Pith. sign in

REVIEW 3 major objections 5 minor 68 references

A two-channel deep network — fed both frequency-differenced and PCA-cleaned maps — keeps the 21-cm cross-power spectrum unbiased at large scales under a realistic cosine beam, where either preprocessing alone loses 5–8% or more.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-03 21:26 UTC pith:JXHZPEOO

load-bearing objection The hybrid FD+PCA UNet is a real improvement under the cosine beam, but the abstract's mismatched-beam robustness claim has no experiment behind it. the 3 major comments →

arxiv 2511.15072 v3 pith:JXHZPEOO submitted 2025-11-19 astro-ph.CO

Deep learning with hybrid frequency differencing and principal component analysis for 21-cm foreground and beam mitigation

Zitong Wang , Feng Shi , Le Zhang , Yanming Liu , Xiaoping Li , Shulei Ni , Ming Jiang , Xiaofan Ma This is my paper
classification astro-ph.CO
keywords 21-cm cosmologyintensity mappingforeground removalbeam effectsdeep learningU-Netfrequency differencingPCA cleaning
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper sets out to show that the two standard ways of suppressing smooth foregrounds in 21-cm intensity mapping — spectral differencing between adjacent frequency channels, and subtracting the leading principal components of the data cube — preserve different parts of the cosmological signal, and that a U-shaped convolutional network fed both as separate input channels outperforms a network fed either one. In simulations with realistic foregrounds and a frequency-dependent cosine beam, each single-channel network underestimates the cross-correlation power spectrum by 5–8% on large scales (k < 0.1 h Mpc^-1) and by more than 20% near k = 0.2 h Mpc^-1, while the two-channel network stays consistent with unity within 1σ. The payoff, if correct, is that large-scale modes needed for baryon acoustic oscillation measurements survive a realistic chromatic beam without the signal loss that aggressive linear cleaning causes.

Core claim

The central claim is that frequency differencing and PCA are not competing preprocessing options but complementary views: differencing preserves diffuse large-scale emission but adds striping artifacts at bright pixels, while PCA preserves compact bright structures but subtracts large-scale modes. Under the cosine beam, the UNet trained on either view alone systematically suppresses the recovered cross-power spectrum on large scales; the hybrid two-channel network does not, and the paper attributes this to the network learning to draw on FD's large-scale fidelity and PCA's small-scale fidelity simultaneously. The result is stated as a bias-free large-scale HI reconstruction, with the caveat

What carries the argument

The mechanism is a two-channel input cube for a 13-layer UNet: channel one is the frequency-differenced cube (adjacent 1 MHz channels subtracted after smoothing the higher-frequency map to the lower-frequency resolution), channel two is the cube after subtracting three principal components along frequency. The contrast between the channels — FD retaining diffuse large-scale structure, PCA retaining compact small-scale structure — is what gives the network the information to avoid the 5–8% large-scale bias each channel alone would bias it toward.

Load-bearing premise

The entire evaluation is in-distribution: training and test cubes come from the same simulation pipeline with the same foreground model and the same beam models, so the claimed accuracy and the abstract's assertion of robustness to imperfect beams have not been tested against genuinely different beam or foreground conditions.

What would settle it

Train the two-channel UNet on cosine-beam maps and test it on maps made with a different beam (e.g., a cosine beam with altered ripple amplitude/period, or a beam from real holographic measurements). If the large-scale cross-correlation ratio departs from unity beyond the 1σ band in that mismatched setting, the paper's robustness claim fails. A cheap version is to train on Gaussian-beam maps and test on cosine-beam maps; the abstract would predict near-unity recovery on large scales, but the body currently contains no such test.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Large-scale HI power, the part most affected by beam-induced spectral structure, can be recovered without the mode-subtraction cost of PCA-only cleaning.
  • The hybrid advantage is specific to the realistic cosine beam: with a Gaussian beam all three U-Net variants are comparable, so chromatic sidelobe structure is the regime where the two-channel design matters.
  • Because the network is trained on matched beam models in this work, the gain at k<0.1 h Mpc^-1 should be re-checked when the beam model is varied between training and inference.
  • If this transfers to real data, it would reduce one systematic in 21-cm auto-power measurements, which currently depend heavily on cross-correlations with galaxy surveys.
  • The 5–8% improvement at fixed network size and training cost suggests that other blind cleaning outputs, not just PCA, could be combined in the same two-channel way.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • A natural reading is that any pair of spectral-smoothness filters with opposite scale biases could be combined this way; ICA and SVD variants are obvious candidates, though the paper does not test them.
  • The abstract claims robustness to imperfect or mismatched beams, but the body only reports matched training/test conditions; this is the paper's gap to close, and it is directly testable.
  • The fact that the hybrid gain appears only under the cosine beam hints that the network is using the FD channel as a 'large-scale anchor' that resists the sidelobe-induced mode mixing; that interpretation could be probed by ablating the FD channel at selected scales.
  • The fixed 3-component PCA subtraction and fixed 1 MHz differencing are hyperparameters of the preprocessing; varying them in tandem with the loss function might push the residual large-scale bias even closer to zero, but no such exploration is reported.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This paper develops a deep-learning foreground and beam mitigation pipeline for 21-cm intensity mapping, comparing three preprocessing strategies feeding a UNet: frequency differencing (FD), principal component analysis (PCA), and a hybrid two-channel combination. Using CRIME simulations of HI plus foregrounds, with optional Gaussian or Cosine beam convolution, the authors report that under the Cosine beam the single-channel FD+UNet and PCA+UNet underestimate the cross-correlation power spectrum by roughly 5–8% at k < 0.1 h Mpc^{-1} and by more than 20% at k ~ 0.2 h Mpc^{-1}, while the hybrid method remains consistent with unity within 1σ at large scales. The paper also claims in the abstract that the method is robust to imperfect or mismatched beams between training and testing, but this claim is not supported by any experiment in the body.

Significance. If the in-body results are correct, the hybrid two-channel preprocessing is a useful, practical contribution: it combines the large-scale fidelity of FD with the small-scale fidelity of PCA, and the improvement under a realistic Cosine beam is directly relevant to ongoing MeerKAT-era intensity mapping analyses. The paper is honest in its in-body comparison and reports concrete power-spectrum ratios with error bars. However, the advertised headline claim of robustness to train/test beam mismatch is absent from the experiments, and the evaluation is entirely in-distribution with respect to the simulation pipeline, so the significance of the paper as written is lower than the abstract suggests.

major comments (3)
  1. [Abstract and Sections IV–V] The abstract states that the method 'can robustly recover the HI signal even when the beam model is imperfect and differs between training and testing,' but no such experiment appears anywhere in Sections IV or V. All reported results train and test on the same beam model (no beam, Gaussian, or Cosine) in matched conditions. This is a load-bearing overclaim: the central advertised generalization is unsupported. The authors should either add an explicit mismatched-beam experiment (e.g., train on Gaussian, test on Cosine, or perturb the Cosine beam parameters between train and test) or soften the abstract and conclusions to describe matched-beam robustness only.
  2. [Section III.B.2, Eq. (8)] The mismatched-beam claim is not merely an untested extrapolation; the FD preprocessing itself depends on the assumed beam. Eq. (8) requires smoothing adjacent-frequency maps to a common resolution using Δθ_FWHM computed from the beam FWHM. If the test-time beam differs from the training beam, the FD channel will contain residual chromatic structure with statistics the network has not seen, and the PCA channel will also change. Thus the advertised robustness is coupled to a beam-dependent input construction. At minimum, the paper should quantify the sensitivity of the hybrid method to beam-model mismatch, or explicitly restrict the claim to the matched-beam case.
  3. [Section IV and Section III.B] The evaluation is entirely in-distribution with respect to the simulation pipeline: training and test samples come from the same CRIME lognormal HI fields, the same Haslam-based foreground model, and the same beam models. The held-out test set measures generalization across random realizations of the same process, not robustness to different foreground morphology, spectral-index variations, or beam systematics. The numbers of PCA modes (fixed at 3) and the FD spacing (fixed at 1 MHz) are free parameters, but no sensitivity analysis is provided. These choices may affect the magnitude of the hybrid improvement. The paper should at least discuss this limitation and, ideally, vary the number of PCA modes or the FD spacing in a robustness check.
minor comments (5)
  1. [Figure 12 caption] The caption of Figure 12 says 'under Cosine beam convolution,' but the surrounding text (Section IV.B) and the third panel describe the Gaussian beam case; Figure 16 is the Cosine-beam counterpart. The caption should be corrected.
  2. [Section IV.C] The text 'which is already visible at large scales (k < hMpc^{-1})' appears to be missing a factor of 0.1; it should read k < 0.1 h Mpc^{-1} for consistency with the rest of the paper.
  3. [Figure 11 caption] The caption says 'correlation coefficient (left panel)' but the correlation coefficient is the right panel; the left panel is the temperature distribution.
  4. [Section III.B.1] The phrase 'Followingdeep21' should cite the deep21 paper (reference [51]) explicitly, as it currently appears as a plain text mention without a citation.
  5. [General] The polynomial coefficients in Eq. (6) are given with a very wide dynamic range; a table or scientific-notation formatting would improve readability.

Circularity Check

0 steps flagged

No circular derivation; the method is an empirical supervised comparison. The abstract's mismatched-beam robustness claim is unsupported, but under-support is not circularity.

full rationale

The central result (hybrid FD+PCA UNet outperforms single-channel FD/PCA under a Cosine beam) is an empirical comparison on held-out test cubes drawn from the same CRIME simulation pipeline, not a quantity enforced by construction. FD (Eq. 7) and PCA are preprocessing definitions, and the UNet is trained against the simulation's own HI truth in the standard supervised way; test cubes are disjoint from training cubes, so the comparison is not a fitted-input-called-prediction. The self-citations to Paper I (simulation pipeline, UNet structure, 1 MHz differencing choice) continue prior published work and are re-tested here rather than assumed as the conclusion. The abstract's claim that 'the method can robustly recover the HI signal even when the beam model is imperfect and differs between training and testing' is not substantiated: Sections IV.B and IV.C train and test with matched Gaussian or Cosine beams only, and no beam-mismatch experiment appears. This is a generalization/validity gap, aggravated by Eq. 8, where the frequency-differencing input is smoothed using the assumed beam FWHM; a different test-time beam would change the FD-channel statistics. But missing support is not a reduction of the result to its inputs, so it does not constitute circularity. The score reflects the normal self-citation and the in-sample evaluation caveat, not a circular derivation.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 0 invented entities

The central claim rests on the fidelity of the simulation pipeline (CRIME lognormal HI + Haslam-based foregrounds), the cosine beam model fitted to MeerKAT holography, and the in-distribution assumption that a UNet trained on 102 patches generalizes to the held-out set and, ultimately, real data. No code or data artifacts are shipped. The free parameters listed (3 PCA modes, 1 MHz differencing, UNet hyperparameters) are choices that are not sensitivity-tested.

free parameters (3)
  • Number of PCA modes removed = 3
    Fixed at 3 components following deep21-style cleaning; affects how much large-scale HI signal is removed and hence what the UNet sees. No sensitivity test is shown.
  • Frequency-differencing spacing = 1 MHz
    Chosen as a trade-off between foreground suppression and signal preservation (Section III.B.2); no scan over spacings shown.
  • UNet training hyperparameters = not reported (deferred to Paper I)
    Architecture and training configuration are stated to be the same as Paper I; epochs, optimizer, learning rate, batch size are not given in this paper.
axioms (3)
  • domain assumption The CRIME simulation pipeline (lognormal HI + Haslam-based foregrounds + point-source models) is a faithful representation of low-redshift 21-cm observations.
    All training/test cubes draw from these simulations (Section II); the claims of 'realistic observational conditions' and generalization to MeerKAT/SKA rely on this.
  • domain assumption The cosine beam model from Matshawule et al. [65] captures the relevant frequency-dependent beam distortions of MeerKAT.
    Equations (5)-(6) adopt empirically fitted beam coefficients; the core result (hybrid superiority under cosine beam) is only meaningful if this model is representative.
  • domain assumption A UNet trained with MSE loss on this simulated dataset will generalize to unseen patches and, by extension, real observations.
    Section III.A trains on 102 samples and validates on 20 from the same pipeline; no distribution shift is tested in the body despite the abstract's claim.

pith-pipeline@v1.3.0-alltime-deepseek · 20387 in / 11539 out tokens · 104780 ms · 2026-08-03T21:26:48.516048+00:00 · methodology

0 comments
read the original abstract

Twenty-one-centimeter intensity mapping is a powerful probe of the large-scale distribution of neutral hydrogen (HI) and cosmological observables such as baryon acoustic oscillations. A major challenge is contamination from bright foregrounds and frequency-dependent beam effects, which can lead to signal loss in traditional methods such as principal component analysis (PCA). We develop a hybrid approach that trains a U-shaped convolutional neural network (UNet) on two input channels derived from frequency differencing (FD) and PCA cleaning, enabling it to exploit their complementary behavior across different scales. This two-channel strategy achieves improved performance, maintaining the cross-correlation power spectrum close to unity on large scales under a cosine beam and improving by 5\%-8\% relative to either FD- or PCA-based UNet alone. We further show that the method can robustly recover the HI signal even when the beam model is imperfect and differs between training and testing, with the large-scale cross-correlation remaining close to unity within the $1\sigma$ level. These results demonstrate that the proposed approach provides a robust framework for HI signal reconstruction under realistic observational conditions.

Figures

Figures reproduced from arXiv: 2511.15072 by Feng Shi, Le Zhang, Ming Jiang, Shulei Ni, Xiaofan Ma, Xiaoping Li, Yanming Liu, Zitong Wang.

Figure 1
Figure 1. Figure 1: FIG. 1. Comparison of beam convolution effects on a simulated point-source map. Left: the input map containing point-like [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. (Left) Normalized angular beam profiles of the Gaussian (blue) and Cosine (red) beam models as a function of angular [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Visualization of the UNet architecture. The input is a cube of size 64 [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Comparison of preprocessing methods applied to pure HI signal fields. From left to right, the panels show the target [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Comparison of preprocessing methods applied to the pure HI fields. Left: cross-correlation coefficient [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Comparison of reconstructed HI maps without beam effects. From left to right, the panels show the true HI signal [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7. Diagnostics of reconstruction performance without beam effects. Left: temperature distribution of the reconstructed [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: FIG. 8. Auto- (left) and cross-power (right) spectrum ratios without beam convolution. The results of FD+UNet (blue dashed), [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: FIG. 9. Comparison of two-dimensional power spectra [PITH_FULL_IMAGE:figures/full_fig_p010_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: FIG. 10. Same as Figure [PITH_FULL_IMAGE:figures/full_fig_p010_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: FIG. 11. Same as Figure [PITH_FULL_IMAGE:figures/full_fig_p010_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: FIG. 12. Auto- and cross-power spectrum ratios under Cosine beam convolution. The first two panels show the results of [PITH_FULL_IMAGE:figures/full_fig_p011_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: FIG. 13. Same as Figure [PITH_FULL_IMAGE:figures/full_fig_p011_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: FIG. 14. Same as Figure [PITH_FULL_IMAGE:figures/full_fig_p012_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: FIG. 15. Same as Figure [PITH_FULL_IMAGE:figures/full_fig_p012_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: shows the auto- and cross-power spec￾trum ratios under Cosine beam convolution. Compared with the no-beam and Gaussian beam cases, all three UNet–based methods experience stronger suppression, which is already visible at large scales (k < h Mpc−1 ) [PITH_FULL_IMAGE:figures/full_fig_p013_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: FIG. 17. Same as Figure [PITH_FULL_IMAGE:figures/full_fig_p013_17.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

68 extracted references · 55 linked inside Pith

  1. [1]

    In our case, PCA is applied to each of the 192 simulated sky patches, with each 643 data cube reshaped into a collection of one-dimensional spectra along the fre- 7 quency axis

    Principal Component Analysis PCA is used to exploit the spectral smoothness of fore- grounds compared to the rapidly fluctuating cosmologi- cal signal. In our case, PCA is applied to each of the 192 simulated sky patches, with each 643 data cube reshaped into a collection of one-dimensional spectra along the fre- 7 quency axis. A frequency–frequency covar...

  2. [2]

    Frequency Differencing Frequency differencing is a model-independent tech- nique that exploits the spectral smoothness of fore- grounds relative to the cosmological 21-cm signal. By taking differences between adjacent frequency channels along each line of sight, the spectrally coherent fore- ground components, which vary slowly with frequency, are effecti...

  3. [3]

    We preprocess the input data cube inde- pendently using PCA and frequency differencing, yield- ing two separate residual maps

    Hybrid Method: Dual-Channel PCA and Frequency Differencing To leverage the complementary strengths of PCA and frequency differencing, we construct a hybrid preprocess- ing approach in which the outputs of both methods are used in parallel. We preprocess the input data cube inde- pendently using PCA and frequency differencing, yield- ing two separate resid...

    2021

  4. [4]

    F. Shi, H. Chang, L. Zhang, H. Shan, J. Zhang, S. Zhou, M. Jiang, and Z. Wang, 21-cm foreground removal using AI and the frequency-difference technique, Phys. Rev. D 109, 063509 (2024), arXiv:2310.06518 [astro-ph.CO]

    Pith/arXiv arXiv 2024

  5. [5]

    Bharadwaj, B

    S. Bharadwaj, B. B. Nath, and S. K. Sethi, Using HI to probe large scale structures at z∼3, Journal of As- trophysics and Astronomy22, 21 (2001), arXiv:astro- ph/0003200 [astro-ph]

    arXiv 2001

  6. [6]

    R. A. Battye, R. D. Davies, and J. Weller, Neutral hydrogen surveys for high-redshift galaxy clusters and protoclusters, MNRAS355, 1339 (2004), arXiv:astro- ph/0401340 [astro-ph]

    arXiv 2004

  7. [7]

    J. B. Peterson, R. Aleksan, R. Ansari, K. Ban- dura, D. Bond, J. Bunton, K. Carlson, T.-C. Chang, F. DeJongh, M. Dobbs, S. Dodelson, H. Darhmaoui, N. Gnedin, M. Halpern, C. Hogan, J.-M. Le Goff, T. T. Liu, A. Legrouri, A. Loeb, K. Loudiyi, C. Magneville, J. Marriner, D. P. McGinnis, B. McWilliams, M. Moniez, N. Palanque-Delabruille, R. J. Pasquinelli, U.-...

    Pith/arXiv arXiv 2010

  8. [8]

    J. S. B. Wyithe and A. Loeb, The 21-cm power spec- trum after reionization, MNRAS397, 1926 (2009), arXiv:0808.2323 [astro-ph]

    Pith/arXiv arXiv 1926

  9. [9]

    Chang, U.-L

    T.-C. Chang, U.-L. Pen, J. B. Peterson, and P. Mc- Donald, Baryon Acoustic Oscillation Intensity Mapping of Dark Energy, Phys. Rev. Lett.100, 091303 (2008), arXiv:0709.3672 [astro-ph]

    Pith/arXiv arXiv 2008

  10. [10]

    H.-J. Seo, S. Dodelson, J. Marriner, D. Mcginnis, A. Stebbins, C. Stoughton, and A. Vallinotto, A Ground- based 21 cm Baryon Acoustic Oscillation Survey, Astro- phys. J.721, 164 (2010), arXiv:0910.5007 [astro-ph.CO]

    Pith/arXiv arXiv 2010

  11. [12]

    P.-J. Wu, Y. Li, J.-F. Zhang, and X. Zhang, Prospects for measuring dark energy with 21 cm intensity map- ping experiments: A joint survey strategy, Science China Physics, Mechanics, and Astronomy66, 270413 (2023), arXiv:2212.07681 [astro-ph.CO]

    Pith/arXiv arXiv 2023

  12. [13]

    Amiri, K

    M. Amiri, K. Bandura, T. Chen, M. Deng, M. Dobbs, M. Fandino, S. Foreman, M. Halpern, A. S. Hill, G. Hinshaw, C. H¨ ofer, J. Kania, T. L. Landecker, J. MacEachern, K. Masui, J. Mena-Parra, N. Miluti- novic, A. Mirhosseini, L. Newburgh, A. Ordog, U.-L. Pen, T. Pinsonneault-Marotte, A. Polzin, A. Reda, A. Re- nard, J. R. Shaw, S. R. Siegel, S. Singh, K. Van...

    Pith/arXiv arXiv 2023

  13. [14]

    Amiri, K

    CHIME Collaboration, M. Amiri, K. Bandura, A. Boskovic, T. Chen, J.-F. Cliche, M. Deng, N. Den- man, M. Dobbs, M. Fandino, S. Foreman, M. Halpern, D. Hanna, A. S. Hill, G. Hinshaw, C. H¨ ofer, J. Ka- nia, P. Klages, T. L. Landecker, J. MacEachern, K. Masui, J. Mena-Parra, N. Milutinovic, A. Mirhos- seini, L. Newburgh, R. Nitsche, A. Ordog, U.-L. Pen, T. P...

    Pith/arXiv arXiv 2022

  14. [15]

    Chen, Radio detection of dark energy—the Tianlai project, Scientia Sinica Physica, Mechanica &amp; As- tronomica41, 1358 (2011)

    X. Chen, Radio detection of dark energy—the Tianlai project, Scientia Sinica Physica, Mechanica &amp; As- tronomica41, 1358 (2011)

    2011

  15. [16]

    Y. Xu, X. Wang, and X. Chen, Forecasts on the Dark En- ergy and Primordial Non-Gaussianity Observations with the Tianlai Cylinder Array, Astrophys. J.798, 40 (2015), arXiv:1410.7794 [astro-ph.CO]

    Pith/arXiv arXiv 2015

  16. [17]

    F. Wu, J. Li, S. Zuo, X. Chen, S. Das, J. P. Mar- riner, T. M. Oxholm, A. Phan, A. Stebbins, P. T. Tim- bie, R. Ansari, J.-E. Campagne, Z. Chen, Y. Cong, Q. Huang, J. Kwak, Y. Li, T. Liu, Y. Liu, C. Niu, C. Osinga, O. Perdereau, J. B. Peterson, J. Podczerwin- ski, H. Shi, G. Siebert, S. Sun, H. Tian, G. S. Tucker, Q. Wang, R. Wang, Y. Wang, Y. Wu, Y. Xu, ...

    Pith/arXiv arXiv 2021

  17. [18]

    Perdereau, R

    O. Perdereau, R. Ansari, A. Stebbins, P. T. Timbie, X. Chen, F. Wu, J. Li, J. P. Marriner, G. S. Tucker, Y. Cong, S. Das, Y. Li, Y. Liu, C. Magneville, J. B. Pe- terson, A. Phan, L. Robinthal, S. Sun, Y. Wang, Y. Wu, Y. Xu, K. Yu, Z. Yu, J. Zhang, J. Zhang, and S. Zuo, The Tianlai dish array low-z surveys forecasts, MNRAS 517, 4637 (2022), arXiv:2205.0608...

    Pith/arXiv arXiv 2022

  18. [19]

    R. A. Battye, I. W. A. Browne, C. Dickinson, G. Heron, B. Maffei, and A. Pourtsidou, H I intensity mapping: a single dish approach, MNRAS434, 1239 (2013), arXiv:1209.0343 [astro-ph.CO]

    Pith/arXiv arXiv 2013

  19. [20]

    Abdalla, E

    E. Abdalla, E. G. M. Ferreira, R. G. Landim, A. A. Costa, K. S. F. Fornazier, F. B. Abdalla, L. Barosi, F. A. Brito, A. R. Queiroz, T. Villela, B. Wang, C. A. Wuen- sche, A. Marins, C. P. Novaes, V. Liccardo, C. Shan, J. Zhang, Z. Zhang, Z. Zhu, I. Browne, J. Delabrouille, L. Santos, M. V. dos Santos, H. Xu, S. Anton, R. Bat- tye, T. Chen, C. Dickinson, Y...

    Pith/arXiv arXiv 2022

  20. [21]

    Santos, P

    M. Santos, P. Bull, S. Camera, S. Chen, J. Fonseca, I. Heywood, M. Hilton, M. Jarvis, G. I. G. Jozsa, K. Knowles, L. Leeuw, R. Maartens, E. Malefahlo, K. McAlpine, K. Moodley, P. Patel, A. Pourtsidou, M. Prescott, K. Spekkens, R. Taylor, A. Witzemann, and I. H. Whittam, A Large Sky Survey with MeerKAT, in MeerKAT Science: On the Pathway to the SKA(2016) p...

    Pith/arXiv arXiv 2016

  21. [22]

    J. Wang, M. G. Santos, P. Bull, K. Grainge, S. Cunning- ton, J. Fonseca, M. O. Irfan, Y. Li, A. Pourtsidou, P. S. Soares, M. Spinelli, G. Bernardi, and B. Engelbrecht, H I intensity mapping with MeerKAT: calibration pipeline for multidish autocorrelation observations, MNRAS505, 3698 (2021), arXiv:2011.13789 [astro-ph.CO]

    Pith/arXiv arXiv 2021

  22. [23]

    L. B. Newburgh, K. Bandura, M. A. Bucher, T. C. Chang, H. C. Chiang, J. F. Cliche, R. Dav´ e, M. Dobbs, C. Clarkson, K. M. Ganga, T. Gogo, A. Gumba, N. Gupta, M. Hilton, B. Johnstone, A. Karastergiou, M. Kunz, D. Lokhorst, R. Maartens, S. Macpherson, M. Mdlalose, K. Moodley, L. Ngwenya, J. M. Parra, J. Peterson, O. Recnik, B. Saliwanchik, M. G. San- tos, ...

    2016

  23. [24]

    Santos, P

    M. Santos, P. Bull, D. Alonso, S. Camera, P. Fer- reira, G. Bernardi, R. Maartens, M. Viel, F. Villaescusa- Navarro, F. B. Abdalla, M. Jarvis, R. B. Metcalf, A. Pourtsidou, and L. Wolz, Cosmology from a SKA HI intensity mapping survey, inAdvancing Astrophysics with the Square Kilometre Array (AASKA14)(2015) p. 19, arXiv:1501.03989 [astro-ph.CO]

    Pith/arXiv arXiv 2015

  24. [25]

    Chang, U.-L

    T.-C. Chang, U.-L. Pen, K. Bandura, and J. B. Peterson, An intensity map of hydrogen 21-cm emission at redshift z˜0.8, Nature (London)466, 463 (2010)

    2010

  25. [26]

    K. W. Masui, E. R. Switzer, N. Banavar, K. Bandura, C. Blake, L. M. Calin, T. C. Chang, X. Chen, Y. C. Li, Y. W. Liao, A. Natarajan, U. L. Pen, J. B. Peterson, J. R. Shaw, and T. C. Voytek, Measurement of 21 cm Brightness Fluctuations at z ˜0.8 in Cross-correlation, ApJL763, L20 (2013), arXiv:1208.0331 [astro-ph.CO]

    Pith/arXiv arXiv 2013

  26. [27]

    L. Wolz, A. Pourtsidou, K. W. Masui, T.-C. Chang, J. E. Bautista, E.-M. M¨ uller, S. Avila, D. Bacon, W. J. Percival, S. Cunnington, C. Anderson, X. Chen, J.-P. Kneib, Y.-C. Li, Y.-W. Liao, U.-L. Pen, J. B. Peterson, G. Rossi, D. P. Schneider, J. Yadav, and G.-B. Zhao, H I constraints from the cross-correlation of eBOSS galaxies and Green Bank Telescope i...

    Pith/arXiv arXiv 2022

  27. [28]

    C. J. Anderson, N. J. Luciw, Y. C. Li, C. Y. Kuo, J. Ya- dav, K. W. Masui, T. C. Chang, X. Chen, N. Opper- mann, Y. W. Liao, U. L. Pen, D. C. Price, L. Staveley- Smith, E. R. Switzer, P. T. Timbie, and L. Wolz, Low- amplitude clustering in low-redshift 21-cm intensity maps cross-correlated with 2dF galaxy densities, MNRAS476, 3382 (2018), arXiv:1710.00424...

    Pith/arXiv arXiv 2018

  28. [29]

    Cunnington, Y

    S. Cunnington, Y. Li, M. G. Santos, J. Wang, I. P. Carucci, M. O. Irfan, A. Pourtsidou, M. Spinelli, L. Wolz, P. S. Soares, C. Blake, P. Bull, B. Engelbrecht, J. Fon- seca, K. Grainge, and Y.-Z. Ma, H I intensity map- ping with MeerKAT: power spectrum detection in cross- correlation with WiggleZ galaxies, MNRAS518, 6262 (2023), arXiv:2206.01579 [astro-ph.CO]

    Pith/arXiv arXiv 2023

  29. [30]

    Mazumder, L

    A. Mazumder, L. Wolz, Z. Chen, S. Paul, M. G. Santos, M. Jarvis, J. Townsend, S. Sekhar, and R. Taylor, HI intensity mapping with the MIGHTEE Survey: first re- sults of the HI power spectrum, MNRAS541, 476 (2025), arXiv:2501.17564 [astro-ph.CO]

    Pith/arXiv arXiv 2025

  30. [31]

    Di Matteo, R

    T. Di Matteo, R. Perna, T. Abel, and M. J. Rees, Radio Foregrounds for the 21 Centimeter Tomography of the Neutral Intergalactic Medium at High Redshifts, Astro- phys. J.564, 576 (2002), arXiv:astro-ph/0109241 [astro- ph]

    Pith/arXiv arXiv 2002

  31. [32]

    S. P. Oh and K. J. Mack, Foregrounds for 21-cm obser- vations of neutral gas at high redshift, MNRAS346, 871 (2003), arXiv:astro-ph/0302099 [astro-ph]

    Pith/arXiv arXiv 2003

  32. [33]

    M. G. Santos, A. Cooray, and L. Knox, Multifrequency Analysis of 21 Centimeter Fluctuations from the Era of Reionization, Astrophys. J.625, 575 (2005), arXiv:astro- ph/0408515 [astro-ph]

    arXiv 2005

  33. [34]

    de Oliveira-Costa, M

    A. de Oliveira-Costa, M. Tegmark, B. M. Gaensler, J. Jonas, T. L. Landecker, and P. Reich, A model of dif- fuse Galactic radio emission from 10 MHz to 100 GHz, MNRAS388, 247 (2008), arXiv:0802.1525 [astro-ph]

    Pith/arXiv arXiv 2008

  34. [35]

    Liu and J

    A. Liu and J. R. Shaw, Data Analysis for Precision 21 cm Cosmology, Publ. Astron. Soc. Pac.132, 062001 (2020), arXiv:1907.08211 [astro-ph.IM]

    Pith/arXiv arXiv 2020

  35. [36]

    Liu and M

    A. Liu and M. Tegmark, A method for 21 cm power spectrum estimation in the presence of foregrounds, Phys. Rev. D83, 103006 (2011), arXiv:1103.0281 [astro- ph.CO]

    Pith/arXiv arXiv 2011

  36. [37]

    E. R. Switzer, T.-C. Chang, K. W. Masui, U.-L. Pen, and T. C. Voytek, Interpreting the unresolved intensity of cosmologically redshifted line radiation, Astrophys. J. 815, 51 (2015), arXiv:1504.07527 [astro-ph.CO]

    Pith/arXiv arXiv 2015

  37. [38]

    S. Zuo, X. Chen, and Y. Mao, A Semiblind PCA-based Foreground Subtraction Method for 21 cm Intensity Mapping, Astrophys. J.945, 38 (2023), arXiv:2208.14675 [astro-ph.CO]

    Pith/arXiv arXiv 2023

  38. [39]

    Chapman, F

    E. Chapman, F. B. Abdalla, G. Harker, V. Jeli´ c, P. Labropoulos, S. Zaroubi, M. A. Brentjens, A. G. de Bruyn, and L. V. E. Koopmans, Foreground removal us- ing F ASTICA: a showcase of LOF AR-EoR, MNRAS423, 2518 (2012), arXiv:1201.2190 [astro-ph.CO]

    Pith/arXiv arXiv 2012

  39. [40]

    Cunnington, M

    S. Cunnington, M. O. Irfan, I. P. Carucci, A. Pourtsidou, and J. Bobin, 21-cm foregrounds and polarization leak- age: cleaning and mitigation strategies, MNRAS504, 208 (2021), arXiv:2010.02907 [astro-ph.CO]

    Pith/arXiv arXiv 2021

  40. [41]

    Chapman, F

    E. Chapman, F. B. Abdalla, J. Bobin, J. L. Starck, G. Harker, V. Jeli´ c, P. Labropoulos, S. Zaroubi, M. A. Brentjens, A. G. de Bruyn, and L. V. E. Koopmans, The scale of the problem: recovering images of reioniza- tion with Generalized Morphological Component Anal- ysis, MNRAS429, 165 (2013), arXiv:1209.4769 [astro- ph.CO]

    Pith/arXiv arXiv 2013

  41. [42]

    L. C. Olivari, M. Remazeilles, and C. Dickinson, Ex- tracting H I cosmological signal with generalized needlet internal linear combination, MNRAS456, 2749 (2016), arXiv:1509.00742 [astro-ph.CO]. 17

    Pith/arXiv arXiv 2016

  42. [43]

    Marins, F

    A. Marins, F. B. Abdalla, K. S. F. Fornazier, E. Abdalla, L. H. F. Assis, M. Remazeilles, C. Alexandre Wuensche, L. Barosi, A. R. Queiroz, T. Villela, B. Wang, C. Feng, R. Landim, V. Liccardo, C. P. Novaes, L. Santos, M. V. dos Santos, and J. Zhang, Foreground removal and 21 cm signal estimates: comparing different blind methods for the BINGO Telescope, a...

    arXiv 2022

  43. [44]

    Kern and A

    N. Kern and A. Liu, Gaussian process foreground sub- traction and power spectrum estimation for 21 cm cos- mology, Mon. Not. Roy. Astron. Soc.501, 1463 (2021), arXiv:2010.15892 [astro-ph.CO]

    Pith/arXiv arXiv 2021

  44. [45]

    J. R. Shaw, K. Sigurdson, U.-L. Pen, A. Stebbins, and M. Sitwell, All-Sky Interferometry with Spherical Har- monic Transit Telescopes, Astrophys. J.781, 57 (2014), arXiv:1302.0327 [astro-ph.CO]

    Pith/arXiv arXiv 2014

  45. [46]

    Ghosh, L

    A. Ghosh, L. V. E. Koopmans, E. Chapman, and V. Jeli´ c, A Bayesian analysis of redshifted 21-cm H I signal and foregrounds: simulations for LOF AR, Mon. Not. Roy. Astron. Soc.452, 1587 (2015), arXiv:1506.04982 [astro- ph.CO]

    Pith/arXiv arXiv 2015

  46. [47]

    Zhang, E

    L. Zhang, E. F. Bunn, A. Karakci, A. Korotkov, P. M. Sutter, P. T. Timbie, G. S. Tucker, and B. D. Wandelt, Bayesian Semi-blind Component Separation for Fore- ground Removal in Interferometric 21 cm Observations, ApJS222, 3 (2016), arXiv:1505.04146 [astro-ph.CO]

    Pith/arXiv arXiv 2016

  47. [48]

    P. H. Sims and J. C. Pober, Joint estimation of the Epoch of Reionization power spectrum and fore- grounds, Mon. Not. Roy. Astron. Soc.488, 2904 (2019), arXiv:1907.02608 [astro-ph.CO]

    Pith/arXiv arXiv 2019

  48. [49]

    Burba, P

    J. Burba, P. H. Sims, and J. C. Pober, All-sky mod- elling requirements for Bayesian 21 cm power spectrum estimation with BAYESEOR, MNRAS520, 4443 (2023), arXiv:2302.04058 [astro-ph.IM]

    Pith/arXiv arXiv 2023

  49. [50]

    J. Ding, X. Wang, U.-L. Pen, and X.-D. Li, Correlation- based Beam Calibration of 21 cm Intensity Mapping, ApJS274, 44 (2024), arXiv:2408.06682 [astro-ph.CO]

    Pith/arXiv arXiv 2024

  50. [51]

    Asorey, D

    J. Asorey, D. Parkinson, F. Shi, Y.-S. Song, K. Ahn, J. Kim, J. Yao, L. Zhang, and S. Zuo, HIR4: cosmology from a simulated neutral hydrogen full sky using Hori- zon Run 4, MNRAS495, 1788 (2020), arXiv:2001.00833 [astro-ph.CO]

    Pith/arXiv arXiv 2020

  51. [52]

    Shi, Y.-S

    F. Shi, Y.-S. Song, J. Asorey, D. Parkinson, K. Ahn, J. Yao, L. Zhang, and S. Zuo, HIR4: cosmological signa- tures imprinted on the cross-correlation between a 21-cm map and galaxy clustering, MNRAS499, 4613 (2020), arXiv:2006.01407 [astro-ph.CO]

    Pith/arXiv arXiv 2020

  52. [53]

    W. Li, H. Xu, Z. Ma, R. Zhu, D. Hu, Z. Zhu, J. Gu, C. Shan, J. Zhu, and X.-P. Wu, Separating the EoR signal with a convolutional denoising autoencoder: a deep-learning-based method, MNRAS485, 2628 (2019), arXiv:1902.09278 [astro-ph.IM]

    Pith/arXiv arXiv 2019

  53. [54]

    T. L. Makinen, L. Lancaster, F. Villaescusa-Navarro, P. Melchior, S. Ho, L. Perreault-Levasseur, and D. N. Spergel, deep21: a deep learning method for 21 cm foreground removal, JCAP2021, 081 (2021), arXiv:2010.15843 [astro-ph.CO]

    Pith/arXiv arXiv 2021

  54. [55]

    Gagnon-Hartman, Y

    S. Gagnon-Hartman, Y. Cui, A. Liu, and S. Ravan- bakhsh, Recovering the wedge modes lost to 21-cm fore- grounds, MNRAS504, 4716 (2021), arXiv:2102.08382 [astro-ph.CO]

    Pith/arXiv arXiv 2021

  55. [56]

    S. Ni, Y. Li, L.-Y. Gao, and X. Zhang, Eliminating Pri- mary Beam Effect in Foreground Subtraction of Neutral Hydrogen Intensity Mapping Survey with Deep Learning, Astrophys. J.934, 83 (2022), arXiv:2204.02780 [astro- ph.IM]

    Pith/arXiv arXiv 2022

  56. [57]

    Kennedy, J

    J. Kennedy, J. C. Carr, S. Gagnon-Hartman, A. Liu, J. Mirocha, and Y. Cui, Machine-learning recovery of foreground wedge-removed 21-cm light cones for high-z galaxy mapping, MNRAS529, 3684 (2024), arXiv:2308.09740 [astro-ph.CO]

    Pith/arXiv arXiv 2024

  57. [58]

    Bianco, S

    M. Bianco, S. K. Giri, D. Prelogovi´ c, T. Chen, F. G. Mertens, E. Tolley, A. Mesinger, and J.-P. Kneib, Deep learning approach for identification of H II regions during reionization in 21-cm observations - II. Foreground con- tamination, MNRAS528, 5212 (2024), arXiv:2304.02661 [astro-ph.IM]

    Pith/arXiv arXiv 2024

  58. [59]

    Bianco, S

    M. Bianco, S. K. Giri, R. Sharma, T. Chen, S. P. Kr- ishna, C. Finlay, V. Nistane, P. Denzel, M. De Santis, and H. Ghorbel, Deep learning approach for identifica- tion of H II regions during reionization in 21-cm obser- vations – III. Image recovery, MNRAS541, 234 (2025), arXiv:2408.16814 [astro-ph.CO]

    Pith/arXiv arXiv 2025

  59. [60]

    Sabti, R

    N. Sabti, R. Purandhar Reddy Sudha, J. B. Mu˜ noz, S. Mishra-Sharma, and T. Youn, A generative model- ing approach to reconstructing 21 cm tomographic data, Machine Learning: Science and Technology6, 015039 (2025), arXiv:2407.21097 [astro-ph.CO]

    Pith/arXiv arXiv 2025

  60. [61]

    Chen, K.-F

    S.-F. Chen, K.-F. Chen, and C. Dvorkin, Field- level Reconstruction from Foreground-Contaminated 21- cm Maps, arXiv e-prints , arXiv:2508.13265 (2025), arXiv:2508.13265 [astro-ph.CO]

    arXiv 2025

  61. [62]

    Alonso, P

    D. Alonso, P. G. Ferreira, and M. G. Santos, Fast simu- lations for intensity mapping experiments, MNRAS444, 3183 (2014), arXiv:1405.1751 [astro-ph.CO]

    Pith/arXiv arXiv 2014

  62. [63]

    C. G. T. Haslam, C. J. Salter, H. Stoffel, and W. E. Wilson, A 408-MHZ All-Sky Continuum Survey. II. The Atlas of Contour Maps, A&AS47, 1 (1982)

    1982

  63. [64]

    Delabrouille, M

    J. Delabrouille, M. Betoule, J. B. Melin, M. A. Miville- Deschˆ enes, J. Gonzalez-Nuevo, M. Le Jeune, G. Castex, G. de Zotti, S. Basak, M. Ashdown, J. Aumont, C. Bacci- galupi, A. J. Banday, J. P. Bernard, F. R. Bouchet, D. L. Clements, A. da Silva, C. Dickinson, F. Dodu, K. Dolag, F. Elsner, L. Fauvet, G. Fa¨ y, G. Giardino, S. Leach, J. Lesgourgues, M. ...

    Pith/arXiv arXiv 2013

  64. [65]

    Remazeilles, C

    M. Remazeilles, C. Dickinson, A. J. Banday, M. A. Bigot- Sazy, and T. Ghosh, An improved source-subtracted and destriped 408-MHz all-sky map, MNRAS451, 4311 (2015), arXiv:1411.3628 [astro-ph.IM]

    Pith/arXiv arXiv 2015

  65. [66]

    G. B. Rybicki and A. P. Lightman,Radiative Processes in Astrophysics(1986)

    1986

  66. [67]

    K. M. G´ orski, E. Hivon, A. J. Banday, B. D. Wan- delt, F. K. Hansen, M. Reinecke, and M. Bartelmann, HEALPix: A Framework for High-Resolution Discretiza- tion and Fast Analysis of Data Distributed on the Sphere, Astrophys. J.622, 759 (2005), arXiv:astro-ph/0409513 [astro-ph]

    Pith/arXiv arXiv 2005

  67. [68]

    S. D. Matshawule, M. Spinelli, M. G. Santos, and S. Ngobese, H I intensity mapping with MeerKAT: pri- mary beam effects on foreground cleaning, MNRAS506, 5075 (2021), arXiv:2011.10815 [astro-ph.CO]. 18

    Pith/arXiv arXiv 2021

  68. [69]

    Y.-S. Song, Y. Zheng, and A. Taruya, Toward a more stringent test of gravity with the redshift space power spectrum: Simultaneous probe of growth and ampli- tude of large-scale structure, PhRvD104, 043528 (2021), arXiv:2102.01785 [astro-ph.CO]

    Pith/arXiv arXiv 2021