Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

ZTF SN Ia DR2: Improved SN Ia colors through expanded dimensionality with SALT3+

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read Type Ia supernova light curves harbor a third intrinsic variable, x2, that standard SALT fits absorb into the color parameter; ignoring it yields a 0.039\u00b10.005 mag Hubble-residual trend while leaving current distance measurements…

desk verdict A solid, honestly-scoped model-development paper: SALT3+ gives a real but modest second intrinsic component, and the 0.039 mag Hubble-residual trend is a fit summary, not a measurement. read the letter →

arxiv 2502.09713 v1 pith:OH7REBX4 submitted 2025-02-13 astro-ph.CO astro-ph.HE

classification astro-ph.COastro-ph.HE
keywords TypeIasupernovaelight-curvemodelSALT3+supernovastandardizationHubbleresidualscolorsZTFSNDR2secondarymaximum
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Type Ia supernovae are standardizable distance indicators whose light-curve models traditionally encode only two sources of variation: a stretch parameter $x_1$ and a color parameter $c$. This paper trains an extended model, SALT3+, with a third parameter $x_2$, on a large optical-survey sample combined with the previous training compilation, and argues that real supernova light curves contain a coherent extra mode of variability. The mode appears as phase-dependent changes in $g-r$ and $r-i$ colors, coupled to a boost in the height of the i-band secondary maximum and to spectral features such as line velocities. Fits that ignore $x_2$ fold it into the color parameter $c$, and neglecting it in Hubble-residual analyses produces a trend of $0.039\pm0.005$ mag with $x_2$. The authors conclude that extended models improve the accuracy of color measurements and give modest precision gains, but are unlikely to substantially shift current cosmological results.

What carries the argument

The load-bearing object is the extended SALT flux model $F(p,\lambda)=x_0[M_0+x_1M_1+x_2M_2]\cdot\exp(-0.4c\,CL(\lambda))$, where $M_0,M_1,M_2$ are two-dimensional B-spline spectral surfaces trained by a rewritten training code using automatic differentiation. The new machinery is the second intrinsic surface $M_2$ together with a Gaussian-process error decomposition used to separate correlated, physical light-curve variation (a Mat\'ern 3/2 kernel with a 5-day length scale) from uncorrelated photometric noise; the fitted amplitudes set per-band error floors and provide evidence that the coherent $x_2$ signal is real. Model definitions fix the means, variances, and correlations of $x_1$, $x_2$, and $c$ on the training sample, and a post-training rotation minimizes mutual information between $x_1$ and $x_2$ so the two axes are separated as cleanly as possible.

What would settle it

Rerun the SALT3+ training on the same supernovae with an independent photometric calibration, such as scene-modeled photometry that replaces difference imaging, and on simulated light curves generated with no extra variability: if a coherent $x_2$ surface of comparable amplitude is still recovered, or if the i-band secondary-maximum correlation disappears while the residuals remain, the claim that the variability is intrinsic fails.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that the residual scatter of Type Ia supernova light curves around the SALT3 model is not pure noise but a structured, coherent second axis of diversity. The paper constructs SALT3+, whose flux model is $F(p,\lambda)=x_0[M_0+x_1M_1+x_2M_2]\exp(-0.4c\,CL(\lambda))$, with $M_2$ a second spectral surface trained from combined survey photometry and spectroscopy. In this model, $x_2$ mostly shifts the slope of the $r-i$ color curve and the amplitude of the i-band secondary maximum relative to the primary, while $x_1$ controls stretch; the new axis correlates with spectral line velocities, especially calcium and silicon features. The paper shows that a standard SALT3 fit of the same supernovae absorbs much of this variation into the color parameter $c$, which is why colors from SALT3 are less informative: the color dispersion drops to the millimagnitude level in rest-frame V when $x_2$ is included. Neglecting $x_2$ in the standardization leaves a trend $\Delta\mu = 0.039\,x_2 + 0.003$ mag in Hubble residuals, but no significant mean shift of $x_2$ with redshift is found, so the authors judge current distance measurements to be unbiased while flagging the trend as a potential systematic.

Load-bearing premise

The detection of x2 rests on the assumption that time-correlated photometric errors are absent, so that the correlated 5-day light-curve residuals are physical rather than arising from calibration or image-subtraction systematics; the paper itself notes that leave-one-out outliers occur at about five times the predicted rate, meaning the noise model is incomplete.

Editorial extensions

If this is right

  • Multi-filter coverage matters: with only two photometric bands, $c$ and $x_2$ remain degenerate, so robust extinction measurements require at least three filters.
  • Ignoring $x_2$ in standardization creates a residual trend of $0.039\pm0.005$ mag in Hubble residuals; if the mean of $x_2$ evolves with redshift, this becomes a systematic, though the data show no significant redshift trend.
  • Including $x_2$ reduces the model color dispersion to millimagnitude level in rest-frame V and increases the fitted color-standardization coefficient by $0.22\pm0.03$, bringing it closer to typical dust-law values.
  • Current cosmological distance measurements appear unbiased: with no significant shift of mean $x_2$ with redshift, the implied systematic is only $2.1\pm1.5$ mmag under the present sample.
  • The extra component correlates with spectral line velocities and calcium feature strengths, linking the empirical axis to physical ejecta properties.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If future high-cadence surveys add near-infrared or ultraviolet bands, the degeneracy between $x_2$ and $c$ could be broken more fully than with the three optical bands used here, which would test whether the improved color accuracy translates into a tighter Hubble diagram.
  • Because the Gaussian-process decomposition assumes a 5-day correlation timescale, a natural stress test is to repeat the analysis with longer kernels; if the recovered $x_2$ amplitude depends strongly on that choice, part of the signal may be slow calibration drift rather than supernova physics.
  • The spectral-velocity correlation suggests $x_2$ may be tied to explosion asymmetry or viewing angle; a testable prediction is that spectropolarimetric observations of high-$|x_2|$ supernovae would show enhanced polarization compared with low-$|x_2|$ objects.
  • The paper's conclusion that higher-dimensional models are unnecessary for cosmology applies to the current generation of optical surveys; if demographic evolution of $x_2$ were found at higher redshift, the estimated $2.1\pm1.5$ mmag systematic could grow and the conclusion would need revisiting.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper trains SALT3+, an extension of the SALT3 light-curve model with an additional intrinsic variation parameter x2, using a combined training sample of the K21 compilation and ZTF SN Ia DR2 photometry and spectroscopy. The model is used to search for coherent phase-dependent color variability beyond SALT3, and the authors report that such variability is present, mainly in g-r and r-i color curves correlated with an i-band secondary-maximum boost. They also report spectral line-velocity correlations with x2, a reduced color dispersion relative to SALT3, and a 0.039 +/- 0.005 mag trend in SALT3-based Hubble residuals as a function of x2, while concluding that current cosmological measurements show no evidence of bias. The paper is explicitly framed as exploratory and not ready for cosmological use because of ongoing ZTF calibration work.

Significance. If the x2 component is real, this is a valuable contribution to SN Ia light-curve modeling and to the systematic-error budget for dark energy measurements: the trained model is publicly available in SNANA and sncosmo, the training code has been rewritten in JAX and is released, and the spectral velocity correlations provide an independent physical anchor for the photometric component. The paper is candid about the calibration limitations and about the exploratory nature of the cosmology results. However, the quantitative detection of x2 rests on a Gaussian-process decomposition that cannot by itself separate correlated calibration systematics from physical variability, and the headline Hubble-residual trend is measured in-sample on the same survey used for training; these issues are load-bearing for the central claims and need to be addressed with additional tests.

major comments (4)
  1. [2.2.1, Table 1] The detection of additional coherent variability rests on the Gaussian-process decomposition of Sec. 2.2.1, which assumes photometric noise is uncorrelated in time while physical variations are correlated on a 5-day timescale. The text itself states that the GP cannot distinguish time-correlated photometric errors from variation in the true light-curve, and Sec. 2.2 notes that ZTF calibration is work in progress with unbudgeted uncertainties greater than 0.01 mag. This is particularly relevant for i-band, where the fitted correlated amplitude is 8.4 centimag (compared with 2.2 centimag uncorrelated) and where the M2 surface produces its most distinctive feature, the secondary-maximum boost (Sec. 4.1.1). The leave-one-out test also finds 3-sigma outliers at roughly five times the predicted rate, indicating that the noise model is incomplete. Please provide a quantitative test that a substantial part of this correlated i-band power is not calibration drift; for example, repeat the GP analysis on residuals from an improved calibration solution, compare the inferred x2 values against observing-condition or image-subtraction diagnostics, or demonstrate that the x2 signal survives when the i-band is excluded from the training.
  2. [4.3, Fig. 15, Table 2] The headline Hubble-residual trend of 0.039 +/- 0.005 mag (Sec. 4.3, Fig. 15) is an in-sample fit summary rather than an out-of-sample prediction. The x2 component is trained on the same ZTF sample used to compute the residuals, and the cosmology sample is selected with post-hoc cuts on SALT3+ parameters, with the x2-c correlation cut alone removing 727 of 2214 objects. A trend in residuals is expected from overfitting even if x2 contained no astrophysical signal, so the quoted slope and the Delta-AIC = 66.4 preference in Table 2 do not by themselves establish that the effect is real. Please provide an out-of-sample assessment, for example by training on one half of the sample and measuring the residual trend on the other half, or by reporting the slope separately for the non-ZTF subsample with a full significance statement; the current roughly 2.5-sigma preference for SALT3+ in the non-ZTF subsample does not directly quantify the 0.039 mag slope.
  3. [3.2, 4.2] The priors used in training and fitting are not innocuous for the central trend. Section 3.2 imposes unit normal priors on x1 and x2 and N(0, 0.2) on c, and Sec. 4.2 states that the SNANA light-curve fits also include unit normal priors on x1 and x2. Since x2 is poorly constrained for most objects and is dominated by the N(0,1) prior (Sec. 4.2.2), the x2 values entering Fig. 15 and the 0.039 mag slope are partly shaped by the prior. Please show that the slope and its significance are stable under reasonable variations of the prior width, or repeat the analysis using only objects whose x2 is well constrained, for example those observed in at least three filters as discussed in Sec. 4.2.2.
  4. [4.3, Sec. 5] The conclusion that there is no bias in current cosmological measurements (Abstract and Sec. 5) is not quantitatively supported by the analysis as presented, because the Tripp fits in Sec. 4.3 ignore selection effects and regression dilution, as the authors acknowledge in Sec. 5. This does not weaken the detection of x2, but it means the no-evidence-of-bias statement should either be restricted to a null result under the simplified estimator, or be accompanied by a forward-modeling calculation that estimates the size of a bias that could be hidden by these effects.
minor comments (6)
  1. [3.3.1] The reference 'Johannson et al. (2024)' appears to be a misspelling of 'Johansson et al. (2024)' and is not listed in the reference list; please correct and add the reference.
  2. [4.3, Eq. (6)] The symbol M is used both for the absolute magnitude in Eq. (5) and for host galaxy mass in Eq. (6); this overloads the notation and makes the definition of theta(M) confusing, especially for objects with no mass available.
  3. [2.2.1] The phrase 'cannot correct distinguish' should read 'cannot distinguish'.
  4. [4.2] The sentence 'although with we include unit normal priors centered at 0 on x1 and x2' contains a word-order error and should read 'we include unit normal priors centered at 0 on x1 and x2'.
  5. [Fig. 15] The y-axis label is ambiguous; please state explicitly that the residuals are computed from SALT3.K21 fits with x1 and c standardization and are plotted against x2 from SALT3+ fits.
  6. [4.3.2] The statement that there is no effective host-galaxy correlation with x2 is made without a quantitative measure; please report a correlation coefficient, a p-value, or an upper limit.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the SALT3+ training, GP variability decomposition, and Hubble-residual trend are separate fits with explicit caveats, and external checks give independent content.

full rationale

The paper does not exhibit a circular derivation. The central claim is that an additional SALT component M2 captures coherent, phase-dependent color variation; this is obtained by training SALT3+ with SALTshaker, and the existence of coherent residual power is separately estimated in Sec. 2.2.1 with a Gaussian process that fits nonzero correlated amplitudes. That GP step does assume a 5-day correlation timescale and labels the correlated component as potentially physical, but the paper explicitly disclaims the ambiguity: it cannot distinguish time-correlated photometric errors from true light-curve variation and notes that ZTF calibration is ongoing work. This is a stated limitation on interpretation, not a circular reduction. The Hubble-residual trend of 0.039 +/- 0.005 mag in Sec. 4.3 is an in-sample diagnostic: x2 comes from SALT3+ fits and the residuals from SALT3.K21 fits to the same sample. However, the paper explicitly states that the Tripp estimator plays no role in SALTshaker and is not evaluated during training, and it frames the trend as a potential systematic rather than an out-of-sample prediction. External anchors give the M2 interpretation independent content: the i-band secondary-maximum boost was previously identified by Pessi et al. (2022), x2 correlates with spectral line velocities in Sec. 4.1.2, and the non-ZTF subsample also prefers the extended model at about 2.5 sigma. Self-citations to K21 and Rigault et al. 2024 provide the training code and data release, but the detection and cosmology comparison are reproduced in this paper rather than assumed from those references. No equation defines x2 in terms of the Hubble-residual trend, and no fitted parameter is renamed as a prediction; therefore no circular step meets the required evidentiary bar.

Assumptions & free parameters 4 free parameters · 5 assumptions · 1 invented entities

The central claim rests on several fitted quantities (GP error floors, color scatter, Tripp parameters) and on assumptions about noise structure, training-sample representativeness, and the linear decomposition of SN Ia diversity. The x2 dimension is an empirically derived latent variable with some independent spectral validation.

free parameters (4)
  • GP uncorrelated error floors (sigma_g, sigma_r, sigma_i) = 0.013, 0.018, 0.022 mag (Table 1)
    Fitted via Gaussian process to ZTF light-curve residuals from SALT2; added in quadrature to photometric uncertainties in training.
  • GP correlated amplitudes (sigmaCorr_g, sigmaCorr_r, sigmaCorr_i) = 0.046, 0.049, 0.084 mag (Table 1)
    Fitted amplitudes of a 5-day Matérn kernel attributed to physical variability; used to justify the existence of extra variation.
  • Color scatter polynomial coefficients (a0..a4 of k(lambda)) = not reported in paper
    Coefficients for k(lambda) in Eq. 3, fit during training; affect color dispersion and error floors.
  • Tripp nuisance parameters (alpha1, alpha2, beta, gamma, sigma_int, M) = see Table 2; alpha1=0.111, alpha2=-0.019, beta=2.94, gamma=0.07
    Fit to the same sample to evaluate cosmological implications; not part of model training but support claims about bias.
assumptions (5)
  • domain assumption SN Ia diversity is linearly decomposable into a small number of spectral surfaces (SALT ansatz)
    Used in Eq. 1; the model assumes a linear combination M0 + x1 M1 + x2 M2 describes all SNe.
  • ad hoc to paper Photometric noise is uncorrelated in time while physical variations are correlated on a ~5 day timescale
    Sec 2.2.1; this decomposition determines error floors and the detection of coherent variability.
  • ad hoc to paper ZTF calibration uncertainties greater than 0.01 mag are acceptable for training
    Sec 2.2; the authors note unbudgeted calibration uncertainties, which could mimic or distort the new component.
  • ad hoc to paper Unit normal priors on x1 and x2 and N(0,0.2) on c do not bias the inferred x2 distribution
    Sec 3.2; priors are included to reduce overfitting and shape parameter distributions.
  • domain assumption The training sample is representative of the cosmological SN Ia population
    Sec 2; sample matching is used to reduce known and unknown systematics, but selection functions differ between ZTF and K21.
invented entities (1)
  • x2 intrinsic variation parameter and M2 spectral surface independent evidence
    purpose: Model additional SN Ia light-curve variability beyond x1 (color-curve slopes and i-band secondary maximum)
    The paper shows x2 correlates with Si II line velocity and with the i-band secondary maximum feature noted by Pessi et al. 2022, providing falsifiable external checks. However, the parameter is defined partly via training-sample constraints, so its physical status remains empirical.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ZTF SN Ia DR2: Improved SN Ia colors through expanded dimensionality with SALT3+." pith.science (2026). https://pith.science/paper/OH7REBX4

@misc{pith2026250209713,
  author       = {Pith},
  title        = {Pith review of: ZTF SN Ia DR2: Improved SN Ia colors through expanded dimensionality with SALT3+},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OH7REBX4}},
  note         = {Machine review of arXiv:2502.09713}
}
read the original abstract

Type Ia supernovae (SNe Ia) are a key probe in modern cosmology, as they can be used to measure luminosity distances at gigaparsec scales. Models of their light-curves are used to project heterogeneous observed data onto a common basis for analysis. The SALT model currently used for SN Ia cosmology describes SNe as having two sources of variability, accounted for by a color parameter c, and a "stretch parameter" x1. We extend the model to include an additional parameter we label x2, to investigate the cosmological impact of currently unaddressed light-curve variability. We construct a new SALT model, which we dub "SALT3+". This model was trained by an improved version of the SALTshaker code, using training data combining a selection of the second data release of cosmological SNe Ia from the Zwicky Transient Facility and the existing SALT3 training compilation. We find additional, coherent variability in supernova light-curves beyond SALT3. Most of this variation can be described as phase-dependent variation in g-r and r-i color curves, correlated with a boost in the height of the secondary maximum in i-band. These behaviors correlate with spectral differences, particularly in line velocity. We find that fits with the existing SALT3 model tend to address this excess variation with the color parameter, leading to less informative measurements of supernova color. We find that neglecting the new parameter in light-curve fits leads to a trend in Hubble residuals with x2 of 0.039 +/- 0.005 mag, representing a potential systematic uncertainty. However, we find no evidence of a bias in current cosmological measurements. We conclude that extended SN Ia light-curve models promise mild improvement in the accuracy of color measurements, and corresponding cosmological precision. However, models with more parameters are unlikely to substantially affect current cosmological results.

Figures

Figures reproduced from arXiv: 2502.09713 by the authors.

Figure 1
Figure 1. Density of spectroscopic and photometric data from train [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Light-curve of 2018cvq, shown at maximum light, along [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Illustration of the Gaussian process inference described in Section 2.2.1 as applied to the light-curve of 2019gvw. First [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (14 more)
Figure 4
Figure 4. Figure 4: Early time photometric fluxes, relative to photometric un [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Light-curves for the SALT3+ model. Upper figure shows the variation in three bands as a function of x1 with x2 fixed to 0. Lower figure shows the variation as a function of x2 with x1 fixed to 0. Any other values will be represented by a linear combination of those lig…
Figure 6
Figure 6. Figure 6: Color curves for the SALT3+ model in magnitudes. Upper figure shows the variation in three bands as a function of x1 with x2 fixed to 0. Lower figure shows the variation as a function of x2 with x1 fixed to 0. Any other values will be represented by a linear combinatio…
Figure 7
Figure 7. Figure 7: Synthetic spectra of SALT3+ across phase, x1, and x2. Left panel shows variation with x1, right panel shows variation with x2 from −5 days to +35 days. SN Ia population. Variance in dust population parameters such as RV from galaxy to galaxy may be a significant effect…
Figure 8
Figure 8. Figure 8: Spectral effect of x2 at maximum light. Top left panel show velocity measured from spectral features as a function of x2, while top right panel shows pseudo-equivalent widths (PEW) as a function of x2. Bottom panels show slopes as a function of wavelength. measured pho…
Figure 10
Figure 10. Figure 10: Distribution in two dimensions of the x1,x2 parameter space for the ZTF sample using only SNe Ia within the volume￾limited sample (z < 0.06) Gaussian log-likelihood L = − P log(σµ)/2 + (∆µ/σµ) 2 /2, us￾ing light-curve fits made with both the new model and the most rec…
Figure 11
Figure 11. Figure 11: Distribution of parameter uncertainties as calculated under each SALT3 [PITH_FULL_IMAGE:figures/full_fig_p013_11.png]
Figure 12
Figure 12. Figure 12: Boxplot showing the distribution of reduced correlation [PITH_FULL_IMAGE:figures/full_fig_p014_12.png]
Figure 16
Figure 16. Figure 16: Observed and inferred intrinsic color distributions de [PITH_FULL_IMAGE:figures/full_fig_p015_16.png]
Figure 14
Figure 14. Figure 14: Difference in fitted light-curve parameters between SALT3.K21 and SALT3+ as a function of fitted x2. 4 2 0 2 4 x2 1.00 0.75 0.50 0.25 0.00 0.25 0.50 0.75 1.00 (z) ({ x1, c } sta n d ardiz atio n) = 0.039x2 + 0.003 [PITH_FULL_IMAGE:figures/full_fig_p015_14.png]
Figure 15
Figure 15. Figure 15: Hubble residuals calculated using SALT3.K21, with [PITH_FULL_IMAGE:figures/full_fig_p015_15.png]
Figure 17
Figure 17. Figure 17: Hubble residuals, corrected for x1 and x2 but not color, as compared between models. Upper panel shows residuals calculated with SALT3+, in lower panel with SALT3. Solid line shows a quadratic fit to the data, dashed line shows a linear fit. 2 0 2 4 x2 0.0 0.1 0.2 0.3…
Figure 18
Figure 18. Figure 18: x2 distribution, split between high and low mass galaxies at 1010M⊙. gression dilution. The results on standardization presented here should thus be regarded as preliminary. Correctly accounting for these biases requires either frequentist forward-simulation, or a Bay…
Figure 14
Figure 14. Figure 14: If accounting for x2 at the level of cosmological anal￾ysis is unnecessary, developing a more robust error model for a one-dimensional SALT model may be a promising direction for future work. Acknowledgements. The authors thank Daniel Kasen for useful discussion. Base…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. A Reassessment of the Pantheon+ and DES 5YR Calibration Uncertainties: Dovekie

    astro-ph.CO 2025-06 conditional novelty 6.0 of 10

    A new open-source cross-calibration of 11 supernova surveys cuts the photometric systematic on the dark-energy parameter w by about 1.5x for Pantheon+ and shows small calibration shifts can be amplified in distances.

Reference graph

Works this paper leans on

86 extracted references · 51 canonical work pages · cited by 1 Pith paper

  1. [1]

    Abbott, T. M. C., Allam, S., Andersen, P., et al. 2019, ApJ, 872, L30

  2. [2]

    2002, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, V ol

    Aldering, G., Adam, G., Antilogus, P., et al. 2002, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, V ol. 4836, Survey and Other Telescope Technologies and Discoveries, ed. J. A. Tyson & S. Wol ff, 61–72

  3. [3]

    2015, MNRAS, 453, 3300

    Amanullah, R., Johansson, J., Goobar, A., et al. 2015, MNRAS, 453, 3300

  4. [4]

    2006, A&A, 447, 31

    Astier, P., Guy, J., Regnault, N., et al. 2006, A&A, 447, 31

  5. [5]

    2018, A&A, 614, A134

    Balland, C., Cellier-Holzem, F., Lidman, C., et al. 2018, A&A, 614, A134

  6. [6]

    2016, SNCosmo: Python library for supernova cosmology

    Barbary, K., Barclay, T., Biswas, R., et al. 2016, SNCosmo: Python library for supernova cosmology

  7. [7]

    2015, Sncosmo: V1.0.0

    Barbary, K., rbiswas4, Rodney, S., et al. 2015, Sncosmo: V1.0.0

  8. [8]

    C., Kulkarni, S

    Bellm, E. C., Kulkarni, S. R., Graham, M. J., et al. 2019, PASP, 131, 018002

Show all 86 references
  1. [9]

    2014, A&A, 568, A22

    Betoule, M., Kessler, R., Guy, J., et al. 2014, A&A, 568, A22

  2. [10]

    D., Walters, R., et al

    Blagorodnova, N., Neill, J. D., Walters, R., et al. 2018, PASP, 130, 035003

  3. [11]

    2018, JAX: composable transforma- tions of Python+NumPy programs, http://github.com/google/jax

    Bradbury, J., Frostig, R., Hawkins, P., et al. 2018, JAX: composable transforma- tions of Python+NumPy programs, http://github.com/google/jax

  4. [12]

    & Scolnic, D

    Brout, D. & Scolnic, D. 2021, ApJ, 909, 26

  5. [13]

    R., Parent, E., Phillips, M

    Burns, C. R., Parent, E., Phillips, M. M., et al. 2018, ApJ, 869, 56

  6. [14]

    R., Stritzinger, M., Phillips, M

    Burns, C. R., Stritzinger, M., Phillips, M. M., et al. 2014, ApJ, 789, 32

  7. [15]

    2020, ApJ, 901, 154

    Burrow, A., Baron, E., Ashall, C., et al. 2020, ApJ, 901, 154

  8. [16]

    2011, A&A, 529, L4

    Chotard, N., Gangler, E., Aldering, G., et al. 2011, A&A, 529, L4

  9. [17]

    O., Kenworthy, W

    Dai, M., Jones, D. O., Kenworthy, W. D., et al. 2023, ApJS, 267, 1

  10. [18]

    M., Riddle, R., et al

    Dekany, R., Smith, R. M., Riddle, R., et al. 2020, PASP, 132, 038001

  11. [19]

    G., Jha, S

    Dettman, K. G., Jha, S. W., Dai, M., et al. 2021, ApJ, 923, 267

  12. [20]

    M., Mortlock, D

    Feeney, S. M., Mortlock, D. J., & Dalmasso, N. 2018, MNRAS, 476, 3861 Fernández, A., García, S., Galar, M., et al. 2018, Dimensionality Reduction for Imbalanced Learning (Cham: Springer International Publishing), 227–251

  13. [21]

    J., Scolnic, D., Rest, A., et al

    Foley, R. J., Scolnic, D., Rest, A., et al. 2018, MNRAS, 475, 193

  14. [22]

    2018, RNAAS, 2, 31

    Foreman-Mackey, D. 2018, RNAAS, 2, 31

  15. [23]

    2017, AJ, 154, 220

    Foreman-Mackey, D., Agol, E., Ambikasaran, S., & Angus, R. 2017, AJ, 154, 220

  16. [24]

    A., Sharma, Y ., et al

    Fremling, C., Miller, A. A., Sharma, Y ., et al. 2020, ApJ, 895, 32 GPy. 2012, GPy: A Gaussian process framework in python, http://github. com/SheffieldML/GPy

  17. [25]

    J., Kulkarni, S

    Graham, M. J., Kulkarni, S. R., Bellm, E. C., et al. 2019, PASP, 131, 078001

  18. [26]

    S., et al

    Grayling, M., Thorp, S., Mandel, K. S., et al. 2024, MNRAS, 531, 953

  19. [27]

    2007, A&A, 466, 11

    Guy, J., Astier, P., Baumont, S., et al. 2007, A&A, 466, 11

  20. [28]

    2005, Astronomy & As- trophysics, 443, 781

    Guy, J., Astier, P., Nobili, S., Regnault, N., & Pain, R. 2005, Astronomy & As- trophysics, 443, 781

  21. [29]

    2010, A&A, 523, A7

    Guy, J., Sullivan, M., Conley, A., et al. 2010, A&A, 523, A7

  22. [30]

    M., Suntzeff, N

    Hamuy, M., Phillips, M. M., Suntzeff, N. B., et al. 1996, AJ, 112, 2408

  23. [31]

    2019, ApJ, 871, 219

    Hayden, B., Rubin, D., & Strovink, M. 2019, ApJ, 871, 219

  24. [32]

    2009, ApJ, 700, 331

    Hicken, M., Challis, P., Jha, S., et al. 2009, ApJ, 700, 331

  25. [33]

    P., et al

    Hicken, M., Challis, P., Kirshner, R. P., et al. 2012, ApJS, 200, 12

  26. [34]

    R., Davis, T

    Hinton, S. R., Davis, T. M., Kim, A. G., et al. 2019, ApJ, 876, 15

  27. [35]

    A., Marriner, J., Kessler, R., et al

    Holtzman, J. A., Marriner, J., Kessler, R., et al. 2008, AJ, 136, 2306

  28. [36]

    J., et al

    Hounsell, R., Scolnic, D., Foley, R. J., et al. 2018, ApJ, 867, 23

  29. [37]

    & Roos, M

    James, F. & Roos, M. 1975, Computer Physics Communications, 10, 343

  30. [38]

    P., Challis, P., et al

    Jha, S., Kirshner, R. P., Challis, P., et al. 2006, AJ, 131, 527

  31. [39]

    G., & Kirshner, R

    Jha, S., Riess, A. G., & Kirshner, R. P. 2007, The Astrophysical Journal, 659, 122

  32. [40]

    B., Fox, O

    Johansson, J., Cenko, S. B., Fox, O. D., et al. 2021, ApJ, 923, 237

  33. [41]

    O., Kenworthy, W

    Jones, D. O., Kenworthy, W. D., Dai, M., et al. 2023, ApJ, 951, 22

  34. [42]

    O., Scolnic, D

    Jones, D. O., Scolnic, D. M., Foley, R. J., et al. 2019, ApJ, 881, 19 Jönsson, J., Sullivan, M., Hook, I., et al. 2010, MNRAS, 405, 535

  35. [43]

    & Woosley, S

    Kasen, D. & Woosley, S. E. 2007, ApJ, 656, 661

  36. [44]

    D., Jones, D

    Kenworthy, W. D., Jones, D. O., Dai, M., et al. 2021, ApJ, 923, 265

  37. [45]

    & Scolnic, D

    Kessler, R. & Scolnic, D. 2017, ApJ, 836, 56

  38. [46]

    Khokhlov, A. M. 1995, ApJ, 449, 695

  39. [47]

    L., Rigault, M., Neill, J

    Kim, Y . L., Rigault, M., Neill, J. D., et al. 2022, PASP, 134, 024505

  40. [48]

    2013, AJ, 145, 11

    Krisciunas, K., Bastola, D., Espinoza, J., et al. 2013, AJ, 145, 11

  41. [49]

    R., et al

    Krisciunas, K., Contreras, C., Burns, C. R., et al. 2017, AJ, 154, 211

  42. [50]

    A., & Hlozek, R

    Kunz, M., Bassett, B. A., & Hlozek, R. A. 2007, Phys. Rev. D, 75, 103508 Léget, P. F., Gangler, E., Mondon, F., et al. 2020, A&A, 636, A46 LSST Dark Energy Science Collaboration, Mandelbaum, R., Eifler, T., et al. 2018, arXiv e-prints, arXiv:1809.01669

  43. [51]

    S., Foley, R

    Mandel, K. S., Foley, R. J., & Kirshner, R. P. 2014, ApJ, 797, 75

  44. [52]

    S., Narayan, G., & Kirshner, R

    Mandel, K. S., Narayan, G., & Kirshner, R. P. 2011, The Astrophysical Journal, 731, 120

  45. [53]

    S., Scolnic, D

    Mandel, K. S., Scolnic, D. M., Shari ff, H., Foley, R. J., & Kirshner, R. P. 2017, ApJ, 842, 93

  46. [54]

    C., Trotta, R., Berkes, P., Starkman, G

    March, M. C., Trotta, R., Berkes, P., Starkman, G. D., & Vaudrevange, P. M. 2011, MNRAS, 418, 2308

  47. [55]

    J., Laher, R

    Masci, F. J., Laher, R. R., Rusholme, B., et al. 2019, PASP, 131, 018003 Article number, page 17 of 18 A&A proofs: manuscript no. aanda

  48. [56]

    A., Yao, Y ., Bulla, M., et al

    Miller, A. A., Yao, Y ., Bulla, M., et al. 2020, ApJ, 902, 47

  49. [57]

    2021, A&A, 649, A74

    Nicolas, N., Rigault, M., Copin, Y ., et al. 2021, A&A, 649, A74

  50. [58]

    2020, ApJ, 895, L5

    Pan, Y .-C. 2020, ApJ, 895, L5

  51. [59]

    C., Sullivan, M., Maguire, K., et al

    Pan, Y . C., Sullivan, M., Maguire, K., et al. 2015, MNRAS, 446, 354

  52. [60]

    2019, PhD thesis, Stockholm University

    Papadogiannakis, S. 2019, PhD thesis, Stockholm University

  53. [61]

    A., Fremling, C., Sollerman, J., et al

    Perley, D. A., Fremling, C., Sollerman, J., et al. 2020, ApJ, 904, 35

  54. [62]

    J., Hsiao, E

    Pessi, P. J., Hsiao, E. Y ., Folatelli, G., et al. 2022, MNRAS, 510, 4929

  55. [63]

    Phillips, M. M. 1993, ApJ, 413, L105 Planck Collaboration, Aghanim, N., Akrami, Y ., et al. 2020, A&A, 641, A6

  56. [64]

    2023, ApJ, 945, 84

    Popovic, B., Brout, D., Kessler, R., & Scolnic, D. 2023, ApJ, 945, 84

  57. [65]

    S., Hudson, M

    Rahman, W., Trotta, R., Boruah, S. S., Hudson, M. J., & van Dyk, D. A. 2022, MNRAS, 514, 139

  58. [66]

    J., et al

    Rest, A., Scolnic, D., Foley, R. J., et al. 2014, ApJ, 795, 44

  59. [67]

    G., Kirshner, R

    Riess, A. G., Kirshner, R. P., Schmidt, B. P., et al. 1999, AJ, 117, 707

  60. [68]

    D., Blagorodnova, N., et al

    Rigault, M., Neill, J. D., Blagorodnova, N., et al. 2019, A&A, 627, A115

  61. [69]

    M., Dixon, S., Rubin, D., et al

    Rose, B. M., Dixon, S., Rubin, D., et al. 2020, ApJ, 890, 60

  62. [70]

    2015, ApJ, 813, 137

    Rubin, D., Aldering, G., Barbary, K., et al. 2015, ApJ, 813, 137

  63. [71]

    2023, arXiv e-prints, arXiv:2311.12098

    Rubin, D., Aldering, G., Betoule, M., et al. 2023, arXiv e-prints, arXiv:2311.12098

  64. [72]

    C., et al

    Sako, M., Bassett, B., Becker, A. C., et al. 2018, PASP, 130, 064002

  65. [73]

    Salim, S., Boquien, M., & Lee, J. C. 2018, ApJ, 859, 11

  66. [74]

    2018, ApJ, 869, 167

    Saunders, C., Aldering, G., Antilogus, P., et al. 2018, ApJ, 869, 167

  67. [75]

    M., Jones, D

    Scolnic, D. M., Jones, D. O., Rest, A., et al. 2018, ApJ, 859, 101

  68. [76]

    Shariff, H., Jiao, X., Trotta, R., & van Dyk, D. A. 2016, ApJ, 827, 1

  69. [77]

    R., Foley, R

    Siebert, M. R., Foley, R. J., Jones, D. O., et al. 2019, MNRAS, 486, 5785

  70. [78]

    O., Popovic, B., et al

    Taylor, G., Jones, D. O., Popovic, B., et al. 2023, MNRAS, 520, 5209

  71. [79]

    Taylor, G., Lidman, C., Popovic, B., & Abbot, H. J. 2024, MNRAS, 528, 4643

  72. [80]

    E., et al

    Taylor, G., Lidman, C., Tucker, B. E., et al. 2021, MNRAS, 504, 4111

  73. [81]

    S., Jones, D

    Thorp, S., Mandel, K. S., Jones, D. O., Ward, S. M., & Narayan, G. 2021, MN- RAS, 508, 4310 Tripp & Robert. 1998, Astronomy and Astrophysics, 331, 815

  74. [82]

    2024, arXiv e-prints, arXiv:2401.02945

    Vincenzi, M., Brout, D., Armstrong, P., et al. 2024, arXiv e-prints, arXiv:2401.02945

  75. [83]

    S., Hook, I

    Walker, E. S., Hook, I. M., Sullivan, M., et al. 2011, MNRAS, 410, 1262

  76. [84]

    V ., Ganeshalingam, M., et al

    Wang, X., Filippenko, A. V ., Ganeshalingam, M., et al. 2009, ApJ, 699, L139

  77. [85]

    V ., Zhang, T., & Zhao, X

    Wang, X., Wang, L., Filippenko, A. V ., Zhang, T., & Zhao, X. 2013, Science, 340, 170

  78. [86]

    Wojtak, R., Hjorth, J., & Hjortlund, J. O. 2023, MNRAS, 525, 5187 Article number, page 18 of 18

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.