{"id":"820f84b3-3225-429b-9207-e40efe258228","arxiv_id":"2501.04158","paper_version":2,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"If lensing field rotation also rotates CMB polarization, SPT-3G plus galaxy-survey templates should detect the effect at SNR~7, and CMB-S4 deep could detect it internally at 3.9σ.","lead":"This paper forecasts how well current and future CMB experiments can detect the rotation of the cosmic microwave background polarization caused by gravitational lensing at second order. It finds that SPT-3G data combined with large-scale structure templates could see this rotation at signal-to-noise around 7, if the controversial theory that lensing rotates polarization is correct.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The headline S/N=7.1 is conditional on the exact, contested β=-ω relation; the paper neither derives it nor quantifies the degradation from partial β-ω decorrelation, so this assumption is the least secure link in the central claim.","rationale":"The paper is a careful forecasting study, and the reader's ACCEPT verdict is structurally sound. The strongest claim is explicitly conditional: if β=-ω, SPT-3G-7y will detect the combined signal at S/N≈7 and CMB-S4-deep at S/N≈38. The reader identified the same load-bearing assumption: the exact, contested β=-ω relation. My independent reading confirms this is the least secure link. The paper is transparent about the assumption and supplies the ω-only alternative, so the internal logic is not flawed; however, the abstract's framing that observations can decide the controversy rests on the discriminating power of the β component, and that power is entirely supplied by β=-ω. No degradation for partial correlation is shown, and the relation is not derived here. I do not see a stronger internal inconsistency: the quadratic-estimator responses, N^0 and N^1 calculations, delensing procedure, and LSS-template formulas are consistent with the cited literature and with the full-sky checks the authors report. The self-correlation concern that the CMB κ tracer shares maps with the rotation estimator is mitigated by parity: the disconnected noise covariance between the parity-odd ω estimator and parity-even κ estimator vanishes in the Gaussian limit, and no other shared-noise term is apparent. Therefore the concern is not fatal enough to change the verdict; it supports keeping the forecast conditional, but the paper already states that condition. The proposed coherence test would quantify how tightly the headline depends on the exact relation.","tokens_in":28008,"tokens_out":13562,"duration_ms":147950,"concrete_test":"Recompute Table 4 and Fig. 7 with a one-parameter family β=-r ω, keeping C_ββ=C_ωω but setting C_βω=-r C_ωω for r in {0, 0.3, 0.5, 0.7, 0.8, 0.9, 1}. If the SPT-3G-7y S/N for the combined rotation signal falls below about 5 for r<0.8, then the promised discrimination of the β=-ω controversy is not robust to even modest decorrelation; if it remains high across the full range, the concern about the exact relation is less consequential for the headline claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central forecast for detecting the combined rotation signal at SPT-3G-7y with S/N≈7.1 (Table 4, β=-ω, κ+g+CIB) and for CMB-S4-deep at S/N≈38 assumes C_ββ=C_ωω and C_βω=-C_ωω, i.e. β=-ω exactly. This is asserted in Section 1 and used in Tables 1, 2, 4 and Fig. 7, but the paper explicitly declines to derive it: 'we do not enter in this controversy, but we shall show that observations can actually decide about it.' The relation comes from Refs [18-20], two of whose authors are co-authors of the present paper, and is disputed by Refs [15-17]. The paper provides the ω-only column in Table 3, where SPT-3G-7y drops from 7.1 to 4.9, so the extra discriminating power that would settle the polarization-rotation controversy is entirely supplied by the contested exact relation. No intermediate case with partial coherence, C_βω = -r C_ωω with r<1, is explored. Because β and ω arise from different line-of-sight weightings of the same post-Born lensing history, there is no a priori guarantee that they remain perfectly coherent; even a modest decorrelation could bring SPT-3G from a clear detection of the polarization-rotation component to a marginal one. The estimator formalism, N^0/N^1 treatment, delensing checks, and LSS template construction appear internally consistent, so the load-bearing weak point is specifically the exactness of β=-ω.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies the detectability of the second-order (post-Born) lensing curl omega and the associated polarization rotation beta. It constructs minimum-variance quadratic estimators for beta, omega, and kappa, derives their reconstruction noise N^(0) and the lensing-induced N^(1) bias, validates the noise predictions against full-sky simulations, and forecasts detection significances for LiteBIRD, SO, SPT-3G-7y, PICO, and CMB-S4, both internally and in cross-correlation with LSS templates built from CMB lensing convergence, galaxy clustering, and CIB. The central numerical results are conditional on the contested relation beta=-omega: SPT-3G-7y is forecast to reach S/N about 7.1 and CMB-S4-deep about 38 with the combined kappa+g+CIB template, while the omega-only case gives 4.9 for SPT-3G-7y.","tokens_in":28303,"tokens_out":22417,"duration_ms":211789,"significance":"The paper is significant because it gives a self-contained, quantitative path to detecting a genuinely second-order lensing effect with currently planned experiments, and it frames the beta=-omega controversy as a falsifiable observational question. The estimator formalism is careful, the N^(1) treatment is checked with simulations, and the forecasts are transparent about the assumptions on LSS tracers, delensing residuals, and the beta=-omega hypothesis. If the forecasts are correct, the post-Born lensing curl becomes detectable with SPT-3G in cross-correlation, which would be a first. The main weakness is that the headline significance for the polarization-rotation component rests entirely on an exact relation that the paper does not itself establish, and no intermediate partial-coherence scenario is quantified.","major_comments":[{"comment":"The headline S/N values (SPT-3G-7y: 7.1; CMB-S4-deep: 38.3 in Table 4) assume beta=-omega exactly, i.e., C_beta beta = C_omega omega and C_beta omega = -C_omega omega. The paper explicitly declines to settle this controversy and provides no intermediate case. Because the omega-only column in Table 3 gives SPT-3G-7y S/N=4.9, the additional discriminating power of the claimed detection is supplied entirely by the contested relation. I recommend adding a one-parameter interpolation beta = -r omega (equivalently C_beta beta = r^2 C_omega omega and C_beta omega = -r C_omega omega) with r in [0,1] to Tables 3-4 and Fig. 7, so that the degradation under partial decorrelation is visible. This is needed to support the conclusion that the polarization-rotation component will be observed at high significance soon.","section":"Section 1; Tables 3 and 4"},{"comment":"There is an apparent sign inconsistency in the treatment of beta and omega. Equation (C.1) gives delta B = -2 beta E for polarization rotation, while Eq. (C.8) gives delta B = -2 omega E for the lensing rotation contribution. Under the stated beta=-omega relation, these two rotation-induced B contributions cancel exactly, leaving only the shear part of the curl signal. Yet Section 4.1 and Fig. 4 describe the beta=-omega cross-term as boosting the B-mode power, and Table 1 reports a 3.9 sigma internal detection for this case. The authors should reconcile the sign conventions, for example by explicitly showing how the shear contribution and the partial decorrelation of the large-scale B modes (the reported -0.8 cross-correlation) carry the signal, or correct the equations and figure caption. As written, the reader cannot determine whether the beta=-omega forecasts correspond to the same physical scenario illustrated in Fig. 1.","section":"Section 3.2; Appendix C; Fig. 4"}],"minor_comments":[{"comment":"The first paragraph contains a typo: 'It has let to' should read 'It has led to'.","section":"Introduction"},{"comment":"The word 'degenaracy'/'degenaracies' appears in several places (e.g., Section 3.2 and Section 5) and should be corrected to 'degeneracy'/'degeneracies'.","section":"Throughout"},{"comment":"The text states that positive image rotation by omega is equivalent to polarization rotation by beta=-omega, while the Fig. 1 caption describes a counter-clockwise polarization rotation by the same angle beta; since positive beta is defined as counter-clockwise, these statements appear to conflict unless 'same angle' means same magnitude rather than same signed angle. Please harmonize the wording.","section":"Section 2 and Fig. 1"},{"comment":"The caption says all numbers include internal iterative delensing self-consistently, but the surrounding text emphasizes the delensing procedure mainly for the deep configuration; please state explicitly which experimental configurations are delensed and which are not.","section":"Table 1 caption"},{"comment":"The signal-to-noise sum starts at L>=30 with no stated Lmax; please specify the maximum multipole used for each configuration.","section":"Eq. (4.1)"},{"comment":"The full-sky simulation validation of the N^(1) predictions is mentioned but not shown; a figure or a short table comparing simulated and predicted spectra would make the claim easier to verify.","section":"Section 3.2"},{"comment":"The caption refers to 'the kappa spectrum and our fiducial post-Born lensing curl spectrum' but does not identify which black line is which; please label the curves directly.","section":"Fig. 2 caption"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope and the forecasting machinery is largely sound, but I would not accept before the authors address the beta=-omega robustness issue and the sign-convention inconsistency between Eqs. (C.1), (C.8), Fig. 1, and the beta=-omega forecasts. The latter is particularly important because, if the equations are taken literally, the rotation-induced B-modes cancel under beta=-omega and the forecasts in Tables 1 and 4 need to be revisited. The dependence of the headline numbers on a relation advocated by two co-authors in earlier papers also makes the requested partial-decorrelation forecast important for the paper's credibility as a test of that relation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a careful forecasting paper, not a breakthrough, but it is genuinely useful. What is new: the joint treatment of the lensing curl ω and polarization rotation β estimators, the N^1 bias formula for polarization rotation (Eq. D.13, with a squeezed-limit approximation), and the forecasts for the β=-ω scenario with realistic LSS templates. The ω-only numbers reproduce Ref [21], which is a good sanity check.\n\nThe paper's central claim is conditional: if β=-ω, SPT-3G-7y reaches S/N≈7.1 in cross-correlation with κ+g+CIB, and CMB-S4-deep ≈38; from CMB alone S4-deep gets ~3.9σ. The assumption is flagged in the introduction — \"we do not enter in this controversy\" — and the authors provide the ω-only column (Table 3) where SPT-3G drops to 4.9. So the central claim is honestly presented as a test, not a derivation.\n\nThe soft spots are the usual ones for forecasts. The β=-ω relation is not derived and no intermediate decorrelation is explored; a modest loss of coherence between β and ω could knock the SPT-3G detection down. The LSS template construction follows Ref [21] and inherits its modeling choices (Halofit, bias model, CIB SED); there is no independent validation of the template bispectrum against simulations, though the CMB-side N^0/N^1 predictions were checked with full-sky simulations and agree well.\n\nOverall the math looks sound. The estimators, responses, and biases are derived carefully and the cross-checks are appropriate. The paper would benefit from a short section on partial β-ω coherence, but that is a completeness request, not a fatal flaw.\n\nThis is a paper for CMB lensing and delensing practitioners. It deserves a serious referee. I would send it to review.","headline":"Solid, honest forecasting paper; the headline S/N numbers rest on the contested β=-ω relation, which the authors flag and provide ω-only alternatives for.","tokens_in":28931,"tokens_out":1876,"would_cite":true,"duration_ms":18009,"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":"This paper asks whether the second-order lensing curl of the CMB, and the polarization rotation it may cause, can be measured by current and planned CMB experiments.","keywords":["cosmic microwave background","CMB lensing","post-Born lensing","lensing rotation","polarization rotation","B-mode polarization","quadratic estimators","large-scale structure cross-correlation"],"falsifier":"Measure the cross-spectrum $C^{\\beta\\omega}_L$ between the reconstructed polarization rotation and lensing rotation maps from the same CMB data. Under $\\beta=-\\omega$ it must equal $-C^{\\omega\\omega}_L$, while under no polarization rotation it is zero; equivalently, the SPT-3G-7y template-cross-correlation S/N should fall near 7.1 in the first case and near 4.9 in the second. A measured value between these, with uncertainties excluding both, would falsify the exact relation.","tokens_in":27743,"feed_emoji":"🌀","tokens_out":14521,"duration_ms":113489,"temperature":0.7,"pith_summary":"This paper asks whether the second-order, “post-Born” curl of CMB lensing—the rotational part of the deflection angle that appears when a photon is deflected more than once—can be measured, and whether that rotation also rotates the polarization of the CMB. The authors build quadratic estimators for the polarization rotation angle $\\beta$ and the lensing rotation angle $\\omega$, compute their noise and biases for current and planned experiments, and forecast detections from the CMB alone and from cross-correlating the CMB with large-scale structure templates. Under the contested assumption $\\beta=-\\omega$, they find that SPT-3G after its nominal seven-year survey will detect the combined signal at signal-to-noise $\\approx 7$ through LSS cross-correlation, and that CMB-S4 deep will reach $\\approx 38$; from the CMB alone, CMB-S4 deep reaches about $3.9\\sigma$. If these forecasts hold, the open dispute over whether the lensing curl rotates polarization can be decided with data already being collected.","feed_headline":"Post-Born lensing rotation is forecast at 7σ with SPT-3G","feed_subtitle":"Cross-correlating CMB maps with galaxy and CIB templates can settle whether lensing rotation also rotates polarization.","key_machinery":"The machinery is a pair of optimal quadratic estimators, one for polarization rotation $\\beta$ and one for the lensing curl $\\omega = -\\frac{1}{2}\\epsilon^{ij}\\nabla_i \\alpha_j$, constructed as likelihood gradients of inverse-variance-filtered CMB polarization maps. In the squeezed limit, a long-wavelength rotation of the image and a long-wavelength rotation of the polarization produce the same local EB power, so the estimators are degenerate at the dipole and only partially separated at $L\\geq 2$ by shear/B-mode information; the paper quantifies this with response functions $R^{\\beta\\beta}$, $R^{\\omega\\omega}$, $R^{\\beta\\omega}$ and Gaussian noise $N^{(0)}_L$, plus the lensing-induced $N^{(1)}_L$ bias. It then applies iterative internal delensing, which reduces both $N^{(0)}_L$ and $N^{(1)}_L$ by orders of magnitude, and builds LSS rotation templates from the bispectrum $b^{\\omega ij}$ of $\\omega$ with convergence, galaxy, and CIB tracers, characterized by a correlation coefficient $F_L$ that enters the forecast signal-to-noise.","core_discovery":"At second order in gravitational lensing, the deflection angle acquires a curl component $\\omega$, and a separate body of work claims this curl also rotates the polarization by $\\beta=-\\omega$. The paper's central computational claim is that the two effects can be measured jointly with quadratic estimators acting on the local EB (E-mode/B-mode) polarization cross-power they create, and that the $\\beta=-\\omega$ combination is much easier to detect than the curl alone because the B-mode signals add coherently. Concretely, the paper forecasts S/N $\\approx 7.1$ for SPT-3G-7y, $\\approx 38.3$ for CMB-S4 deep, and $\\approx 39.4$ for PICO when the CMB rotation estimators are cross-correlated with templates built from CMB lensing convergence, galaxy clustering, and the cosmic infrared background; without polarization rotation the same SPT-3G measurement yields S/N $\\approx 4.9$. A purely internal CMB detection remains marginal, at $3.9\\sigma$ for CMB-S4 deep even with iterative delensing.","pith_inferences":["If the true $\\beta$ and $\\omega$ are only partially anti-correlated, the forecast S/N lies between the $\\beta=0$ and $\\beta=-\\omega$ endpoints (4.9 and 7.1 for SPT-3G-7y); the paper does not compute that interpolation, so its headline numbers represent the maximum under the contested relation.","A null $\\beta$-$\\omega$ cross-correlation would not refute the post-Born curl; it would refute the claim that the curl rotates polarization, effectively deciding the dispute in favor of the no-rotation calculations.","The same estimator formalism can separate lensing rotation from other rotation sources such as cosmic birefringence or Faraday rotation using the frequency and redshift dependence of the latter, extending the method beyond the lensing question.","Using redshift-resolved galaxy samples or deeper convergence maps could push the LSS template correlation above the $\\sim 0.8$ large-scale value the paper finds for CMB-S4 deep, strengthening the forecast further."],"forward_implications":["SPT-3G-7y should detect the combined lensing-rotation signal at S/N $\\approx 7.1$ with LSS templates if $\\beta=-\\omega$, so the rotation question can be addressed with data now being collected.","CMB-S4 deep reaches S/N $\\approx 38.3$ and PICO $\\approx 39.4$ under the same assumption, making the post-Born curl a measurable signal rather than a theoretical correction.","A CMB-only measurement stays marginal at $3.9\\sigma$ for CMB-S4 deep, so delensing and external templates are required for a high-significance detection.","If polarization rotation is absent, SPT-3G-7y still detects the image rotation at S/N $\\approx 4.9$ with the same templates, so the curl itself should be detected either way.","The $\\beta$ and $\\omega$ estimators are nearly degenerate on the largest scales and correlated elsewhere, so any joint measurement must include their cross-response to avoid misattributing signal."],"supporting_citations":[{"why":"Predicts that the post-Born lensing curl rotates CMB polarization by $\\beta=-\\omega$; the paper's main forecasts adopt this relation.","marker":"[18–20]"},{"why":"Pioneering forecast for detecting lensing rotation whose LSS tracer choices and template construction the paper follows.","marker":"[21]"},{"why":"Provides the General Minimum Variance quadratic estimators the paper uses for all reconstructions.","marker":"[30]"},{"why":"Introduces the flat-sky quadratic estimator for polarization rotation that this paper generalizes to joint $\\beta$ and $\\omega$ reconstruction.","marker":"[31]"},{"why":"Provides the full-sky formalism for polarization rotation estimators used here.","marker":"[32]"},{"why":"Supplies the spherical bispectrum expansion and squeezed-limit response rules used for noise and $N^{(1)}$ calculations.","marker":"[35]"},{"why":"Defines the iterative internal delensing procedure used to reduce reconstruction noise and biases.","marker":"[38]"},{"why":"Validates iterative delensing on simulated CMB-S4 maps and supplies the deep-patch noise model adopted in forecasts.","marker":"[40]"},{"why":"Provides the flat-sky $N^{(1)}$ bias expressions for lensing estimators that the paper adapts to rotation estimators.","marker":"[36]"},{"why":"Shows that post-Born curl B-modes are detectable and supports the flat-sky LSS template approximations.","marker":"[49]"}],"fun_headline_variants":["Lensing rotation forecast at 7σ with SPT-3G cross-correlation","Post-Born lensing rotation: 7σ prediction for SPT-3G","CMB rotation controversy: joint estimators predict 7σ SPT-3G","Cross-correlate CMB with galaxies to settle lensing rotation","SPT-3G will test if lensing rotates polarization at 7σ"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The headline forecasts assume the exact relation $\\beta=-\\omega$, meaning the polarization rotation and lensing rotation fields have equal power and are perfectly anti-correlated; the paper adopts this contested relation from other work without deriving it and does not show how the forecasts degrade under partial decorrelation.","fun_headline_variants_meta":{"raw":{"variants":["Lensing rotation forecast at 7σ with SPT-3G cross-correlation","Post-Born lensing rotation: 7σ prediction for SPT-3G","CMB rotation controversy: joint estimators predict 7σ SPT-3G","Cross-correlate CMB with galaxies to settle lensing rotation","SPT-3G will test if lensing rotates polarization at 7σ"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000997,"raw_usage":{"total_tokens":4225,"prompt_tokens":954,"completion_tokens":3271,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":570,"completion_tokens_details":{"reasoning_tokens":3180}},"tokens_in":570,"tokens_out":3271,"duration_ms":23165,"temperature":1.0,"reasoning_tokens":3180,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:40:41.257348+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the cross-spectrum $C^{\\beta\\omega}_L$ between the reconstructed polarization rotation and lensing rotation maps from the same CMB data. Under $\\beta=-\\omega$ it must equal $-C^{\\omega\\omega}_L$, while under no polarization rotation it is zero; equivalently, the SPT-3G-7y template-cross-correlation S/N should fall near 7.1 in the first case and near 4.9 in the second. A measured value between these, with uncertainties excluding both, would falsify the exact relation.","supporting_citations":[{"cited_title":"How to detect lensing rotation","cited_arxiv_id":"2303.13313","evidence_quote":"Pioneering forecast for detecting lensing rotation whose LSS tracer choices and template construction the paper follows."}],"review_version":1}