Pantheon+ Type Ia Supernova Compilation

• Pantheon+ is a homogenized Type Ia supernova compilation combining 1701 light-curve fits from 1550 spectroscopically confirmed SNe Ia gathered from 18 surveys.
• It employs uniform SALT2 standardization and a global cross-survey calibration strategy to deliver high-precision distance measurements for cosmological analyses.
• The dataset features rigorous redshift revisions, bias corrections, and a comprehensive covariance structure to support robust Hubble constant and dark-energy inferences.

Searching arXiv for the cited Pantheon and Pantheon+ papers to ground the article and collect citation metadata. arxiv_search.query({"search_query":"id:(Scolnic et al., 2021) OR id:(Carr et al., 2021) OR id:(Scolnic et al., 2017) OR id:(Wang et al., 18 May 2026) OR id:(Wang et al., 13 Apr 2026) OR id:(Rana, 23 Sep 2025) OR id:(Ringermacher et al., 2019)","max_results":10,"sort_by":"relevance","sort_order":"descending"}) Retrieved relevant arXiv records for Pantheon, Pantheon+, and follow-up analyses. Pantheon+ is a homogenized Type Ia supernova compilation comprising 1701 light-curve fits of 1550 spectroscopically confirmed SNe Ia, assembled across 18 different surveys for joint analyses of the Hubble constant and the dark-energy equation-of-state parameter. It extends the original Pantheon sample of 1048 SNe Ia, includes 153 SNe with $z<0.01$, and spans the nearby calibrator regime, the Hubble flow, and the high-redshift domain to $z\approx2.26$–$2.30$; in practical cosmological fits, analyses typically restrict to $z>0.01$ (Scolnic et al., 2021, Scolnic et al., 2017, Rana, 23 Sep 2025).

1. Provenance and sample definition

The original Pantheon sample combined 279 quality-cut Pan-STARRS1 Medium Deep SNe Ia with SDSS, SNLS, low-$z$, and HST samples to form a compilation of 1048 SNe Ia over $0.01$z$ surveys, SDSS, PS1 Medium-Deep, SNLS, and HST (Scolnic et al., 2017). Pantheon+ preserves that role as a combined Hubble-diagram resource while expanding both the redshift coverage at the low-$z$ end and the survey diversity.

In the light-curve release, Pantheon+ is organized across 18 survey datasets: LOSS1, LOSS2, SOUSA, CSP DR3, CfA1, CfA2, CfA3S+K, CfA4, LOWZ, CNIa 0.02, Foundation, SDSS, PS1MD, SNLS, DES 3YR, HDFN, SCP, and CANDELS+CLASH, with GOODS+PANS also listed among the high-$z$ components. The most dramatic growth relative to Pantheon occurs at low redshift, where six new samples raise the number of SNe with $z<0.1$ by $z\approx2.26$0; at higher redshift, the inclusion of the DES three-year sample adds $z\approx2.26$1 additional SNe up to $z\approx2.26$2 (Scolnic et al., 2021).

This construction makes Pantheon+ simultaneously a cosmology sample and a calibration infrastructure. By including nearby events with $z\approx2.26$3, it enables a unified treatment of the local “second rung” calibrator SNe and the $z\approx2.26$4 Hubble-flow SNe used for cosmological inference and $z\approx2.26$5 determination (Scolnic et al., 2021). A plausible implication is that Pantheon+ should be understood not merely as a larger Pantheon, but as a dataset designed to merge distance-ladder and dark-energy use cases within a single covariance-aware framework.

2. Photometric calibration and SALT2 standardization

Pantheon+ adopts a global cross-survey calibration strategy often described as “Supercal.” All light curves are re-fitted with the same SALT2 model, zero-point offsets and filter throughput functions are adjusted to a common AB system, empirical color terms and filter transformations are applied to map each instrument’s natural magnitudes onto the fiducial system, and a combined calibration covariance matrix is constructed to encode correlated zero-point uncertainties across filters and surveys. PS1 serves as the calibration backbone in this scheme, with SDSS, SNLS, and low-$z\approx2.26$6 systems tied to PS1 through overlapping fields or common spectrophotometric standards; HST photometry is brought onto the same scale when possible (Rana, 23 Sep 2025).

In the light-curve release, the photometry remains in the native systems released by each survey, while the global cross-calibration ties every system to CALSPEC standards and to one another. The release also records specific calibration revisions, including SDSS zero-point corrections in $z\approx2.26$7 and a revised CfA3 natural-to-standard offset in $z\approx2.26$8 relative to Pantheon (Scolnic et al., 2021).

The SALT2 fit yields the light-curve parameters $z\approx2.26$9, $2.30$0, $2.30$1, and $2.30$2, with

$2.30$3

Pantheon+ then uses the Tripp relation

$2.30$4

where $2.30$5 is the rest-frame $2.30$6-band apparent magnitude at peak, $2.30$7 is stretch, $2.30$8 is color, $2.30$9 is the fiducial absolute magnitude, $z>0.01$0 is the host-mass correction, and $z>0.01$1 is the simulation-based bias correction. In the release summary, nominal nuisance coefficients are $z>0.01$2 and $z>0.01$3 (Scolnic et al., 2021). A closely related Pantheon+ summary writes the standardized modulus as

$z>0.01$4

with $z>0.01$5 defined through a host-mass step at $z>0.01$6 and $z>0.01$7 mag (Rana, 23 Sep 2025).

The standardization step is therefore explicitly hierarchical. Pantheon+ does not treat raw SNe Ia as identical candles; it standardizes them through a calibrated global fit in which light-curve shape, color, host-galaxy environment, and selection biases are all propagated into the distance estimator.

3. Sample cuts, bias corrections, and covariance structure

Pantheon+ applies explicit quality cuts to define the cosmology sample. These include $z>0.01$8, $z>0.01$9, $z$0, survey-dependent thresholds on $z$1 and $z$2, light-curve $z$3 in at least three filters with pre- and post-peak coverage, measured host-galaxy mass, Milky Way reddening $z$4 mag, and exclusion of obvious spectroscopic peculiarities such as 91bg-like SNe. After these cuts, the sample remains $z$5 complete up to $z$6 and retains HST events out to $z$7 (Rana, 23 Sep 2025).

Selection effects are handled through survey-specific SNANA simulations. For each survey, simulated SNe with the same cadence, noise, trigger logic, and spectroscopic selection are SALT2-fitted and compared to their true distances, producing a bias-correction surface $z$8. In the observational sample, each SN’s $z$9 is interpolated from that surface. Typical $0.01Scolnic et al., 2021).

The uncertainty model is fully covariance-based. Pantheon+ provides a $0.01

$0.01

where $0.01Rana, 23 Sep 2025). In the light-curve release summary, 40 systematic perturbations contribute covariance through

$0.01

with additional off-diagonal terms introduced for duplicate fits of the same SN, but not for sibling pairs (Scolnic et al., 2021).

At low redshift, peculiar-velocity errors are added diagonally as

$0.01

with $0.01Rana, 23 Sep 2025). This makes the Pantheon+ covariance not merely a bookkeeping artifact but the central statistical object for downstream inference.

4. Redshift revision and peculiar-velocity modeling

A defining component of the Pantheon+ program is the dedicated redshift review and flow correction analysis. That work updated 990 heliocentric redshifts, fixed 44 cases of incorrect or missing coordinates, added 230 missing heliocentric or CMB-frame redshifts, and assigned 1200 missing redshift uncertainties (Carr et al., 2021).

The redshift framework uses exact multiplicative transformations: $z$0 and

$z$1

with the solar-motion correction defined through the line-of-sight projection of the CMB dipole velocity and the peculiar-velocity-corrected redshift $z$2 used for Hubble-diagram work (Carr et al., 2021).

The peculiar-velocity model is based on a 2M++ velocity-field reconstruction with

$z$3

where $z$4 and $z$5 km s$z$6 in the direction $z$7. Beyond $z$8 Mpc $z$9, the residual bulk flow is taken to decay following the $z$0CDM expectation, anchored to the $z$1 Mpc $z$2 sphere’s bulk flow of $z$3 km s$z$4 (Carr et al., 2021).

The resulting Pantheon+ redshift catalogue supplies, for each SN, $z$5, $z$6, $z$7, their uncertainties, and a peculiar-velocity estimate. The reported cosmological impact of the revisions is small: redshift-related systematics contribute negligibly compared to current statistical and calibration errors, with shifts remaining $z$8 of current SN-only uncertainties in $z$9 and $z$0 in $z$1 (Carr et al., 2021).

5. Internal consistency tests and public release

Pantheon+ includes explicit cross-checks on repeated and environmentally related observations. The compilation contains 151 duplicate SNe, meaning the same SN Ia was observed by two different surveys, almost all at $z$2. The root-mean-square difference in duplicate distance moduli is $z$3 mag, whereas two random SNe would differ by $z$4 mag. This motivates assigning full distance covariance between duplicate measurements of the same event (Scolnic et al., 2021).

The release also identifies 12 host galaxies containing SN sibling pairs or triplets. Because such siblings share progenitor environment, one expects correlated $z$5 and $z$6, but empirically the dispersion in sibling Hubble residuals is $z$7 mag, consistent with the overall SN intrinsic scatter of $z$8 mag. The adopted conclusion is that no extra distance-covariance term is required for siblings (Scolnic et al., 2021).

The public release is correspondingly detailed. It includes per-survey subdirectories, native-system light-curve files in ASCII or FITS format, metadata such as SN and host coordinates, Milky Way $z$9, heliocentric and CMB-frame redshifts, and peculiar velocities, as well as a global FITRES file containing SALT2 fit parameters, uncertainties, covariances, cut flags, survey identifiers, observational epochs, and host metrics. Filter transmission curves, CALSPEC standards, and zero-point offsets are also provided so that users can reproduce the cross-calibration or retrain alternative light-curve models (Scolnic et al., 2021).

These release choices make Pantheon+ more than a Hubble-diagram table. It is an end-to-end data product intended for reproducible cosmological analyses, cross-survey consistency tests, and alternative standardization pipelines.

6. Cosmological inference and information compression

Pantheon+ is routinely used through analytic marginalization over the nuisance offset

$z<0.1$0

which removes direct sensitivity to the $z<0.1$1–$z<0.1$2 degeneracy in distance-only analyses (Wang et al., 13 Apr 2026). One model-independent route converts $z<0.1$3 to comoving distances, forms finite differences $z<0.1$4, propagates the full covariance, and applies the Best Linear Unbiased Estimator to recover binned $z<0.1$5 estimates that are insensitive to $z<0.1$6 (Wang et al., 13 Apr 2026).

A second route compresses the entire Pantheon+ distance-redshift relation into eleven Gaussian-distributed parameters $z<0.1$7 at the knot set

$z<0.1$8

with $z<0.1$9 and a natural-boundary cubic spline used between knots. Because the forward model is linear in $z\approx2.26$00, analytic marginalization over $z\approx2.26$01 and application of BLUE give

$z\approx2.26$02

For Pantheon+ without flux averaging, the resulting compressed posterior reproduces the full distance-modulus likelihood within the statistical sampling noise of the chains (Wang et al., 18 May 2026).

The practical effect is large. The closed-form compression takes $z\approx2.26$03 s per dataset, or $z\approx2.26$04 s including flux averaging, and the downstream MCMC cost becomes the evaluation of an 11-dimensional Gaussian likelihood with per-step cost $z\approx2.26$05, independent of the SN sample size. In flat $z\approx2.26$06CDM, the Pantheon+ compression yields $z\approx2.26$07, compared with the full-$z\approx2.26$08 result $z\approx2.26$09. In flat $z\approx2.26$10CDM combined with Planck 2015 distance priors and DESI DR2 BAO, the compressed Pantheon+ likelihood gives $z\approx2.26$11, $z\approx2.26$12, and $z\approx2.26$13, identical to the corresponding full-$z\approx2.26$14 chain to $z\approx2.26$15 (Wang et al., 18 May 2026).

7. Interpretation, residual anomalies, and current debates

Pantheon+ is often treated as the benchmark SN Ia dataset for late-time cosmology, but later analyses emphasize that its interpretation remains contingent on tests of residual luminosity evolution and inter-probe consistency. A Gaussian-process analysis comparing Pantheon+ to a model-independent distance-modulus baseline $z\approx2.26$16 from cosmic chronometer $z\approx2.26$17 data finds that SNe Ia are consistent with being standard candles within $z\approx2.26$18, while also identifying localized departures, including a mild positive bump around $z\approx2.26$19 in Pantheon+ (Rana, 23 Sep 2025).

Model-independent dark-energy reconstructions using Pantheon+, DESI DR2 BAO, and Planck CMB distance priors likewise show mild structure. In a five-node reconstruction of

$z\approx2.26$20

Pantheon+ gives

$z\approx2.26$21

corresponding to a $z\approx2.26$22 upward deviation from the $z\approx2.26$23CDM value $z\approx2.26$24. In the same analysis, Pantheon+ combined with Planck15 CMB distance priors and DESI DR2 BAO yields $z\approx2.26$25 and $z\approx2.26$26, a $z\approx2.26$27 deviation from $z\approx2.26$28CDM in the $z\approx2.26$29 plane; after flux averaging, the fit shifts to $z\approx2.26$30 and $z\approx2.26$31 (Wang et al., 13 Apr 2026).

The same study argues that these departures track each dataset’s preferred $z\approx2.26$32: Pantheon+ favors $z\approx2.26$33 in SN-only flat $z\approx2.26$34CDM, while DESI DR2 BAO gives $z\approx2.26$35, and flux averaging reduces the Pantheon+ tension with DESI from $z\approx2.26$36 to $z\approx2.26$37 (Wang et al., 13 Apr 2026). This suggests a conservative reading of Pantheon+ residual features. The data support high-precision dark-energy inference, but the distinction between genuine dark-energy evolution, residual calibration or selection systematics, and inter-probe $z\approx2.26$38 inconsistency remains unresolved at the few-$z\approx2.26$39 level.

Pantheon+ therefore occupies a dual role in contemporary cosmology. It is both the largest homogeneous SN Ia compilation currently available and a stress test for the sub-percent control of calibration, redshift, bias-correction, and population-evolution systematics required by next-generation inferences from Euclid, Roman, LSST, DESI, and related programs.