# Alpha-enhanced SSP grids

Simple stellar population (SSP) spectra with an [α/Fe] axis, in the DSPS/tengri HDF5 layout extended by one
axis. Every file holds

| Dataset | Shape | Meaning |
|---|---|---|
| `ssp_flux` | (n_met, n_alpha, n_age, n_wave) | L_sun / Hz per M_sun formed |
| `ssp_alpha_fe` | (n_alpha,) | [α/Fe] = −0.2, 0.0, +0.2, +0.4, +0.6 |
| `ssp_lgmet` | (n_met,) | log10 of the metallicity node (see "Metallicity axis") |
| `ssp_lg_age_gyr` | (n_age,) | log10(age / Gyr) |
| `ssp_wave` | (n_wave,) | vacuum wavelength, Å |

None of the grids contains nebular emission (there are no α-enhanced CLOUDY tables). Each file records its
source, licence, build flags and caveats in its HDF5 attributes.

## Files

| File | Models | IMF | Shape (Z, α, age, λ) | Wavelength | Ages |
|---|---|---|---|---|---|
| `fsps_amist_c3k_lr_afe_{chabrier,kroupa,salpeter}.h5` | α-MC built with FSPS v4.0: MIST v2.5 + C3K v2.3 low resolution | Chabrier, Kroupa, Salpeter | 13 × 5 × 107 × 1936 | 100 Å – 1 cm, R = 100 (R = 50 beyond 24 μm) | 0.1 Myr – 20 Gyr |
| `amist_c3k_hr_kroupa_afe.h5` | α-MC high resolution (M. J. Park SSPs of 2025-07-22, from the CERIDWEN release) | Kroupa | 13 × 5 × 107 × 10992 | 100 Å – 1 cm, λ/Δλ = 6000 per pixel over 3000–9000 Å | 0.1 Myr – 20 Gyr |
| `fsps_amist_c3k_lr_afe_{chabrier,kroupa,salpeter}_totalz.h5`, `amist_c3k_hr_kroupa_afe_totalz.h5` | the four α-MC grids above, resampled to fixed total metallicity (see below) | as above | 10 × 5 × 107 × (1936 or 10992) | as above | as above |
| `smiles_bsti_<imf>_afe.h5` (29 files) | sMILES semi-empirical SSPs | `kroupa` (universal), `kroupa_revised`, `chabrier`, `unimodal<slope>` and `bimodal<slope>` with slopes 0.30–3.50 (13 each). Unimodal is a single power law with logarithmic slope Γ (`unimodal1.30` is Salpeter); Bimodal has the same slope above 0.6 M_sun and flattens below it (Vazdekis et al. 1996) | 10 × 5 × 53 × 4300 | 3541.5–7411.6 Å, 0.9 Å pixels, FWHM 2.5 Å | 0.03 – 14 Gyr |
| `bpss_v23_{bin,sin}_imf135_300_afe.h5` | BPASS v2.3, binary and single-star populations | BPASS `imf135_300` | 13 × 5 × 51 × 100000 | 1 – 100000 Å, 1 Å steps | 1 Myr – 100 Gyr |

## Metallicity axis

- **α-MC grids (both):** `ssp_lgmet = [Fe/H] + log10(0.0185)`, the same at every [α/Fe]. A node is a
  fixed [Fe/H], so total metallicity rises with [α/Fe]. [Fe/H] runs from −2.5 to +0.5 in steps of 0.25.
- **α-MC `*_totalz.h5` versions:** `ssp_lgmet = log10(Z)` with Z the total metallicity, Z = 0.0185 × 10^[Z/H],
  [Z/H] = −2.0 to +0.25 in steps of 0.25 at every [α/Fe] (the range covered at every [α/Fe]). Made from the
  native files by linear interpolation in log Z at each [α/Fe] between the two bracketing [Fe/H] nodes, with
  [Z/H] = [Fe/H] + log10(1 − x + x·10^[α/Fe]), x = 0.68749 (shifts −0.127, 0, +0.147, +0.310, +0.484 dex for
  [α/Fe] = −0.2 … +0.6). The [α/Fe] = 0 models are identical to the native nodes. `ssp_zh` holds [Z/H] and
  `ssp_feh` (n_met, n_α) the [Fe/H] of each node; the surviving-mass tables were resampled the same way.
  A leave-one-out test on the native grid puts the interpolation error at a median 0.4–0.8% in broad bands
  (10 Myr – 10 Gyr). These files share the convention of the sMILES, MILES and BPASS grids.
- **sMILES:** `ssp_lgmet = log10(Z)` with Z the total metallicity (0.0003 to 0.0300); `ssp_mh` holds
  [M/H]_SSP (−1.79 to +0.26).
- **BPASS v2.3:** `ssp_lgmet = log10(Z)` with Z the total metallicity (1e-5 to 0.040, Z_sun = 0.020).
  [α/Fe] varies at fixed total Z, so α-rich models are Fe-poor.

## Caveats

- **FSPS low-resolution grids are for photometry.** R = 100 does not resolve Mg b or the Fe lines.
- **FSPS v4.0 normalization.** Spectra synthesized by FSPS up to v4.0 are about 2% brighter than the bolometric
  luminosity of the isochrones, because of an inaccurate constant in the code (cconroy20/fsps#83; a fix is
  proposed in cconroy20/fsps#88 for v4.1). The three FSPS-built files here carry that offset, as do all
  FSPS-based SSPs built on v4.0 or earlier. Their fluxes are in units of the FSPS solar luminosity,
  3.839e33 erg/s (attribute `lsun_erg_per_s`).
- **Surviving mass.** `ssp_mass_remaining_alpha` (n_met, n_alpha, n_age) holds living stars plus remnants per
  solar mass formed, and `ssp_mass_remaining` (n_met, n_age) is its scaled-solar ([α/Fe] = 0) slice.
  - α-MC (FSPS v4.0): FSPS `stellar_mass`. It exceeds 1 on MIST v2.5 below about 10^6.5 yr; those cells are
    capped at 1 and the raw table is kept as `ssp_mass_remaining_alpha_fsps_raw`.
  - α-MC high resolution: the FSPS v4.0 Kroupa table (same isochrones, IMF and nodes; the table does not
    depend on the spectral library).
  - sMILES: column M(*+remn) of the MILES v11.0 mass tables for BaSTI isochrones, scaled-solar for
    [α/Fe] ≤ +0.2 and α-enhanced for +0.4 and +0.6, per IMF.
  - BPASS v2.3: none; the release has no stellar-mass files.
- **The two α-MC lineages agree.** Band-averaged fluxes of the Kroupa low- and high-resolution grids agree
  within 1% at 1 and 10 Gyr for every [Fe/H] and [α/Fe] (rebuilt files, 2026-11-13), and differ by about 5% in
  the ultraviolet for ages ≤ 10 Myr. The FSPS v4.0 files first published here were up to 2.8× brighter over
  1500–2500 Å at [Fe/H] ≥ +0.25: FSPS v4.0 read only 11 of the 13 C3K metallicities (nzinit = 11), so the
  super-solar models used the solar stellar spectra. The bug is fixed upstream (cconroy20/fsps#89, commit
  7572834, after the v4.0 release and python-fsps 0.5.0); the `fsps_amist_c3k_lr_afe_*` files were rebuilt with
  the fix on 2026-11-13 and differ from the first release only at [Fe/H] = +0.25 and +0.5.
- **[Fe/H] = +0.5, [α/Fe] = +0.6** uses the same isochrone as [α/Fe] = +0.4 (FSPS ships one file for both).
- **BPASS v2.3: only the atmospheres are α-enhanced.** The spectra of main-sequence and giant stars change with
  [α/Fe]; the stellar evolution models are the solar-scaled BPASS v2.2 models, and the hot low-gravity,
  Wolf-Rayet and white-dwarf spectra are also solar-scaled (BPASS v2.3 README). [α/Fe] varies at fixed
  total Z, so the α-enhanced models are iron-poor.
- **BPASS v2.3** is on the native 1 Å grid.
  The BPASS manual does not state whether wavelengths are in air or vacuum; they are stored as distributed.
  The IMF is the BPASS default (slope −1.30 over 0.1–0.5 M_sun, −2.35 over 0.5–300 M_sun).
- **sMILES** covers the optical only and starts at 30 Myr. Wavelengths were converted from air to vacuum
  and fluxes from L_sun/Å to L_sun/Hz (the MILES website states air wavelengths and L_λ/L_sun per Å per
  M_sun for a total SSP mass of 1 M_sun; an FSPS MILES SSP agrees to 0.95–1.02 over 4000–7300 Å at 10 Gyr).
  MILES relates iron and total metallicity by [Fe/H] = [M/H] − 0.75 [Mg/Fe]. Isochrones
  are scaled-solar for [α/Fe] ≤ +0.2 and [α/Fe] = +0.4 for the two highest [α/Fe] values.

## Citations and licences

Please cite the original work when using these files. Each file repeats its sources, licence and the changes
made here in its HDF5 attributes.

- **FSPS** (MIT License, © 2009–2021 Charlie Conroy & contributors; https://github.com/cconroy20/fsps). FSPS
  asks users to cite Conroy C., Gunn J. E., White M., 2009, ApJ, 699, 486 and Conroy C., Gunn J. E., 2010, ApJ,
  712, 833. Built here with FSPS v4.0, https://github.com/cconroy20/fsps/releases/tag/v4.0.
- **python-fsps** (MIT License) asks users to follow the FSPS citation requirements and to reference the bindings:
  Johnson B., Foreman-Mackey D., Sick J., Leja J., Walmsley M., Tollerud E., Leung H., Park M., 2026,
  dfm/python-fsps: v0.5.0, Zenodo, doi:10.5281/zenodo.21778582 (all versions: doi:10.5281/zenodo.591505).
- **α-MC models:** Park M., Conroy C., Johnson B. D., Leja J., Dotter A., Cargile P. A., 2025, ApJ, 994, 165,
  doi:10.3847/1538-4357/ae0cba (arXiv:2410.21375).
- **α-enhanced MIST isochrones:** Dotter A., Bauer E., Park M. J., Conroy C., et al., 2026, "MESA Isochrones
  and Stellar Tracks. II. Models with alpha-enhanced chemical composition", ApJS, 283, 64 (arXiv:2602.22012).
- **CERIDWEN grids** (`amist_c3k_hr_kroupa_afe.h5`): adapted from `amist_c3k_hr_krou_afe.h5` in Stoffers A.,
  2026, "CERIDWEN Stellar Population Grids: Standard SSPs, BPASS, and Alpha Enhancement" (version v6), Zenodo,
  doi:10.5281/zenodo.22937956, licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).
  Changes made: flux axes reordered from (afe, met, age, wave) to (met, alpha, age, wave), `ssp_afe` renamed
  `ssp_alpha_fe`, `ssp_resolution` not copied, attributes added, surviving-mass tables attached. Flux values
  are unchanged. The CERIDWEN code is Stoffers A., "CERIDWEN: Fast and Flexible GPU-Accelerated Stellar
  Population Inference", https://github.com/Espe13/ceridwen (MIT License).
- **sMILES** (`smiles_bsti_*_afe.h5`): adapted from Knowles A. T., Sansom A. E., Vazdekis A., Allende Prieto C.,
  2023, "sMILES SSPs - A library of semi-empirical MILES stellar population models with variable [alpha/Fe]
  abundances" [Dataset], UCLanData, doi:10.17030/uclan.data.00000388, licensed under CC BY-NC 4.0
  (https://creativecommons.org/licenses/by-nc/4.0/): attribution required, non-commercial use only. The
  adapted files here are shared under the same terms. Changes made: spectra assembled into one array per
  IMF, wavelengths converted from air to vacuum, flux from L_sun/Å to L_sun/Hz, [M/H] expressed as log10(Z),
  MILES surviving-mass tables added. The authors state: "Whilst these semi-empirical SSP spectra are made
  publicly available for anyone to download and use, this is at the users own risk. The author does not
  guarantee that they are free from errors. If you make use of these data please acknowledge Knowles et al.
  [2023]." That paper is Knowles A. T., Sansom A. E., Vazdekis A., Allende Prieto C., 2023, MNRAS, 523, 3450,
  doi:10.1093/mnras/stad1647. Full library from the MILES website (https://research.iac.es/proyecto/miles/).
  Mass tables: MILES SSP models v11.0 (Vazdekis A., et al., 2010, MNRAS, 404, 1639; Vazdekis A., et al., 2015,
  MNRAS, 449, 1177).
- **BPASS v2.3:** Byrne C. M., Stanway E. R., Eldridge J. J., McSwiney L., Townsend O. T., 2022, MNRAS, 512,
  5329, doi:10.1093/mnras/stac807; Stanway E. R., Eldridge J. J., 2018, MNRAS, 479, 75; Eldridge J. J., et al.,
  2017, PASA, 34, e058. The BPASS team asks for: "This work made use of v2.3 of the Binary Population and
  Spectral Synthesis (BPASS) models as last described in Byrne et al (2022) and Stanway & Eldridge (2018)."
- **Surviving mass (FSPS):** Renzini A., Ciotti L., 1993, ApJ, 416, L49 (as implemented in FSPS).
- **DSPS file layout:** Hearin A. P., et al., 2023, MNRAS, 521, 1741 (arXiv:2112.06830).
