MESA Isochrones and Stellar Tracks (MIST)

• MESA Isochrones and Stellar Tracks (MIST) is a comprehensive stellar-model library that combines MESA’s 1D evolution calculations with the EEP formalism to produce consistent isochrones and evolutionary tracks.
• The framework spans a wide parameter space including non-rotating and rotating models across solar-scaled, α-enhanced, and white-dwarf regimes, ensuring detailed coverage from the pre-main sequence to advanced evolutionary stages.
• MIST employs robust treatments of microphysics—such as convection, mass loss, and diffusion—and is validated through comparisons with observational data and asteroseismic benchmarks.

Searching arXiv for recent and foundational MIST papers to ground the article. MESA Isochrones and Stellar Tracks (MIST) is a stellar-model library and interpolation framework built on the one-dimensional stellar evolution code MESA, with the aim of providing self-consistent evolutionary tracks and isochrones across broad ranges of mass, age, composition, and evolutionary phase. In its solar-scaled release, MIST covers $0.1 \leq M/M_\odot \leq 300$, $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$, and $-2.0 \leq [Z/H] \leq 0.5$, with an extension to lower metallicity for models evolved from the pre-main sequence to the end of core helium burning; later releases add self-consistent $\alpha$-enhanced compositions and an explicit white-dwarf cooling-sequence extension (Choi et al., 2016, Dotter et al., 25 Feb 2026, Bauer et al., 26 Sep 2025). Methodologically, MIST is closely identified with the Equivalent Evolutionary Point (EEP) formalism for constructing isochrones from heterogeneous stellar tracks, allowing a single interpolation framework to span phases from the pre-main sequence to the white-dwarf cooling sequence (Dotter, 2016).

1. Historical development and project architecture

MIST emerged as a sequence of closely related releases rather than a single static grid. Paper 0 formalized the EEP-based interpolation machinery used to transform raw stellar tracks onto a uniform basis for track interpolation and isochrone construction (Dotter, 2016). Paper I presented the solar-scaled model library computed with MESA, including both theoretical and observational isochrones and extensive comparisons to other model sets and empirical constraints (Choi et al., 2016). Subsequent work used the MIST framework to analyze cluster ages in the Gaia era, emphasizing how different stellar-model parameters project onto different observables and how multi-observable fitting can break classical degeneracies (Choi et al., 2018). The project was then extended to $\alpha$-enhanced compositions over a two-dimensional abundance grid in $[\mathrm{Fe/H}]$ and $[\alpha/\mathrm{Fe}]$ (Dotter et al., 25 Feb 2026), and later to carbon–oxygen-core white dwarfs with hydrogen atmospheres descended from full progenitor calculations (Bauer et al., 26 Sep 2025).

The underlying computational engine is MESA (Modules for Experiments in Stellar Astrophysics), which solves the coupled stellar-structure and composition equations from the pre-main sequence through advanced phases. In the cluster-age analysis, MIST is described as using non-rotating tracks for the cluster fits, with a rotating grid published separately; the solar-scaled release itself includes both non-rotating and rotating families (Choi et al., 2018, Choi et al., 2016). The project therefore combines three layers: interior stellar-evolution calculations, EEP-based interpolation, and atmosphere/bolometric-correction machinery for observational predictions.

A central organizational feature is the distinction between model families defined by composition and physics assumptions. The solar-scaled grids adopt solar-scaled heavy-element mixtures and a helium-enrichment law. The later $\alpha$-enhanced database adds explicit abundance variation in O, Ne, Mg, Si, S, Ar, Ca, and Ti, with microphysics tables and synthetic spectra recomputed for each abundance pattern (Dotter et al., 25 Feb 2026). The white-dwarf extension adds a further endpoint-specific branch to the library rather than a separate stand-alone product (Bauer et al., 26 Sep 2025).

2. Stellar-physics content

The solar-scaled MIST release adopts a composite equation of state joining OPAL, SCVH, MacDonald, HELM, and PC tables across different thermodynamic regimes, together with OPAL Type I and Type II high-temperature opacities, Ferguson low-temperature opacities, and Cassisi electron-conduction opacities (Choi et al., 2016). Its default nuclear network is mesa_49.net, containing pp-chains, cold and hot CNO cycles, triple-$\alpha$, $\alpha$-captures up to $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$0, Ne-Na, Mg-Al, and C/O burning, with rates drawn from JINA REACLIB (Choi et al., 2016).

Convection is treated with Henyey MLT under the Ledoux criterion. In MIST I, the mixing length is

$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$1

and diffusive exponential overshoot is written

$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$2

with $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$3 and $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$4 (Choi et al., 2016). The same exponential-overshoot formalism appears in the cluster-age analysis as the parameterization used for H-core and He-core overshoot (Choi et al., 2018). Semiconvection and thermohaline mixing are also included in the solar-scaled release, with explicit diffusion prescriptions and adopted efficiencies listed in the model description (Choi et al., 2016).

Mass loss is versioned by evolutionary regime. For low-mass stars, the solar-scaled release uses Reimers on the RGB and Bloecker on the AGB,

$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$5

with $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$6, and

$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$7

with $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$8; for high-mass stars it adopts the Dutch scheme, including Vink, Nugis–Lamers, and de Jager prescriptions, with a mass-loss cap of $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$9 (Choi et al., 2016). The cluster-age analysis likewise treats the Reimers efficiency as a parameter relevant to age inference, noting that it primarily affects the mass difference between RGB and red-clump stars while leaving CMD loci nearly unchanged (Choi et al., 2018).

Composition is parameterized through a helium-enrichment law. In MIST I,

$-2.0 \leq [Z/H] \leq 0.5$0

with $-2.0 \leq [Z/H] \leq 0.5$1 and $-2.0 \leq [Z/H] \leq 0.5$2 (Choi et al., 2016). The cluster-age analysis presents the same linear form and notes that it ties the initial helium mass fraction to metallicity in the standard solar-scaled framework (Choi et al., 2018).

The $-2.0 \leq [Z/H] \leq 0.5$3-enhanced release modifies this baseline in several ways. It recomputes EOS, OPAL, Ferguson, and atmosphere boundary tables for each $-2.0 \leq [Z/H] \leq 0.5$4, replaces the earlier atomic-diffusion turbulence prescription, adopts a hybrid atmosphere boundary condition with a $-2.0 \leq [Z/H] \leq 0.5$5–$-2.0 \leq [Z/H] \leq 0.5$6 integration at $-2.0 \leq [Z/H] \leq 0.5$7 for $-2.0 \leq [Z/H] \leq 0.5$8, and uses a metallicity-dependent envelope overshoot parameter,

$-2.0 \leq [Z/H] \leq 0.5$9

constrained to $\alpha$0 (Dotter et al., 25 Feb 2026). Its solar calibration on the GS98 mixture yields $\alpha$1, $\alpha$2, $\alpha$3, and $\alpha$4 (Dotter et al., 25 Feb 2026). These release-to-release changes are part of the model content and should not be conflated with a single immutable “MIST physics” specification.

3. EEP formalism and isochrone construction

The methodological core of MIST is the EEP formalism introduced by Dotter. The basic reparameterization replaces direct interpolation in age or timestep with a mapping

$\alpha$5

where $\alpha$6 is an index labeling equivalent evolutionary phases across tracks of different initial mass (Dotter, 2016). This avoids phase mismatch when adjacent tracks have entered different evolutionary states.

Primary EEPs are physically defined landmarks such as the pre-main-sequence start, ZAMS, IAMS, TAMS, RGB tip, onset and termination of core-helium burning, TPAGB or carbon-burning onset, post-AGB crossing, and WD cooling sequence (Dotter, 2016). Between adjacent primary EEPs, fixed numbers of secondary EEPs are inserted using a positive-definite metric,

$\alpha$7

with coordinates such as $\alpha$8, $\alpha$9, or $\alpha$0 and tunable weights $\alpha$1 (Dotter, 2016). Typical tracks use roughly $\alpha$2–$\alpha$3 EEP points (Dotter, 2016).

Isochrone construction proceeds in two stages. First, each raw track is converted to the EEP basis. Second, for a desired age, one solves for the initial mass corresponding to each EEP by interpolating the age–mass relation at fixed EEP, then reads off $\alpha$4, $\alpha$5, $\alpha$6, $\alpha$7, surface abundances, and derived photometric quantities (Dotter, 2016). The solar-scaled MIST paper summarizes the same logic as interpolation in the three-dimensional grid $\alpha$8 on a uniform EEP mesh, with final isochrones containing more than 500 EEP points and more than 50 stellar properties (Choi et al., 2016).

A technically important feature is the treatment of non-monotonic age–mass behavior. Dotter identifies two physical transitions that can violate the usual monotonic assumption: the onset of a convective core near $\alpha$9 and the degenerate–nondegenerate He-ignition boundary near $[\mathrm{Fe/H}]$0 (Dotter, 2016). MIST handles incidental non-monotonicities by isotonic regression using the Pool Adjacent Violators algorithm and can alternatively export secondary branches when users wish to retain genuinely multi-valued isochrone structure (Dotter, 2016).

The same EEP framework remains operative in later releases. The $[\mathrm{Fe/H}]$1-enhanced models state explicitly that MESA tracks are converted to uniform EEP tracks via iso, after which EEP tracks are interpolated in mass for a given age to form isochrones (Dotter et al., 25 Feb 2026). The continuity of the interpolation formalism across versions is one of the project’s central design features.

4. Grid coverage and released model families

The solar-scaled MIST library spans initial masses from $[\mathrm{Fe/H}]$2 to $[\mathrm{Fe/H}]$3 in roughly 100 models with finer spacing near critical masses, an age grid of $[\mathrm{Fe/H}]$4 to $[\mathrm{Fe/H}]$5 in steps of $[\mathrm{Fe/H}]$6, and metallicities from $[\mathrm{Fe/H}]$7 to $[\mathrm{Fe/H}]$8 in steps of $[\mathrm{Fe/H}]$9 dex, plus an extension from $[\alpha/\mathrm{Fe}]$0 to $[\alpha/\mathrm{Fe}]$1 evolved from the pre-main sequence to the end of core-helium burning (Choi et al., 2016). Evolutionary endpoints depend on mass: low-mass tracks may terminate at TAMS if they do not reach helium burning within the modeled timespan; intermediate-mass low-mass-type tracks are followed through TPAGB, post-AGB, and into WD cooling; high-mass tracks are followed through carbon burning (Choi et al., 2016).

Rotation is represented by two model families in the v1.2-era description: $[\alpha/\mathrm{Fe}]$2 and $[\alpha/\mathrm{Fe}]$3 (Brandner et al., 2022). In MIST I, shellular rotation is imposed with solid-body rotation on the ZAMS and a ramp from $[\alpha/\mathrm{Fe}]$4 to $[\alpha/\mathrm{Fe}]$5 up to $[\alpha/\mathrm{Fe}]$6, with diffusive transport coefficients following Heger et al. and rotationally enhanced mass loss following Langer (Choi et al., 2016). The later $[\alpha/\mathrm{Fe}]$7-enhanced release again provides non-rotating and rotating families, with initial $[\alpha/\mathrm{Fe}]$8 for $[\alpha/\mathrm{Fe}]$9, tapered to zero at $\alpha$0, and adds a Tayler–Spruit dynamo prescription for $\alpha$1 (Dotter et al., 25 Feb 2026).

The $\alpha$2-enhanced grid expands the composition space to 74 compositions: $\alpha$3 from $\alpha$4 to $\alpha$5 in steps of $\alpha$6 dex and $\alpha$7, with the $\alpha$8, $\alpha$9 point omitted (Dotter et al., 25 Feb 2026). The abundance changes are applied self-consistently to both interior models and atmosphere/synthetic-spectrum grids. Bolometric-correction tables carry the same $\alpha$0 dimension, and the atmosphere grid spans $\alpha$1–$\alpha$2 and $\alpha$3 to $\alpha$4 (Dotter et al., 25 Feb 2026).

Observationally, MIST distributes both theoretical and photometric isochrones. The solar-scaled release provides .iso and .bc.iso files with magnitudes in approximately 30 photometric systems, including UBVRI, SDSS, 2MASS, Gaia, HST, JWST, Spitzer, and WISE (Choi et al., 2016). The later $\alpha$5-enhanced release preserves the general file structure while adding columns for gravity darkening, convective turnover times, and apsidal-motion $\alpha$6 (Dotter et al., 25 Feb 2026).

5. Empirical performance, validation, and known discrepancies

Validation has been a defining component of MIST since the solar-scaled release. Paper I reports solar calibration, comparisons to open clusters including M 67, Praesepe, the Pleiades, NGC 6791, and Ruprecht 106, detached eclipsing binaries, asteroseismic diagnostics, massive-star population ratios, and other model sets such as PARSEC, $\alpha$7, DSEP, BaSTI, and Lyon (Choi et al., 2016). In the open-cluster comparison, the models reproduce the MS, MSTO, RGB, and red-clump loci in multiband CMDs, but also exhibit the well-known mismatch below $\alpha$8–$\alpha$9, where the models are too blue in $\alpha$0 and $\alpha$1, attributed there to missing line opacities in bolometric corrections (Choi et al., 2016).

The cluster-age analysis sharpened this by examining how stellar age and model parameters such as $\alpha$2, $\alpha$3, $\alpha$4, and $\alpha$5 imprint differently on CMD morphology, mass–radius relations, surface abundances, asteroseismic $\alpha$6, and population ratios (Choi et al., 2018). For NGC 6819, MIST fits were reported as showing excellent agreement in the turn-off, subgiant, and clump; in M 67, the Henyey-hook morphology constrained core overshoot near $\alpha$7 but mild RGB-color and [C/N] tensions remained; in NGC 6791, no single fit reproduced all CMDs simultaneously, and Gaia DR2 parallax zero-point uncertainties were identified as a leading systematic (Choi et al., 2018).

The Hyades benchmark provides a more differential test of the single-star sequence. Using Gaia EDR3, non-rotating MESA models at $\alpha$8 fit stars above $\alpha$9 and below $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$00 well, but systematically underpredict the luminosity of stars between $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$01 and $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$02, by up to $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$03 mag (Brandner et al., 2022). That study points to the fixed solar-calibrated mixing-length treatment as a potential limitation for partially convective stars and notes increased scatter near the fully convective boundary, possibly related to the convective-kissing instability driven by $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$04He burning (Brandner et al., 2022).

A related empirical issue concerns Gaia colors. An analysis based on Hyades, Pleiades, and Praesepe found that current MIST isochrones are systematically too blue at low masses, with typical deviations of $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$05–$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$06 mag and $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$07–$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$08 mag for $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$09 (Wang et al., 2024). Polynomial color-correction functions were then derived for non-rotating, solar-scaled MIST isochrones, reducing the RMS color difference from $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$10–$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$11 mag to $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$12 mag and removing a systematic age bias of about $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$13 dex in $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$14 when the low-mass sequence is included (Wang et al., 2024). The same work explicitly states that the underlying cause of the offset remains unclear, listing model-atmosphere opacities, low-mass convection, and Gaia photometric calibration as possibilities (Wang et al., 2024).

MIST is also used as a forward model in downstream Bayesian inference. BRUTUS embeds MIST v1.2 isochrones in a joint photometric and astrometric likelihood with Galactic priors on distance, metallicity, age, dust, and $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$15, and applies empirical corrections to $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$16 and $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$17 on the lower main sequence together with photometric zero-point offsets derived from nearby field stars (Speagle et al., 4 Mar 2025). In that framework, the calibration is reported to reduce lower-main-sequence systematic errors to $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$18, and tests on mock and real data recover distances to $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$19, $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$20 to $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$21, and $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$22 to $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$23 mag when calibrated models and Gaia astrometry are used (Speagle et al., 4 Mar 2025). These results are not changes to MIST itself, but they document how MIST grids are operationally calibrated in precision inference workflows.

6. White-dwarf cooling sequence extension and current data ecosystem

The white-dwarf extension substantially broadens the evolutionary endpoint coverage of MIST. Previous versions ended at the end of the AGB, around $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$24; the updated library extends tracks and isochrones down the white-dwarf cooling sequence to $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$25 and cooling ages beyond $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$26 Gyr (Bauer et al., 26 Sep 2025). The released models are carbon–oxygen-core DA/DC white dwarfs descended from full progenitor calculations. For each metallicity point $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$27 in the 84-point MIST grid, approximately 100 WD tracks are produced with $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$28–$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$29, descended from progenitors of $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$30–$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$31 (Bauer et al., 26 Sep 2025).

The WD physics goes beyond a simple Mestel-like cooling treatment. Core compositions are inherited from full progenitor evolution, with typical central oxygen mass fractions $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$32–$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$33 and $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$34 from CNO $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$35 processing (Bauer et al., 26 Sep 2025). Envelope hydrogen and helium masses are computed self-consistently, with representative starting values around $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$36 for a $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$37 WD and $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$38 (Bauer et al., 26 Sep 2025). Time-dependent element diffusion is solved in Lagrangian form,

$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$39

and heavy-element sedimentation, residual nuclear burning, crystallization, latent heat, and carbon–oxygen phase separation are all included self-consistently (Bauer et al., 26 Sep 2025).

The cooling problem is summarized by

$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$40

with residual burning capable of delaying cooling by up to about $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$41 Gyr for low-$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$42, low-mass white dwarfs, and C/O phase separation delaying cooling by $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$43–$5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$44 Gyr around $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$45 (Bauer et al., 26 Sep 2025). For practical use, cooling-age contour files are distributed in $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$46 space, with both raw and LOWESS-smoothed versions and a stated bilinear interpolation recipe for retrieving $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$47 from measured mass and luminosity (Bauer et al., 26 Sep 2025).

Validation of the WD extension is both internal and external. The resulting initial–final mass relation emerges self-consistently from the full progenitor calculations and is reported to reproduce recent empirical constraints; the cooling tracks show good agreement with LPCODE, BaSTI, and STELUM to approximately $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$48 Gyr in cooling age, with later differences attributed to more modern treatments of crystallization and phase separation in MESA/Skye EOS (Bauer et al., 26 Sep 2025). A $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$49 WD track aligns precisely with the A-branch peak in the Gaia HRD, which is presented as confirmation that the adopted DA/DC bolometric corrections are well calibrated (Bauer et al., 26 Sep 2025).

Current access reflects the layered history of the project. Solar-scaled tracks and isochrones were originally distributed through the MIST project website at http://waps.cfa.harvard.edu/MIST/ (Choi et al., 2016). The $5 \leq \log(\mathrm{Age})\,[\mathrm{yr}] \leq 10.3$50-enhanced release states that data products are available from https://mist.science, with a direct data portal at the earlier WAPS site and full reproducibility files on Zenodo (Dotter et al., 25 Feb 2026). The WD extension distributes model tracks, isochrones, and cooling-age contours via the MIST website and Zenodo DOI 10.5281/zenodo.15242047, with full MESA inlists and run directories at DOI 10.5281/zenodo.15196934 (Bauer et al., 26 Sep 2025). Across these releases, MIST functions both as a published stellar-model database and as a reproducible framework for generating new tracks, bolometric-correction tables, and isochrones under controlled physics assumptions.