{"id":"0a073e0a-612d-4c2e-8a7e-caa059b52d5a","arxiv_id":"2412.00826","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"An automatic differentiation scheme constructs CMB-lite likelihoods from one minimisation and one Hessian evaluation, reproducing full multi-frequency posterior constraints on SPT-3G data within 0.1 sigma.","lead":"This paper shows how to build compressed cosmic microwave background likelihoods in about a minute using automatic differentiation, replacing slow Monte Carlo sampling. It releases a fast 'lite' likelihood for the SPT-3G 2018 data that matches the cosmological constraints of the full multi-frequency analysis.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Hessian covariance is a Laplace approximation with no general error bound; the concern is real but empirically tested for SPT-3G, so the verdict stands.","rationale":"The reader's weakest_assumption correctly identifies the Laplace approximation as the most load-bearing assumption. I agree: the Hessian-based covariance is exact only for Gaussian posterior distributions, and no general error bound is provided. However, the paper does not ignore this limitation. It flags it explicitly in Section 3, quantifies the systemic bias with scatter-free mocks (best-fit shifts <0.01 sigma, error bars <1%), and validates the full pipeline with 1000 Gaussian realisations, finding realisation-dependent scatter confined to within ±0.1 sigma. These checks are direct and appropriate for the SPT-3G 2018 dataset, which is the stated application. The central claim is therefore supported for the demonstrated case, and the residual concern is a well-acknowledged limitation rather than a flaw. The reader's MODERATE confidence already reflects the absence of a general proof and the lack of independent code execution. A conditional acceptance requiring an additional non-Gaussian test would be defensible, but the paper's empirical validation meets the standard for this methodological contribution. I therefore recommend leaving the verdict unchanged.","tokens_in":11376,"tokens_out":5686,"duration_ms":53195,"concrete_test":"Generate a mock dataset from the multi-frequency likelihood using a fiducial cosmology and a foreground model with a running spectral index (or otherwise more nonlinear than the fiducial SPT-3G model). Run the reconstruction to obtain the Hessian covariance, then sample the reconstruction likelihood with an MCMC (e.g., Hamiltonian Monte Carlo) and compute the sample covariance of the CMB band powers. Compare the two covariance matrices: if the relative difference in the largest eigenvalue or in the diagonal errors exceeds 10%, the Laplace approximation is not adequate for that regime. Repeating this over a range of foreground spectral indices would delineate the method's validity.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The covariance of the CMB-only band powers is obtained from the inverse of the Hessian of the reconstruction likelihood at its best-fit point (Section 3, step 2). This is a Laplace approximation: it equals the Fisher matrix only if the likelihood is exactly Gaussian in all parameters (CMB band powers plus 33 nuisance parameters). The paper explicitly acknowledges in Section 3 that non-Gaussian contributions are not captured, and it offers empirical checks (scatter-free mocks and 1000 Gaussian realisations) for SPT-3G 2018, but it provides no general bound on the approximation error. For applications where foreground parameters are nonlinearly related to the data or poorly constrained, the Hessian at the best fit can misestimate the true marginal covariance of the CMB band powers, biasing the lite likelihood's parameter errors. This is the weakest link in the central claim that one minimisation plus one Hessian evaluation yields a sampling noise-free covariance.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an automatic-differentiation-based implementation of the CMB-lite framework. Using the differentiable SPT-3G 2018 TT/TE/EE likelihood from the candl library, the author minimizes the reconstruction likelihood once and evaluates its Hessian to obtain foreground-marginalised CMB-only band powers and their covariance. The resulting lite likelihood is benchmarked against the multi-frequency likelihood for ΛCDM, ΛCDM+Neff, and ΛCDM+AL, finding best-fit shifts below 0.1σ and posterior-width ratios within 10%. The author also performs scatter-free mock reconstructions and 1000 Gaussian realisations to quantify systematic and realisation-dependent biases, and publicly releases the lite likelihood and a construction notebook.","tokens_in":11590,"tokens_out":4689,"duration_ms":43633,"significance":"If the result holds, this is a practical and timely contribution: it reduces lite-likelihood construction from expensive MCMC sampling to about a minute on a personal computer and removes sampling noise from the covariance estimate. The paper is transparent about its approximations, including the Laplace approximation for the covariance, the treatment of frequency-dependent window functions, the omission of beam terms in the lite likelihood, and the representation of super-sample lensing through an additive covariance term. The empirical validation is a strength: scatter-free mocks, 1000 Gaussian realisations, three cosmological models, and public code. The main limitation is that no general error bound is given for the Laplace approximation, but the paper does not overclaim beyond the SPT-3G 2018 data set and the tests support the central claim for that application.","major_comments":[],"minor_comments":[{"comment":"The text contains spacing/formatting artifacts such as 'TT /TE/E E' and 'Elikelihood'; these should be corrected to 'TT/TE/EE' and 'EE likelihood'.","section":"Abstract and Section 1"},{"comment":"Please specify explicitly whether the Hessian is evaluated for the log-likelihood or the log-posterior, since priors on nuisance parameters such as the calibration parameters affect the resulting covariance matrix.","section":"Section 3, step 2"},{"comment":"The phrase 'systemic bias' appears twice (in the scatter-free mock paragraph and in Figure 4); the intended term is 'systematic bias'.","section":"Section 4"},{"comment":"There is a typo: 'Mulitpole' should be 'Multipole'.","section":"Figure 1 caption"},{"comment":"The header 'ln(1010A)' omits the subscript 's'; it should read 'ln(10^10 A_s)' to match the text.","section":"Table 1"}],"recommendation":"accept","confidential_remarks":"The paper fits the journal scope and the public release of code strengthens reproducibility. No concerns about novelty or citation practice."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Lennart's paper is a clean, practical advance: it reduces CMB-lite reconstruction to one minimization plus one Hessian evaluation, runs in about a minute on a laptop, and eliminates the sampling noise that plagues MCMC-based covariance estimates. That is genuinely new relative to D13 and Prince et al. 2024, who sample the reconstruction likelihood. The validation is solid: scatter-free mocks, 1000 Gaussian realisations, posterior comparisons for three cosmological models, and the agreement is within the claimed 0.1 sigma and 10% error margins. The lite likelihood and notebook are public, so the work is reproducible. It also does a good job of discussing the real approximations: window function mismatch, beam term, aberration, super-sample lensing. That transparency is worth something.\n\nThe soft spot is exactly what the stress-test says: the Hessian at the best fit gives a Laplace-approximation covariance. The paper says it doesn't capture non-Gaussian contributions and then validates empirically for SPT-3G 2018. No general bound, and the mocks are Gaussian, so non-Gaussianity in foregrounds isn't stressed. For this dataset and this foreground model, the checks are convincing, but a reader applying this to a different experiment should redo the validation. That's a real limitation, not a fatal flaw. The central claim holds.\n\nI'd send it to peer review. It's a methods paper with a reproducible result and honest limitations. The Laplace approximation point is worth asking the author to state more prominently, maybe with a small set of experiments that stress the non-Gaussian regime, but it doesn't undermine the paper.","headline":"Clean AD-based CMB-lite construction that actually works on SPT-3G data; the Laplace-approximation caveat is real but empirically tested.","tokens_in":12015,"tokens_out":2502,"would_cite":true,"duration_ms":19263,"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":"A CMB-lite likelihood can be assembled in about a minute on a laptop by replacing sampling-based reconstruction with one minimisation and one Hessian evaluation, producing a sampling-noise-free covariance and recovering the reference…","keywords":["CMB-lite","likelihood compression","automatic differentiation","cosmic microwave background","band powers","foreground marginalisation","covariance estimation","SPT-3G"],"falsifier":"Construct a synthetic multi-frequency data set whose reconstruction posterior is visibly non-Gaussian, for instance low signal-to-noise band powers or strong degenerate foregrounds, then compare parameter constraints from the Hessian-based lite likelihood against those from a long, converged MCMC sampling of the same reconstruction likelihood; if best-fit shifts or error-bar ratios exceed $0.1\\sigma$ and $10\\%$, the core Gaussian assumption fails.","tokens_in":11124,"feed_emoji":"🌌","tokens_out":7999,"duration_ms":87300,"temperature":0.7,"pith_summary":"CMB experiments observe the sky at several frequencies, so their power-spectrum likelihoods carry both cosmological information and nuisance parameters describing foregrounds and calibration. The CMB-lite idea is to first compress those multi-frequency measurements into a shorter vector of foreground-marginalised, CMB-only band powers, which can then be analysed quickly with any cosmological model. This paper claims that the expensive and noisy Markov-chain reconstruction step can be replaced by a single minimisation of the reconstruction likelihood followed by one evaluation of its Hessian, using automatic differentiation, so the whole compressed likelihood is built in about a minute on a personal computer with no sampling noise in its covariance. Applied to the 2018 SPT-3G temperature and polarisation data, the resulting lite likelihood recovers the cosmological constraints of the full multi-frequency likelihood for Lambda CDM and two extensions, with best-fit shifts below 0.1 sigma and error bars matching to better than 10 percent. If it holds, this makes lite likelihood construction cheap enough to be rerun routinely for robustness tests and for the much larger, lower-noise data sets now being planned.","feed_headline":"CMB-lite likelihoods in a minute: one minimisation, one Hessian","feed_subtitle":"Builds the compressed likelihood in about a minute and matches full multi-frequency constraints to 0.1 sigma.","key_machinery":"The object that carries the argument is the Gaussian reconstruction likelihood $\\mathcal{L}_{\\rm recon}(\\mathbf{C}^{\\rm CMB},\\boldsymbol{\\theta})$ over CMB-only band powers and nuisance parameters, together with the identity connecting its Hessian at the best-fit point to the covariance: the negative inverse Hessian of a Gaussian likelihood is its covariance, equivalently the Hessian is the negative Fisher matrix. Because the likelihood is written in a differentiable form, its gradient and Hessian can be evaluated exactly and cheaply; a truncated-Newton minimiser finds the best fit, and the Hessian at that point gives the sampling-noise-free covariance of the compressed band powers. This replaces the alternating Gibbs-sampling scheme originally used for CMB-lite construction, which is what removes the computational cost and the sampling noise.","core_discovery":"The central claim is that the CMB-lite reconstruction, the step that turns multi-frequency band powers into foreground-marginalised CMB-only band powers and their covariance, can be performed from one minimisation and one Hessian evaluation rather than from Markov-chain Monte Carlo sampling. Writing the reconstruction likelihood over both the CMB band powers and the nuisance parameters, the author minimises the joint likelihood to obtain the compressed band powers, then evaluates the Hessian at the best-fit point; the negative inverse Hessian is the covariance of the compressed band powers, with nuisance marginalisation folded in. The paper demonstrates this on the differentiable SPT-3G 2018 TT/TE/EE likelihood: the compressed data vector shrinks from 728 to 123 bins, the nuisance parameters from 33 to two, likelihood evaluation is about four times faster, and marginalised posteriors for $\\Lambda$CDM, $\\Lambda$CDM+$N_{\\rm eff}$, and $\\Lambda$CDM+$A_L$ agree with the reference likelihood to within $<0.1\\sigma$ in best-fit shifts and $<10\\%$ in error bars. One thousand Gaussian mock realisations show that the observed shifts are consistent with the expected realisation-dependent bias.","pith_inferences":["If the Gaussian (quadratic) approximation of the reconstruction likelihood degrades in low-signal-to-noise regimes or with strongly non-Gaussian foregrounds, the inverse-Hessian covariance could understate the true uncertainty; a direct comparison with a converged MCMC reconstruction on such a case would bound the error.","The same single-minimisation recipe could be applied to any differentiable likelihood, potentially producing 'lite' versions of other nuisance-heavy cosmological data vectors such as galaxy clustering or lensing measurements.","Because no cosmological model enters the reconstruction, the compressed band powers and covariance form a model-independent data product that can be reused for many cosmological models without rebuilding.","A natural testable extension is to replace the fixed Hessian covariance with a Laplace-corrected or non-Gaussian covariance for cases where the realisation-dependent bias exceeds the quoted thresholds."],"forward_implications":["Lite likelihood construction goes from a high-dimensional sampling problem to a local optimisation problem, so it can be repeated on a laptop in about a minute for different data cuts, foreground models, or robustness tests.","The covariance of the CMB-only band powers no longer carries sampling noise, so subsequent cosmological analyses avoid the need for sampling-noise corrections in this step.","Because the compressed data vector has 123 bins and two nuisance parameters instead of 728 bins and 33 nuisance parameters, MCMC exploration of cosmological parameters is faster and the likelihood is easier to interpret.","Parameter constraints from the lite likelihood are claimed to match the full multi-frequency likelihood within $<0.1\\sigma$ best-fit shifts and $<10\\%$ error bars for the three models tested, with the realisation-dependent offset shown to be statistically normal.","The recipe should transfer to future multi-frequency experiments with many channels, where the reduction in data-vector length and nuisance dimensionality is larger."],"supporting_citations":[{"why":"Introduces the CMB-lite framework and the alternating Gibbs-sampling reconstruction process that this paper replaces.","marker":"D13"},{"why":"Supplies the SPT-3G 2018 TT/TE/EE data, the multi-frequency likelihood, and the differentiable likelihood library used for the demonstration.","marker":"Balkenhol et al. (2023)"},{"why":"Provides the Fisher-information relation that identifies the negative inverse Hessian at the best fit with the parameter covariance.","marker":"Heavens et al. (2014)"},{"why":"Supplies the automatic differentiation framework that makes exact gradient and Hessian evaluations possible.","marker":"Bradbury et al. (2018)"},{"why":"Documents the wider use of automatic differentiation in cosmology and the behaviour of second derivatives relied on here.","marker":"Campagne et al. (2023)"},{"why":"Quantifies the sampling-noise penalty in covariance estimates that the Hessian-based construction avoids.","marker":"Hartlap et al. (2007)"},{"why":"Points out that differing window functions across frequencies mean no single set of CMB-only band powers exists, motivating the paper's window-function combination.","marker":"Prince et al. (2024)"}],"fun_headline_variants":["Autodiff CMB-lite: one minimisation, one Hessian, one minute","One minimisation + one Hessian builds CMB-lite likelihood in a minute","Automatic differentiation shrinks CMB-lite construction to a minute","From 728 to 123 bins: CMB-lite likelihoods via autodiff","CMB-lite likelihoods: one step, one minute, <0.1 sigma shift"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the reconstruction likelihood is sufficiently Gaussian near its best fit that the negative inverse Hessian there is an unbiased covariance for the CMB-only band powers, including nuisance marginalisation; the paper validates this on the SPT-3G data and Gaussian mocks but does not give a general bound on the approximation error.","fun_headline_variants_meta":{"raw":{"variants":["Autodiff CMB-lite: one minimisation, one Hessian, one minute","One minimisation + one Hessian builds CMB-lite likelihood in a minute","Automatic differentiation shrinks CMB-lite construction to a minute","From 728 to 123 bins: CMB-lite likelihoods via autodiff","CMB-lite likelihoods: one step, one minute, <0.1 sigma shift"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001144,"raw_usage":{"total_tokens":4793,"prompt_tokens":1039,"completion_tokens":3754,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":655,"completion_tokens_details":{"reasoning_tokens":3644}},"tokens_in":655,"tokens_out":3754,"duration_ms":23439,"temperature":1.0,"reasoning_tokens":3644,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T04:58:06.235467+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct a synthetic multi-frequency data set whose reconstruction posterior is visibly non-Gaussian, for instance low signal-to-noise band powers or strong degenerate foregrounds, then compare parameter constraints from the Hessian-based lite likelihood against those from a long, converged MCMC sampling of the same reconstruction likelihood; if best-fit shifts or error-bar ratios exceed $0.1\\sigma$ and $10\\%$, the core Gaussian assumption fails.","supporting_citations":[{"cited_title":"2023, Phys","cited_arxiv_id":null,"evidence_quote":"Supplies the SPT-3G 2018 TT/TE/EE data, the multi-frequency likelihood, and the differentiable likelihood library used for the demonstration."},{"cited_title":"F., Seikel, M., Nord, B","cited_arxiv_id":null,"evidence_quote":"Provides the Fisher-information relation that identifies the negative inverse Hessian at the best fit with the parameter covariance."},{"cited_title":"2023, The Open Journal of Astrophysics, 6, 15","cited_arxiv_id":null,"evidence_quote":"Documents the wider use of automatic differentiation in cosmology and the behaviour of second derivatives relied on here."}],"review_version":1}