{"id":"90d836a6-79be-401c-9a0d-b8baea3db633","arxiv_id":"2608.01844","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Metastable dark energy, where the vacuum energy decays at a constant rate, fits the data but is not required: some combinations show 2-3 sigma hints of decay, none decisively beats the cosmological constant.","lead":"Tests whether dark energy can slowly decay like a radioactive substance, using the newest galaxy clustering, supernova, and cosmic microwave background data. The results allow dying dark energy but do not prove it is real, and show that full galaxy-shape data can tell decay scenarios apart.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"FS likelihood for Models 2/3 is built on standard EFT kernels; the modified DM/DR perturbation equations (Eqs. 13,17) alter second-order time evolution, so the reported >2σ FS deviations may be biased unless validated.","rationale":"The paper's central claim—that metastable DE remains viable and that no decisive evidence for Γ/H0≠0 exists—is supported by the BAO/SN/CMB constraints and is not endangered by the FS issue. However, the paper's boldest new claim is that DESI DR1 FS discriminates the decay channels and that Model 1 shows a >2σ deviation. That claim depends entirely on the correctness of the FS likelihood for modified perturbation equations. The standard EFT templates in velocileptors assume the usual second-order kernels and time-rescaled growth; Eqs. (13)–(14) and (17) add scale-independent source/damping terms that alter the time evolution of δ_2 relative to the standard D^2 scaling. For couplings Γ/H0 ~ 0.2, these corrections can be at the percent level, comparable to the statistical errors on the FS multipoles (which drive Γ/H0 uncertainties of ~0.05–0.1). The paper's own caveat in Section VI about projection effects documents that FS nuisance marginalization is delicate for extended models, but it does not address the kernel-level validity. Therefore, the FS discrimination result should be regarded as conditional until validated. This does not overturn the paper's main conclusion, so I leave the reader's CONDITIONAL verdict unchanged.","tokens_in":24764,"tokens_out":16078,"duration_ms":197519,"concrete_test":"Derive the exact second-order density and velocity kernels from Eqs. (13)–(14) for Model 2 and implement them in a standalone time-dependent SPT code. Compute the one-loop matter power spectrum for a fiducial cosmology with Γ/H0=0.2 and compare against the velocileptors prediction used in the paper. If the monopole/quadrupole differ by more than ~1% at k<0.2 h/Mpc (roughly the DESI DR1 FS precision), then the reported Γ/H0 constraints in Table VII are biased and the FS discrimination claim is not supported. Alternatively, run a COLA/MG-GADGET simulation with the damping term to validate the EFT template.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section VI applies the DESI DR1 full-shape likelihood via velocileptors to Models 1–3. The pipeline was calibrated for standard ΛCDM-like tracers. For Model 2, the modified DM density equation (Eq. 13) contains a scale-independent damping term -aΓ(ρ_DE/ρ_DM)δ_DM, which changes the time evolution of δ_DM and hence the second-order perturbative kernels used in the one-loop EFT. Velocileptors assumes the standard kernels with a common growth factor; it does not account for the modified time dependence of δ_2 and θ_2. For Γ/H0 ~ 0.2 (the reported 1σ range), the linear growth is modified at the ~10% level and the second-order kernels can shift the one-loop P(k) by a comparable fraction of the statistical error, biasing Γ/H0. The paper acknowledges projection effects in Section VI but does not validate the pipeline against simulations or an exact PT calculation. The FS-based claim (Model 1 >2σ, Models 2–3 within 2σ) is therefore not yet robust.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constrains three metastable dark-energy scenarios governed by a constant radioactive-like decay rate Γ: an effective exponentially decaying DE component (Model 1), decay of DE into non-baryonic dark matter (Model 2), and decay of DE into dark radiation (Model 3). Using DESI DR2 BAO, three SNIa compilations, Planck or P-ACT CMB data, a BBN prior, and—for the first time—DESI DR1 full-shape clustering, the authors find that BAO+SNIa combinations mildly prefer positive Γ/H0 at the ≳2σ level, while CMB-inclusive combinations are generally consistent with Γ/H0=0. The full-shape analysis is used to argue that the three decay channels have distinct growth signatures, with Model 1 showing the largest residual deviation (Γ/H0=0.163±0.055 for Planck+FS+DES-Dovekie) while Models 2 and 3 are reported as consistent with ΛCDM within 2σ. The overall conclusion is that metastable DE remains phenomenologically viable but not decisively required. The paper includes modified CLASS perturbation equations, extensive tables of marginalized constraints, and Δχ²/DIC model comparisons.","tokens_in":25187,"tokens_out":6224,"duration_ms":78612,"significance":"If the full-shape pipeline is valid for the modified perturbation equations, the paper makes a useful contribution: it extends metastable-DE constraints beyond background probes and demonstrates that growth/clustering data can in principle discriminate decay channels. The treatment is honest in the important respect that Γ/H0 is fitted rather than predicted, and reconstructed derived quantities such as Om(z), w_DE(z), and fσ8(z) are not presented as independent confirmations; I found no equation-level circularity. The comprehensive Tables II–VII and the inclusion of multiple SNIa and CMB combinations are strengths. The main significance risk is the unvalidated use of standard full-shape EFT kernels for Models 2 and 3, which is load-bearing for the claimed FS discrimination.","major_comments":[{"comment":"The FS discrimination claim rests on applying the DESI DR1 full-shape likelihood through velocileptors, whose one-loop EFT kernels assume standard ΛCDM second-order growth. For Model 2, Eq. (13) adds a scale-independent damping term -aΓ(ρ_DE/ρ_DM)δ_DM to the DM density contrast; for Model 3, Eqs. (17)–(18) modify the DR perturbation hierarchy with analogous -aΓ(ρ_DE/ρ_DR) terms. These changes alter the time evolution of δ and θ relative to the standard kernels. The paper notes possible projection effects in Section VI, but it does not validate the pipeline against N-body simulations or an exact perturbation-theory calculation. The reported values such as Γ/H0=0.163±0.055 (Model 1) and 0.21±0.10 (Model 2) in Table VII are therefore not yet shown to be unbiased. Since the central claim that FS discriminates the three decay channels depends on this, the FS-based conclusions are not robust w","section":"Section VI, Eqs. (13) and (17), Tables VI–VII"},{"comment":"The abstract and Section VIII state that for CMB+FS+DES-Dovekie, Models 2 and 3 remain consistent with the ΛCDM limit within 2σ. Table VII gives Γ/H0=0.21±0.10 for Model 2, which is 2.1σ from zero if the reported 68% interval is interpreted as Gaussian. This contradicts the summary statement and matters because the paper's model-discrimination narrative highlights Model 1 as the only channel with a >2σ residual. The authors should either correct the wording or derive the significance directly from the posterior (e.g. asymmetric or truncated intervals) and report it explicitly.","section":"Abstract, Section VIII, Table VII"},{"comment":"The paragraph acknowledging that 'marginalized FS constraints can be sensitive to projection effects from nuisance-parameter marginalisation' is an important caveat, but it is not turned into a quantitative test. For extended models such as Models 2 and 3, the DESI baseline nuisance parameterization (bias, counterterms, stochastic terms) was developed for standard tracer templates. A concrete robustness check—e.g. analyzing synthetic FS data generated from the modified CLASS power spectra, or comparing the one-loop FS likelihood with a direct perturbation-theory computation—is needed before the claimed Model 1 >2σ deviation and the Model 2/3 discrimination are treated as reliable.","section":"Section VI, 'marginalized FS constraints' caveat"}],"minor_comments":[{"comment":"The notation uses H for both the Hubble rate and the conformal Hubble rate (H=aH). This is confusing in the perturbation equations; suggest using \\mathcal{H} for the conformal quantity.","section":"Section II, Eq. (13)"},{"comment":"The entry '<0.701' for Model 3 in the BBN+DESI row carries an unexplained footnote marker '1'. Either define the footnote or remove the marker; similar markers appear in other tables.","section":"Table II"},{"comment":"The phrase 'a slight deviation from Γ/H0=0 at the ≳2σ level' is internally awkward: a 2.96σ deviation (0.163/0.055) is not 'slight'. Rephrase to 'a deviation at the ~3σ level' or similar.","section":"Abstract"},{"comment":"The DESI full-shape fσ8(z) points are described as 'extracted using the ShapeFit compression', but the figure does not state whether these are measured values with error bars or reconstruction-band inputs. Clarify the provenance and error treatment in the caption.","section":"Section V, Fig. 8"},{"comment":"The CLASS modification is described only by reference; no code release or validation against an independent Boltzmann code is indicated. A short validation appendix (e.g. reproducing ΛCDM limits and checking energy conservation) would strengthen reproducibility.","section":"Section III"}],"recommendation":"major_revision","confidential_remarks":"The central background-level constraints and the no-detection conclusion are likely sound, but the full-shape analysis is the main risk. The paper would be acceptable after either validating the FS pipeline for the modified perturbation equations or substantially softening the FS-discrimination claims. The 2σ inconsistency for Model 2 in the abstract is a fixable wording issue, but it should be corrected before publication. The manuscript fits the journal's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a competent, well-hedged constraints paper, not a discovery claim. The central result—current data allow late-time DE decay but do not provide decisive evidence for a nonzero decay rate—is supported by the tables and the abstract is honest about it. The new piece is the first application of DESI DR1 full-shape measurements to these metastable DE models, together with the perturbation-level implementation of the three decay channels in CLASS. That is real work and a useful step beyond earlier background-only studies.\n\nWhat it does well: the model taxonomy is clear (effective decaying DE, decay into DM, decay into DR), the linear-perturbation equations are laid out, and the data handling is standard. The BAO+SN mild preference for positive Γ/H0 (around 2σ) versus CMB-included consistency with zero is presented carefully, and the DIC/Δχ² values are modest—no overclaiming. Γ/H0 is honestly fitted, not presented as a prediction; the reconstructed w(z), Om(z), and fσ8(z) are derived, not independent confirmations.\n\nThe soft spot is the FS analysis. The paper applies the DESI DR1 full-shape likelihood through velocileptors, which assumes standard EFT perturbation kernels. For Model 2, the modified DM density equation contains a scale-independent damping term that changes the time evolution of δ_DM and hence the second-order kernels; velocileptors doesn't account for that. The paper does flag projection effects from nuisance marginalisation in Section VI, but that's a related yet different issue. There's no validation against simulations or an exact perturbation calculation, and no code or configuration files are released. So the reported >2σ FS discrimination between models (Model 1 outside 2σ, Models 2–3 within) is the least robust part of the paper. This is a legitimate concern, not a demonstrated error. It doesn't undermine the central viability conclusion, which rests mainly on distance and CMB data, but it does mean the FS-based discrimination claim should be treated as suggestive until validated.\n\nWho it's for: anyone tracking dark-energy constraints with DESI, especially people building growth-data analyses. It deserves a serious referee. The issues are fixable: validate the FS pipeline for extended models, release the modified CLASS code, or soften the FS-discrimination wording. I'd send it to review, not desk-reject.","headline":"Solid constraints study; the central viability conclusion holds up, and the first DESI DR1 full-shape constraints are the new contribution, but the FS discrimination claim needs validation before it carries weight.","tokens_in":25628,"tokens_out":3118,"would_cite":true,"duration_ms":33309,"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":"Metastable dark energy—a dark-energy component that decays at a constant rate like a radioactive substance—remains a viable explanation of cosmic acceleration, with current data showing only mild, dataset-dependent hints of a nonzero decay","keywords":["metastable dark energy","dark energy decay","DESI BAO","full-shape clustering","cosmological perturbations","structure growth","dark matter interaction","dark radiation"],"falsifier":"Generate mock DESI-like full-shape data from a known metastable model with a chosen nonzero Γ/H0, then re-analyse with the standard full-shape likelihood; if the recovered Γ/H0 is biased by more than the reported uncertainty, the claimed 2–3σ deviations are pipeline artefacts. Alternatively, a DESI DR2 full-shape measurement with the same modelling that returns Γ/H0 consistent with zero for Model 1 would directly undercut the paper's strongest deviation.","tokens_in":24680,"feed_emoji":"🔭","tokens_out":6215,"duration_ms":67850,"temperature":0.7,"pith_summary":"This paper tests whether dark energy could be metastable, decaying at a constant rate like a radioactive substance, rather than being an exactly constant cosmological constant. It constrains three decay scenarios—fading dark energy, decay into dark matter, and decay into dark radiation—using DESI DR2 baryon acoustic oscillations, supernovae, CMB data, and, for the first time, DESI DR1 full-shape galaxy clustering. The late-time distance data mildly prefer a positive decay rate, which would make dark energy quintessence-like at low redshift, but including CMB data pulls the decay rate back to zero. The full-shape clustering data discriminate the decay channels: the pure-fading model shows a deviation beyond 2σ in the combined analysis, while the dark-matter and dark-radiation channels remain consistent with ΛCDM within 2σ. The paper concludes metastable dark energy is still viable, but there is no decisive evidence for a nonzero decay rate.","feed_headline":"Metastable dark energy survives DESI DR2 and full-shape tests","feed_subtitle":"New distance and clustering data show decaying-dark-energy models remain consistent, with only mild hints of a nonzero decay rate.","key_machinery":"The central object is the dimensionless decay rate Γ/H0, defined by the radioactive-like law ρ_DE ∝ exp(-Γt) for the dark-energy density. This single parameter controls three different physical channels: Model 1 lets the dark-energy density fade with an effective equation of state w = -1 + Γ/(3H); Model 2 transfers energy to non-baryonic dark matter, altering both the background matter abundance and the dark-matter perturbation equations; Model 3 creates a dark-radiation component with no primordial abundance. The analysis machinery consists of a modified Boltzmann solver that evolves the linear perturbations for each channel and feeds the power spectra and growth quantities into the DESI DR","core_discovery":"The paper establishes that a constant 'radioactive' decay rate Γ for dark energy, measured in units of the Hubble rate H0, is compatible with the full current data set. For distance data alone (DESI DR2 BAO plus supernovae), the posterior for Γ/H0 shifts positive by about 2σ, corresponding to a dark-energy density that decreases with time and an effective equation of state w > -1 at low redshift. Adding CMB data from Planck or Planck+ACT removes this preference, making Γ/H0 consistent with the ΛCDM value of zero. The DESI DR1 full-shape analysis, which is new for these models, breaks the degeneracy between decay channels: decaying into dark matter changes the matter abundance and perturbatio","pith_inferences":["A natural next test is to apply the same full-shape pipeline to DESI DR2 full-shape data when they become public; if the Model 1 deviation grows or shrinks, it will separate a real background-level decay from the projection effects the paper flags in Section VI.","The paper's three scenarios could be embedded in a broader framework where dark energy also couples to baryons or where the decay rate varies in time; those extensions would break the current degeneracy between Γ/H0 and nuisance parameters.","The projection-effect caveat could be checked directly with mock catalogs: generate DESI-like clustering data from a known metastable model and see whether the standard nuisance parameterization recovers the input Γ/H0. This is a falsifiable prediction of the analysis pipeline itself.","If future data confirm positive Γ/H0 in Model 1, a single constant decay rate would be hard to distinguish from a slowly rolling quintessence field using distances alone; growth and clustering data would then be the primary way to tell the two apart."],"forward_implications":["If the mild positive Γ/H0 preference is real, dark energy behaves like an unstable component with a half-life comparable to the Hubble time, producing a quintessence-like effective equation of state at low redshift.","CMB data are the main anchor: any metastable model that noticeably adds dark radiation or changes the early matter density faces strong constraints, so late-time hints must survive CMB+SN combination to matter.","Full-shape clustering provides a growth-based discriminator: the dark-matter decay channel (Model 2) leaves the largest imprint on the power-spectrum amplitude and scale dependence, while dark-radiation and pure-fading channels are harder to see.","Model 1's >2σ deviation in the combined analysis, if confirmed, would point to a background-level modification of dark energy rather than an interacting dark sector, since Models 2 and 3 stay within 2σ.","Metastable dark energy fits the data slightly better than ΛCDM in several combinations (e.g., Δχ² about -12 for BBN+DESI+DES-Dovekie), but the improvement is not decisive after penalizing model complexity."],"fun_headline_variants":["Dark energy decay: no decisive signal in DESI data","Metastable DE passes DESI DR2 and full-shape tests","New data: decaying dark energy still compatible","DESI full-shape data keep metastable DE alive","Radioactive dark energy: evidence fades with CMB"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The full-shape analysis assumes the galaxy bias and noise model tuned for standard cosmology remains valid when dark-matter or dark-radiation perturbation equations are modified—a caveat the paper itself raises in Section VI, where it notes marginalized full-shape constraints can be sensitive to projection effects from nuisance-parameter marginalisation.","fun_headline_variants_meta":{"raw":{"variants":["Dark energy decay: no decisive signal in DESI data","Metastable DE passes DESI DR2 and full-shape tests","New data: decaying dark energy still compatible","DESI full-shape data keep metastable DE alive","Radioactive dark energy: evidence fades with CMB"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000287,"raw_usage":{"total_tokens":1597,"prompt_tokens":891,"completion_tokens":706,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":635,"completion_tokens_details":{"reasoning_tokens":625}},"tokens_in":635,"tokens_out":706,"duration_ms":8051,"temperature":1.0,"reasoning_tokens":625,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T19:48:11.312494+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Generate mock DESI-like full-shape data from a known metastable model with a chosen nonzero Γ/H0, then re-analyse with the standard full-shape likelihood; if the recovered Γ/H0 is biased by more than the reported uncertainty, the claimed 2–3σ deviations are pipeline artefacts. Alternatively, a DESI DR2 full-shape measurement with the same modelling that returns Γ/H0 consistent with zero for Model 1 would directly undercut the paper's strongest deviation.","supporting_citations":[],"review_version":1}