{"id":"e783a431-87d5-41d6-b13e-09e0d50c8d95","arxiv_id":"2501.06890","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"The full four-year SPTpol 500 deg² dataset yields TE and EE power spectra from ℓ=50 to 8000 that are consistent with ΛCDM, with the most sensitive damping-tail measurements at high ℓ and a fitted H0 of 70.48 ± 2.16 km/s/Mpc.","lead":"This paper presents new measurements of the cosmic microwave background's temperature and polarization patterns from four years of data taken by the SPTpol telescope at the South Pole, covering 500 square degrees of sky. The results confirm the standard cosmological model and give the most precise view yet of the CMB's small-scale damping tail, a regime relevant to the Hubble constant tension.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"TE transfer function is a geometric mean of TT and EE plus a simulation-derived 'TE bias'; if that correction is cosmology-dependent, the headline damping-tail TE/EE spectra are biased and the TE-only PTE would not catch it.","rationale":"The reader's CONDITIONAL verdict is appropriate. My stress-test converges on the transfer-function/TE-bias element of the reader's weakest_assumption, which is the most load-bearing part because it directly affects the high-ell TE bandpowers behind the sensitivity claim and is connected to the unexplained TE-only PTE. The alternate-cosmology test in Section 6.3 is a good start, but it does not isolate the TE bias: it only reports parameter shifts, which can be insensitive to a bias whose shape is orthogonal to the parameter directions. The paper is otherwise exemplary in disclosure, including the artificially tight point-source upper limit, the 57 erased eigenmodes, and the low PTE, so no stricter verdict is warranted. If the proposed targeted re-analysis shows the TE bias is robust, the central claim stands.","tokens_in":739,"tokens_out":711,"duration_ms":76095,"concrete_test":"Recompute the TE bias from the 90 alternate-cosmology simulations (same pipeline, Section 4.6) and apply it to the real data instead of the Planck-cosmology TE bias; compare the resulting TE bandpowers and the TE-only LambdaCDM PTE. If any high-ell TE bin shifts by more than ~0.3 sigma of its quoted error, or if the TE-only PTE changes from 0.24% to above 1%, the TE bias correction is cosmology-dependent and the headline damping-tail claim is not secure.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's headline claim rests on TE and EE bandpowers at 1700 1700 in TE and ell > 2000 in EE. The paper includes an extensive set of internal consistency checks: seven jackknife null tests, frequency-split consistency, minimum-variance bandpower tests, data-split parameter comparisons, and an alternate-cosmology pipeline recovery.","tokens_in":27459,"tokens_out":7180,"duration_ms":69371,"significance":"If the results hold, the public SPTpol spectra provide the deepest high-ell view of the CMB damping tail in TE and EE from a ground-based experiment, and the cosmological constraints are competitive with other contemporary CMB data sets. The paper's strengths are the breadth and care of its internal consistency program, the external anchoring of the calibration chain to Planck 143 GHz rather than to the paper's own target parameters, the candid discussion of the AL < 1 preference and the unexplained TE-only goodness-of-fit, and the public release of maps, bandpowers, covariance, and likelihood. The main correctness risks are concentrated in the simulation-based TE transfer-function and TE-bias corrections, which rest on a single fiducial cosmology, and in the conditioning of the bandpower covariance matrix, which requires erasing 57 eigenmodes before the internal consistency test is passed. These issues are concrete and testable, and they bear directly on the headline damping-tail claim, so they should be addressed before the paper is accepted.","major_comments":[{"comment":"The TE transfer function is not measured directly; it is defined as the geometric mean of the TT and EE transfer functions (Section 4.3), and the residual 'TE bias' is estimated from 226 simulated datasets generated from a single fiducial cosmology, Planck base_plikHM_TT_lowTEB_lensing (Section 4.6). The alternate-cosmology test in Section 6.3 uses 90 realizations of a very different cosmology, applies the same unbiasing kernel, and checks parameter recovery, but it does not compare the TE bias at the bandpower level. If the true sky's TE spectrum shape, or the foreground polarization levels, differ from the fiducial assumptions in a way that changes the TE bias, the high-ell TE bandpowers that support the headline damping-tail claim would be biased. Please either compute the TE bias for the alternate cosmology and compare it bandpower-by-bandpower with the fiducial TE bias, or measure the TE transfer function directly using a zero-crossing-safe procedure, or provide an analytic argument for why the geometric-mean and bias corrections are insensitive to the TE spectrum shape.","section":"Section 4.3 and Section 4.6"},{"comment":"The bandpower covariance matrix as constructed fails the minimum-variance internal consistency test (chi^2 too high for 280 degrees of freedom), and the paper conditions the matrix by setting 57 eigenvalues below 2.619e-5 of the largest eigenvalue to very large values. The justification is that a Kolmogorov-Smirnov test on 226 simulated chi^2 values passes and that parameter constraints are degraded by less than 6% in width. However, the erased eigenmodes are precisely the directions in which the estimated covariance is smallest; if those small eigenvalues are not purely spurious but reflect genuine near-degeneracies in the data vector, erasing information along them could hide real inconsistencies or underestimate uncertainties in those directions. The current tests check widths, not bias. Please report the eigenvalue spectrum, the projections of the actual data residual onto the 57 erased modes, and the sensitivity of the reported parameters and PTEs to the conditioning threshold.","section":"Section 5.3"},{"comment":"The TE-only data split has a best-fit LCDM chi^2 = 238.11 for 180 degrees of freedom, corresponding to a PTE of 0.24%, and the paper states that the origin of this low PTE is not completely understood. The PTE improves to 1.5% when the highest-ell 95x95 TE bin (7000 < ell < 8000) is removed, so the discrepancy is concentrated in the high-ell TE bandpowers that anchor the damping-tail claim. Because the paper's central claim is the sensitivity of the TE damping-tail measurement, an unresolved systematic in TE directly affects that claim. Please investigate whether the low PTE is connected to the TE bias or transfer-function approximations, the beam and calibration nuisance parameters, or a specific high-ell bin, or alternatively soften the damping-tail claim in the abstract and conclusions until the origin is understood.","section":"Section 7.1, Table 4"}],"minor_comments":[{"comment":"The title contains stray spaces in 'T emperature' and 'F ull' in the manuscript text; these should be fixed.","section":"Title"},{"comment":"The abstract's phrase 'using only the SPTpol data and a Planck-based prior on the optical depth' should be clarified, because the SPTpol temperature calibration and low-ell beam amplitude are fitted against the Planck 143 GHz map in Section 4.4. The text should state explicitly that the calibration is anchored to Planck, not just that a Planck tau prior is used.","section":"Abstract and Section 4.4"},{"comment":"The definition of the effective number of degrees of freedom nu_b in Eq. (6) and its inversion 'backwards' from simulated auto-spectra is described tersely; a brief explanation of how nu_b is determined for the final coarse bins would improve reproducibility.","section":"Section 5.2"},{"comment":"The nuisance parameter E95to150_cal is defined as T95cal * P95cal / (T150cal * P150cal), but the notation is dense; a table listing all nuisance parameters, their definitions, and their priors would improve readability.","section":"Section 6.2"},{"comment":"The captions of Figures 4 and 5 note that small offsets in ell have been added for plotting, but the corresponding subpanels in Figures 6 and 7 do not state this; please add the same note there.","section":"Figures 4 and 5"}],"recommendation":"major_revision","confidential_remarks":"This is a careful and transparent data-release paper, and I do not see grounds for rejection. The three major concerns—TE transfer-function and bias cosmology dependence, covariance matrix conditioning, and the unexplained TE-only PTE—are all load-bearing for the headline damping-tail claim, but each can be addressed with additional tests or by suitably qualifying the claims. The AL < 1 preference is discussed honestly and is not itself a reason for rejection, particularly given the paper's comparison with the lensing-reconstruction result on the same field."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this as a data-release paper with a cosmological fit attached. The spectra are the product, and they are real. The full four-year SPTpol dataset, both 95 and 150 GHz, gives TE and EE from 50 < ell < 8000, and the damping-tail measurements at ell > 1700 (TE) and ell > 2000 (EE) are the most sensitive in those regimes. The paper does the usual SPT-level careful work: seven jackknife null tests, frequency-split consistency, alternate-cosmology recovery with doubled foregrounds, beam and calibration treated as nuisance parameters, and the data products are public. It is also honest about its own limitations, flagging the artificially tight point-source upper limit, the TE-only PTE of 0.24%, and the covariance conditioning. That candor counts.\n\nThe main soft spot is the one the stress-test note hits. The TE transfer function is not measured directly; it is the geometric mean of the TT and EE transfer functions, and the residual is absorbed into a 'TE bias' computed from 226 simulations of a single fiducial cosmology. The alternate-cosmology test recovers parameters, not per-bandpower TE bias, so a cosmology-dependent bias in that additive correction is not fully excluded. The unexplained TE-only PTE, which improves when the highest-ell 95x95 TE bin is removed, is a fair place to point. This is a real concern, but it is not a demonstrated flaw. The paper bounds it with multiple checks, and the TE-only fit is not driving the main claims.\n\nThe covariance conditioning is also mildly uncomfortable: the matrix fails internal consistency until 57 eigenmodes are erased, and the quoted errors carry that step. The authors show parameter constraints degrade by at most about 6%, which is reassuring. The AL = 0.70 ± 0.13 preference is a roughly 2 sigma tension with unity; the paper argues it is more likely statistical than a measurement of low lensing power, citing the lensing reconstruction from the same field (AL * AL_phi = 0.995 ± 0.090). That argument holds.\n\nNo load-bearing flaw here. The cuts of 228 observations and a 2D Fourier mask are post-hoc but disclosed, and the jackknife PTE thresholds are stated. The central claims, that the spectra are consistent with Lambda CDM, that they are the most sensitive damping-tail measurements in the quoted ranges, and that the parameters are consistent with Planck under AL marginalization, are supported.\n\nThis paper is for anyone doing CMB cosmology with small-scale TE/EE, and for methodologists looking at transfer-function systematics. It deserves a serious referee; the review should push on the TE transfer function and the TE-only PTE, but this is not a desk reject.","headline":"A careful, honest data-release paper whose headline damping-tail claims hold up; the TE transfer-function construction is the soft spot to push on in review.","tokens_in":28534,"tokens_out":2548,"would_cite":true,"duration_ms":25569,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Using the full four-year SPTpol 500-square-degree dataset, this paper presents TE and EE CMB power spectra over $50<\\ell<8000$ and argues they are the most sensitive lensed-damping-tail measurements for $\\ell>1700$ in TE and $\\ell>2000$…","keywords":["cosmic microwave background","E-mode polarization","angular power spectrum","damping tail","SPTpol","Lambda-CDM model","CMB lensing","Hubble constant"],"falsifier":"Recompute the TE bandpowers using a directly simulated TE transfer function (rather than the geometric mean of TT and EE) and compare with the published spectra; a shift of the $\\ell>1700$ TE bandpowers by more than their quoted uncertainties would refute the headline sensitivity claim. Separately, an independent high-resolution CMB experiment covering the same patch and multipole range would settle whether $A_L=0.70\\pm0.13$ is a real preference or a statistical fluctuation.","tokens_in":26935,"feed_emoji":"🌌","tokens_out":8335,"duration_ms":67155,"temperature":0.7,"pith_summary":"This paper uses the complete four-year SPTpol observation of a 500-square-degree patch of sky, in both the 95\\,GHz and 150\\,GHz bands, to measure the temperature$-$E-mode polarization cross-spectrum (TE) and the E-mode auto-spectrum (EE) over angular multipoles $50<\\ell<8000$. Its central claim is that these are the deepest measurements of the lensed CMB damping tail to date, for roughly $\\ell>1700$ in TE and $\\ell>2000$ in EE. The authors show the dataset passes internal consistency checks across maps, frequency bands, and spectrum types, and that the full dataset is well fitted by the $\\Lambda$CDM model, giving $H_0=70.48\\pm2.16$\\,\\text{km}\\,\\text{s}^{-1}\\,\\text{Mpc}^{-1}$ and $\\Omega_m=0.271\\pm0.026$ when a Planck-based prior on the reionization optical depth is adopted. They also report a persistent $\\sim2\\$\\sigma$$ preference for a low lensing amplitude, $A_L=0.70\\pm0.13$, and marginalizing over $A_L$ brings the parameters into agreement with Planck. If correct, the released spectra become the reference ground-based measurements of the small-scale CMB damping tail and a new cosmological constraint.","feed_headline":"Deepest CMB damping-tail spectra yet, from 500 sq deg","feed_subtitle":"Full four-year SPTpol TE/EE data fit Lambda-CDM with H0 = 70.5 +/- 2.2 km/s/Mpc.","key_machinery":"The pipeline is a pseudo-spectrum analysis: cross-spectra of 50 map bundles are computed at $\\Delta\\ell=5$ and unbiassed by inverting the kernel $K=MFB^2$, where $M$ is the flat-sky analytic mode-coupling matrix from the apodization mask, $F$ is the filter transfer function, and $B$ is the beam. The transfer functions are solved iteratively from 226 simulated datasets generated from a single fiducial cosmology, with the TE transfer function constructed as the geometric mean of the TT and EE transfer functions because of TE zero-crossings; a 'TE bias' correction absorbs residual misrecovery of simulated TE spectra. The bandpower covariance combines a simulated signal-only part with an analytic noise form, and is 'further conditioned' by erasing 57 near-null eigenvectors before inversion. Cosmological fits use a CosmoPower-trained emulator for $\\Lambda$CDM with nuisance parameters for calibration, beam eigenmodes, foregrounds, and super-sample lensing.","core_discovery":"The paper's discovery claim is that the SPTpol 500\\,\\text{deg}^2 field, observed for four years in two frequency bands, produces TE and EE power spectra that are the most sensitive probes of the lensed CMB damping tail for $\\ell>1700$ (TE) and $\\ell>2000$ (EE). On the paper's own terms, the full dataset is self-consistent, fits $\\Lambda$CDM well (goodness-of-fit PTE of 0.10), and yields $H_0=70.48\\pm2.16$\\,\\text{km}\\,\\text{s}^{-1}\\,\\text{Mpc}^{-1}$, $\\Omega_m=0.271\\pm0.026$, and $\\sigma_8=0.758\\pm0.022$ under a Planck-based $\\tau$ prior. The data favor a lensing amplitude $A_L=0.70\\pm0.13$, about $2\\$\\sigma$$ below unity; this low-$A_L$ preference drives most of the mild inconsistency between $\\ell<1000$ and $\\ell>1000$ parameter constraints, and when $A_L$ is floated the SPTpol constraints become consistent with Planck. The authors interpret the low $A_L$ as more likely a statistical fluctuation than evidence for reduced lensing, because lensing-reconstruction analyses of the same field are consistent with $A_L=1$.","pith_inferences":["Because the TE transfer function is a geometric mean of TT and EE rather than a directly measured TE response, any scale-dependent difference between TE and TT/EE filtering would imprint a bias in the TE damping tail; a direct TE transfer-function calibration in simulations would test this and could alter the $\\ell>2000$ TE claim.","The flat-sky mode-coupling matrix is cross-checked against a curved-sky version only over $500<\\ell<3000$ in the 150\\,GHz EE band; recomputing the full bandpower set with a curved-sky $M$ would test whether the low-$A_L$ preference could be a mode-coupling artifact.","The public release of maps, bandpowers, and covariance invites external reanalyses, e.g., with different $\\tau$ priors or free neutrino mass, which would test the robustness of the reported $H_0$ and $A_L$ constraints."],"forward_implications":["The released TE/EE bandpowers over $50<\\ell<8000$ become the deepest ground-based reference for the lensed damping tail, sharpening constraints on parameters sensitive to small-scale damping (e.g., $n_s$, $H_0$, $\\Omega_m$) in the multipole range 1700$-$8000.","The reported $H_0=70.48\\pm2.16$\\,\\text{km}\\,\\text{s}^{-1}\\,\\text{Mpc}^{-1}$ provides an independent ground-based CMB estimate (with a Planck $\\tau$ prior) that sits between Planck and late-universe values.","If the $A_L\\approx0.7$ preference is real, it would indicate that the acoustic peaks in this patch are less smoothed by lensing than $\\Lambda$CDM predicts; the paper's own lensing-reconstruction cross-check indicates this is likely a statistical fluctuation.","The consistency between 95\\,GHz and 150\\,GHz, and between TE and EE, supports the foreground and calibration models used in the analysis, making the systematic budget credible for future combined analyses."],"supporting_citations":[{"why":"Supplies the predecessor analysis, data selection, and pseudo-spectrum method that this work extends to the full four-year two-band dataset.","marker":"H18"},{"why":"Provides the TE bias treatment and the curved-sky mode-coupling cross-check used to validate the flat-sky mode-coupling matrix.","marker":"Dutcher et al. 2021"},{"why":"Provides the pseudo-spectrum formalism and mode-coupling matrix used to unbias the bandpowers.","marker":"Hivon et al. 2002"},{"why":"Supplies the fiducial simulation cosmology, the Gaussian prior on optical depth, and the calibration reference for beam and temperature calibration.","marker":"Planck Collaboration et al. 2020"},{"why":"Provides the method for constructing the signal-only part of the bandpower covariance from simulated auto-spectra.","marker":"Crites et al. 2015"},{"why":"Supplies the analytic noise-variance formula used for the diagonal of the bandpower covariance matrix.","marker":"Lueker et al. 2010"}],"fun_headline_variants":["Deepest CMB damping-tail spectra yet from SPTpol's 500 sq deg","SPTpol full dataset gives best TE/EE damping-tail sensitivity","Full SPTpol data yield most sensitive CMB damping-tail TE/EE to date","SPTpol's 4-year 500-deg^2 CMB spectra set a damping-tail record","Best-ever CMB damping-tail TE/EE from SPTpol's entire field"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis assumes that the 226 simulated time streams, generated from a single fiducial cosmology, reproduce the real telescope's filtering and noise response well enough that the derived transfer functions and TE bias corrections are unbiased; if the real data respond differently in a way the tests do not catch, the headline TE and EE damping-tail bandpowers would be biased.","fun_headline_variants_meta":{"raw":{"variants":["Deepest CMB damping-tail spectra yet from SPTpol's 500 sq deg","SPTpol full dataset gives best TE/EE damping-tail sensitivity","Full SPTpol data yield most sensitive CMB damping-tail TE/EE to date","SPTpol's 4-year 500-deg^2 CMB spectra set a damping-tail record","Best-ever CMB damping-tail TE/EE from SPTpol's entire field"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000444,"raw_usage":{"total_tokens":2373,"prompt_tokens":1201,"completion_tokens":1172,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":817,"completion_tokens_details":{"reasoning_tokens":1062}},"tokens_in":817,"tokens_out":1172,"duration_ms":9423,"temperature":1.0,"reasoning_tokens":1062,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T20:52:41.592098+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the TE bandpowers using a directly simulated TE transfer function (rather than the geometric mean of TT and EE) and compare with the published spectra; a shift of the $\\ell>1700$ TE bandpowers by more than their quoted uncertainties would refute the headline sensitivity claim. Separately, an independent high-resolution CMB experiment covering the same patch and multipole range would settle whether $A_L=0.70\\pm0.13$ is a real preference or a statistical fluctuation.","supporting_citations":[{"cited_title":"T., Henning , J","cited_arxiv_id":null,"evidence_quote":"Provides the method for constructing the signal-only part of the bandpower covariance from simulated auto-spectra."}],"review_version":1}