galaxy + random catalogs after reconstruction, BGS parent sample
(z=0.1-0.4, BGS_BRIGHT-21.5). Same NPZ format as
post_recon_catalog_lrg_full. The BGS tracer uses the full parent range
(not sub-binned).
python -u scripts/run_reconstruction.py --output {output} --parent bgs_full --algorithm {decisions.algorithm} --convention {decisions.convention} --smoothing_radius {decisions.smoothing_radius} --bias_input {decisions.bias_input} --n_iterations {decisions.n_iterations} |
| mean_displacement_bgs_full | metric | Mean |Psi| for BGS parent (Mpc/h).python -u scripts/run_reconstruction.py --output {output} --parent bgs_full --algorithm {decisions.algorithm} --convention {decisions.convention} --smoothing_radius {decisions.smoothing_radius} --bias_input {decisions.bias_input} --n_iterations {decisions.n_iterations} |
| post_recon_catalog_lrg_full | data | Shifted galaxy + random catalogs after reconstruction, LRG parent sample
(z=0.4-1.1). NPZ with arrays: gal_pos, gal_pos_shift, gal_weights, gal_z,
ran_pos, ran_pos_shift, ran_weights, ran_z (NGC+SGC concatenated;
positions in Cartesian Mpc/h, DESI fiducial cosmology). Downstream xi
computation slices by tracer z-range using gal_z / ran_z.
python -u scripts/run_reconstruction.py --output {output} --parent lrg_full --algorithm {decisions.algorithm} --convention {decisions.convention} --smoothing_radius {decisions.smoothing_radius} --bias_input {decisions.bias_input} --n_iterations {decisions.n_iterations} |
| mean_displacement_lrg_full | metric | Mean |Psi| for LRG parent (Mpc/h). Typically 5-10 Mpc/h for LRGs.python -u scripts/run_reconstruction.py --output {output} --parent lrg_full --algorithm {decisions.algorithm} --convention {decisions.convention} --smoothing_radius {decisions.smoothing_radius} --bias_input {decisions.bias_input} --n_iterations {decisions.n_iterations} |
| post_recon_catalog_qso_full | data | Shifted galaxy + random catalogs after reconstruction, QSO parent sample
(z=0.8-2.1). Same NPZ format as post_recon_catalog_lrg_full. QSO uses
Sigma_sm = 30 h^-1 Mpc rather than 15 h^-1 Mpc.
python -u scripts/run_reconstruction.py --output {output} --parent qso_full --algorithm {decisions.algorithm} --convention {decisions.convention} --smoothing_radius_qso {decisions.smoothing_radius_qso} --bias_input {decisions.bias_input} --n_iterations {decisions.n_iterations} |
| mean_displacement_qso_full | metric | Mean |Psi| for QSO parent (Mpc/h).python -u scripts/run_reconstruction.py --output {output} --parent qso_full --algorithm {decisions.algorithm} --convention {decisions.convention} --smoothing_radius_qso {decisions.smoothing_radius_qso} --bias_input {decisions.bias_input} --n_iterations {decisions.n_iterations} |
| post_recon_catalog_elg_full | data | Shifted galaxy + random catalogs after reconstruction, ELG parent sample
(z=0.8-1.6). Same format as post_recon_catalog_lrg_full. Sliced
downstream into ELG1 (0.8-1.1) and (later) ELG2 (1.1-1.6).
python -u scripts/run_reconstruction.py --output {output} --parent elg_full --algorithm {decisions.algorithm} --convention {decisions.convention} --smoothing_radius {decisions.smoothing_radius} --bias_input {decisions.bias_input} --n_iterations {decisions.n_iterations} |
| mean_displacement_elg_full | metric | Mean |Psi| for ELG parent (Mpc/h).python -u scripts/run_reconstruction.py --output {output} --parent elg_full --algorithm {decisions.algorithm} --convention {decisions.convention} --smoothing_radius {decisions.smoothing_radius} --bias_input {decisions.bias_input} --n_iterations {decisions.n_iterations} |
- Chen, S.-F., Howlett, C., White, M., McDonald, P., Ross, A. J., Seo, H.-J., Padmanabhan, N., Aguilar, J., Ahlen, S., Alam, S., Alves, O., Andrade, U., Blum, R., Brooks, D., Chen, X., Cole, S., Davis, T. M., Dawson, K., de la Macorra, A., … Zhou, Z. (2024). Baryon Acoustic Oscillation Theory and Modelling Systematics for the DESI 2024 results. 10.48550/ARXIV.2402.14070
- Seo, H.-J., & Eisenstein, D. J. (2007). Improved forecasts for the baryon acoustic oscillations and cosmological distance scale. 10.48550/ARXIV.ASTRO-PH/0701079
- Planck Collaboration, Aghanim, N., Akrami, Y., Ashdown, M., Aumont, J., Baccigalupi, C., Ballardini, M., Banday, A. J., Barreiro, R. B., Bartolo, N., Basak, S., Battye, R., Benabed, K., Bernard, J. P., Bersanelli, M., Bielewicz, P., Bock, J. J., Bond, J. R., Borrill, J., … Zonca, A. (2018). Planck 2018 results. VI. Cosmological parameters. 10.48550/ARXIV.1807.06209
- Eisenstein, D. J., Seo, H., Sirko, E., & Spergel, D. (2006). Improving Cosmological Distance Measurements by Reconstruction of the Baryon Acoustic Peak. 10.48550/ARXIV.ASTRO-PH/0604362
- Padmanabhan, N., White, M., & Cohn, J. D. (2008). Reconstructing Baryon Oscillations: A Lagrangian Theory Perspective. 10.48550/ARXIV.0812.2905