PaperlayerView exact revision on arXiv

Source-linked AI summary

Modules for Experiments in Stellar Astrophysics (MESA)

Bill Paxton, Lars Bildsten, Aaron Dotter, Falk Herwig, Pierre Lesaffre, Frank Timmes

arXiv:1009.1622v1astro-ph.SRastro-ph.IM

TL;DR

Stellar structure and evolution models are central to interpreting starlight, but researchers need broadly applicable computational tools for diverse stellar-physics applications. MESA provides modular open-source libraries and MESA star, supporting tested evolutionary calculations across a wide range of stellar objects and phases.

  • Problem

    Interpreting starlight and studying stars requires detailed models of stellar atmospheres, structure, and evolution across diverse astrophysical applications.

  • Method

    MESA combines independently usable physics libraries with the MESA star evolution code, using coupled equations, adaptive meshing, timestep control, and ongoing verification.

  • Results

    MESA supports evolutionary tracks for very-low-mass objects, gas giants, intermediate-mass and massive stars, pulsating stars, and accreting compact objects.

  • Takeaways & Limitations

    MESA offers a portable, open-source framework for astrophysicists to explore stellar evolution across a wide range of masses and metallicities.

  • Takeaways & Limitations

    Incomplete treatment of liquid diffusion prevents verification of MESA star’s resulting white-dwarf cooling sequences.

Abstract

from arXiv · show

Stellar physics and evolution calculations enable a broad range of research in astrophysics. Modules for Experiments in Stellar Astrophysics (MESA) is a suite of open source libraries for a wide range of applications in computational stellar astrophysics. A newly designed 1-D stellar evolution module, MESA star, combines many of the numerical and physics modules for simulations of a wide range of stellar evolution scenarios ranging from very-low mass to massive stars, including advanced evolutionary phases. MESA star solves the fully coupled structure and composition equations simultaneously. It uses adaptive mesh refinement and sophisticated timestep controls, and supports shared memory parallelism based on OpenMP. Independently usable modules provide equation of state, opacity, nuclear reaction rates, and atmosphere boundary conditions. Each module is constructed as a separate Fortran 95 library with its own public interface. Examples include comparisons to other codes and show evolutionary tracks of very low mass stars, brown dwarfs, and gas giant planets; the complete evolution of a 1 Msun star from the pre-main sequence to a cooling white dwarf; the Solar sound speed profile; the evolution of intermediate mass stars through the thermal pulses on the He-shell burning AGB phase; the interior structure of slowly pulsating B Stars and Beta Cepheids; evolutionary tracks of massive stars from the pre-main sequence to the onset of core collapse; stars undergoing Roche lobe overflow; and accretion onto a neutron star. Instructions for downloading and installing MESA can be found on the project web site (http://mesa.sourceforge.net/).

1. Introduction

MESA is an open-source suite for stellar astrophysics that combines modern numerical methods, updated physics, and modular software design. Its MESA star module enables broad stellar-evolution applications, including challenging phases such as the He core flash and advanced nuclear burning.

  • Motivation: Stellar structure and evolution models are essential for interpreting starlight and studying objects including the Sun and pulsating stars.The introduction identifies stellar atmospheres, structure, and evolution as central to interpreting astronomical observations.
  • MESA design: MESA began as an effort to improve the EZ stellar evolution code and evolved toward carefully structured, independently valuable software components.Its design uses modern software engineering techniques targeting contemporary computer architectures.
  • MESA star: MESA star uses sophisticated numerical methods and updated physics to evolve stellar models through challenging phases across a wide range of applications.Examples include the He core flash in low-mass stars and advanced nuclear burning in massive stars.
  • Community and scope: MESA is open source, allowing users to download, compile, and run the code while contributing tests, fixes, capabilities, and shared experience.The project is intended to engage the broader astrophysics community.
  • Paper scope: The paper introduces MESA and demonstrates its capabilities through descriptions of modules, MESA star, tests, code comparisons, and broader applications.The paper assumes familiarity with basic stellar physics and numerical methods.

2. Module design and implementation

MESA modules separately organize the numerics and physics needed for stellar models, with standardized interfaces, implementations, build systems, and verification. They are thread-safe through read-only shared data after initialization, enabling concurrent execution on multicore processors.

  • Module organization: Each MESA module addresses a distinct numerical or physical aspect of constructing computational stellar-astrophysics models.
  • Module organization: Modules share a public interface, private implementation, makefile, test suite, and installation script that builds, tests, and exports successful libraries.
  • Parallel execution: Thread safety permits multiple processes to execute module routines simultaneously, supporting applications that use multicore processors.
  • Parallel execution: Shared data must be read-only after initialization, while working memory is kept in routine-local or dynamically allocated variables.This design prohibits common blocks and SAVE statements.

3. Numerical methods

MESA provides modular numerical methods for linear algebra, interpolation, differential equations, nonlinear root-finding, diagnostics, and utility operations. These components support flexible matrix backends, controlled interpolation and integration, and solution of highly nonlinear systems with tens of thousands of variables.

  • Linear algebra: The mtx module supports dense and sparse linear algebra through BLAS/LAPACK, SPARSKIT, and an interface to Intel PARDISO, while permitting alternative packages such as GotoBLAS or MKL.Its interface can be modified to use other linear algebra libraries.
  • Interpolation: The interp 1d and interp 2d modules provide one- and two-dimensional interpolation, including monotonic methods that prevent values from exceeding the supplied range.The one-dimensional methods include piecewise monotonic cubic and monotonicity-preserving approaches.
  • Differential equations and root-finding: The num module supplies stiff and non-stiff ODE solvers plus multidimensional Newton-Raphson root-finding with adaptive control, Jacobian options, reuse, and line search.The non-stiff solvers include explicit Runge-Kutta integrators of orders 5 and 8 with dense output.
  • Differential equations and root-finding: MESA’s Newton-Raphson solver handles square, banded, and sparse matrices and is used by MESA star for differential-algebraic systems with tens of thousands of variables.Its implementation derives from Lesaffre’s version of the Eggleton stellar evolution code.
  • Diagnostics and utilities: The alert and utils modules provide terminal reporting, invalid-value checks, Fortran I/O tracking, file I/O, array allocation, and hash-tree support for updating the stellar-evolution mesh.The hash tree is implemented in Fortran.

4. Microphysics

MESA microphysics supplies modular treatments of stellar matter, including an EOS assembled from OPAL, SCVH, HELM, and PC sources with tabulated interpolation and regime-specific extensions. Its opacity module combines OPAL, OP, and Ferguson tables, while `other eos` and `other kap` allow users to incorporate alternative physics.

  • Equation of state: The eos module uses density ρ and temperature T as independent variables, while preprocessed Pgas–T tables avoid repeated root finds for pressure-based calculations.A root find recovers ρ from Pgas = P − aT^4/3 and T; tabulation reduces the resulting computational overhead.
  • Equation of state: MESA EOS tables combine the 2005 OPAL update with SCVH tables at lower temperatures and densities, using a smooth transition in their overlapping region.The limited thermodynamic information restricts blending to the output quantities listed in Table 3.
  • Equation of state: Outside the MESA tables, HELM handles high-temperature pair production, while PC treats crystallization and mixtures in cooling white-dwarf cores.HELM reaches log T = 13 and log ρ = 15; PC is selected by default for Γ > 80 and HELM for Γ < 40.
  • User extensions: The `other eos` and `other kap` modules let users replace standard EOS or opacity treatments, enabling experiments with new schemes such as dust-driven-wind physics in karo.A flag directs MESA star to call `other kap` instead of `kap`.
  • Opacity: The kap module builds MESA opacity tables from Type 1 and 2 OPAL tables, OP tables, and Ferguson et al. (2005), blending sources with a smooth interpolation function.The interpolation uses F = (log T − log TL)/(log TU − log TL) and S = (1 − cos(Fπ))/2.
  • Opacity: For log R > 1 and log T < 8.2, kap reuses radiative opacity at log R = 1, introducing errors before conductive opacity becomes dominant.The passage notes that main-sequence stars are always efficiently covered despite this limitation.

5. Macrophysics

MESA’s macrophysics modules model convection, overshoot mixing, stellar atmospheres, and microscopic diffusion, while validation shows close agreement with independent stellar-evolution calculations. These modules provide configurable physical prescriptions and interfaces for use in MESA star.

  • Convection: The mlt module implements standard Cox–Giuli mixing-length convection, computes temperature gradients or convective luminosities, and returns diffusion coefficients and partial derivatives.It also supports the Henyey et al. modification, which accounts for radiative losses and opacity-dependent convective efficiency near stellar surfaces.
  • Overshoot mixing: MESA star treats convective mixing as time-dependent diffusion and models boundary overshoot with an adjustable exponentially declining diffusion coefficient.Overshoot parameters can differ by boundary and burning regime, with controls for minimum diffusion, spatial extent, and enhanced third-dredge-up mixing.
  • Atmospheres: The atm module derives surface pressure and temperature from stellar mass, radius, and luminosity using direct integrations, atmosphere-table interpolation, or a simple recipe.Its tables span log Z/Z⊙ = −4 to +0.5, log(g) = −0.5 to 5.5, and Teff = 2,000–50,000 K.
  • Diffusion: MESA diffusion solves particle diffusion and gravitational settling with Burger’s equations and a semi-implicit finite-difference transport scheme, excluding radiative levitation.For H, O, and Fe, the calculated diffusion velocities agree excellently with Thoul et al. (1994) when Fe is assigned the comparison ionization state.
  • Validation: 1 part in 10^4: DSEP tracks using the atm or mlt modules agree with the DSEP-only track to about this level.Even with kap and eos, the main-sequence lifetime differs by less than 0.3% and Teff by less than 10 K; the largest discrepancy occurs in the central Tc–ρc diagram at ρc > 3 × 10^4 g cm−3.

6. Stellar structure and evolution

MESA star is a full-featured 1-D stellar structure and evolution library that couples structure and composition equations while using adaptive meshing, finite-volume methods, and controlled timestepping. Its numerical design supports stable, convergent stellar evolution calculations and thread-safe parallel execution.

  • Numerical formulation: MESA star simultaneously solves coupled stellar structure and composition equations on adaptively refined, spherically symmetric 1-D meshes.The implementation uses a Henyey-style scheme, analytic Jacobians, and cells ranging from hundreds to thousands depending on model complexity and tolerances.
  • Numerical formulation: Finite-volume, flux-conservation variables improve numerical stability and efficiency in MESA star.Mass-averaged cell variables and outer-face variables are defined consistently with the finite-volume formulation.
  • Mixing and convergence: Holding the Lagrangian diffusion coefficient fixed during implicit iterations significantly improves numerical convergence.The coefficient represents combined convection and overshoot mixing effects and introduces a small inconsistency between final mixing and convection boundaries.
  • Timestep selection: MESA star selects timesteps in two stages, combining digital-control-based proposals with tests that reduce steps when model properties change too rapidly.The procedure balances convergence in relatively few iterations against efficient evolution while avoiding timestep over-corrections.
  • Convergence tests: α ∼1.6 describes super-linear convergence at smaller C, while larger C values converge linearly.These orders are consistent with first-order time integration and second-order finite-volume differencing on uniform grids.
  • Parallel execution: MESA modules are thread-safe, enabling parallel execution across calculations using different numbers of threads.Execution-time scaling differs among modules because parallel overhead varies relative to actual task time.

7. MESA star results: comparisons and capabilities · 7.1. Low mass stellar structure and evolution

MESA star is verified through module, mesh, timestep, and code comparisons, while demonstrating broad capabilities from substellar objects to massive stars and specialized accretion scenarios. Its low-mass results reproduce expected stellar and substellar evolution, agree closely with comparison models, and yield a calibrated Solar model consistent with helioseismic constraints.

  • 7. MESA star results: comparisons and capabilities: MESA star undergoes ongoing verification and validation through module tests, refinement studies, comparisons with specific modules, and stellar evolution code comparisons.The MESA test suite tests code modules individually and, where possible, verifies and validates the integrated MESA star code.
  • 7. MESA star results: comparisons and capabilities: MESA star models single stellar and substellar objects from 10^-3M⊙ to 1000M⊙ and include Solar comparisons, neutron-star He accretion, and cataclysmic-variable mass transfer.The paper presents evolutionary calculations and verification results across sections 7.1–7.3, with additional experiments in §7.4.
  • 7.1. Low mass stellar structure and evolution: MESA star evolves low-mass stars and substellar objects down to approximately 10^-3M⊙, including complete sequences from the pre-main sequence to white-dwarf cooling without intervention.For 1 and 1.25M⊙ models at Z = 0.01, the 1.25M⊙ model exhibits a late He-shell flash during the pre-white-dwarf phase.
  • 7.1. Low mass stellar structure and evolution: Models spanning 0.9–2M⊙ at Z = 0.02 converge to nearly identical degenerate helium cores on the RGB, while the 2M⊙ model ignites helium at lower degeneracy.The calculations continue through the end of the TP-AGB and show subsequent core evolution in the Tc–ρc plane.
  • 7.1.1. Low mass pre-main sequence stars, contracting brown dwarfs and giant planets: For M < 0.3M⊙, fully convective objects separate into main-sequence stars above 0.08M⊙ and brown dwarfs below that threshold through their central-condition and luminosity evolution.Only M > 0.08M⊙ objects asymptote to constant late-time luminosity; lower-mass objects continue fading, with L ∝ t^-2/3 during contraction.
  • 7.1.1. Low mass pre-main sequence stars, contracting brown dwarfs and giant planets: Beyond 10 Myr, MESA star tracks for 0.09–0.001M⊙ objects overlap BCBAH models at almost every point.The comparison uses XD = 2 × 10^-5; the youngest ages remain uncertain because of differing assumptions about deuterium burning.
  • 7.1.2. Code comparisons of 0.8M⊙and 1M⊙models: Code comparisons use matched physics for 0.8M⊙, Z = 10^-4 and 1M⊙, Z = 0.02 models, revealing near-identical central conditions until the red-giant phase but greater track scatter when convection is prominent.For the 1M⊙, Z = 0.02 case, convection treatment and surface-convection-zone resolution primarily cause the main-sequence spread; MESA star and BaSTI/FRANEC have the shortest lifetimes.
  • 7.1.3. The MESA star Solar model: The calibrated Solar model reaches adopted parameters to better than one part in 10^5, matches convection-zone depth within 1-σ and surface helium abundance within 2-σ, and is compared with helioseismic sound-speed data.Calibration varies αMLT, initial Yi, and Zi while including diffusion.

7.2. Intermediate Mass Structure and Evolution

MESA star models intermediate-mass stellar evolution from the pre-main sequence through He burning and the thermal-pulse AGB, with results generally agreeing with established stellar-evolution calculations. Comparisons show close agreement in evolutionary tracks, thermal-pulse properties, and surface C/O evolution, while incomplete liquid diffusion limits verification of white-dwarf cooling sequences.

  • Intermediate-mass evolution: MESA star evolves 2–10 M⊙ models through He-core burning and advanced thermal-pulse AGB phases, producing results compatible with published stellar-evolution codes.The grid uses Z = 0.02 models from the pre-main sequence to the end of the thermally pulsating AGB.
  • Intermediate-mass evolution: 8 and 10 M⊙ models ignite carbon burning off center, whereas 2–7 M⊙ models produce C/O white dwarfs.The models span 2–10 M⊙ at Z = 0.02.
  • Limitations: Incomplete treatment of liquid diffusion prevents verification of the resulting white-dwarf cooling sequences.This limitation applies to the 2–7 M⊙ models that produce C/O white dwarfs.
  • MESA star–EVOL comparison: Key properties of the 2 M⊙, Z = 0.01 MESA star and EVOL sequences differ by less than 5%, with tracks comparing well in the H-R diagram.The EVOL sequence is approximately 4% older at the first thermal pulse because it mixes more helium into the core.
  • Thermal-pulse AGB: MESA star reproduces thermal-pulse AGB behavior, including complex He-shell mixing and nucleosynthesis, with excellent agreement in interpulse time, core growth, and dredge-up amount.The calculations remain smooth and continuous during advanced thermal pulses and beyond.
  • Thermal-pulse AGB: The MESA star and EVOL models show good agreement in He-shell-flash intershell abundances and surface C/O evolution, including the timing and magnitude of C/O increases from repeated third dredge-up.The 2 M⊙ models do not undergo hot-bottom burning, making C/O a probe of cumulative third-dredge-up efficiency.

7.3. High Mass Stellar Structure and Evolution

MESA star evolves massive stars through advanced burning stages to core collapse, producing core tracks consistent with other stellar evolution codes and detailed pre-supernova structures. Comparisons show broadly similar hydrogen-burning lifetimes, while a zero-metallicity 1000M⊙ model demonstrates the module’s extreme-mass capability.

  • Massive-star calculations: MESA star evolves 15M⊙, 20M⊙, 25M⊙ solar-metallicity models and a 1000M⊙ zero-metallicity model from the Hayashi track to core collapse.The massive-star calculations use a 21-isotope nuclear reaction network; the models are non-rotating and include exponential diffusive overshoot with f=0.01.
  • Code comparisons: The 25M⊙ MESA star core-evolution track is consistent with Kepler, Hirschi et al. (2004), and FRANEC through iron-core collapse.A bump around carbon burning results from central convection developing in MESA star but not in the other codes.
  • Pre-supernova structure: At core collapse, the 25M⊙ model has an iron core at m ≤1.61M⊙, silicon shell at 1.61 ≤m/M⊙≤1.88, and oxygen shell at 1.88 ≤m/M⊙≤2.5.The infall speed reaches ≈1000 km s−1 just inside the iron core, while Ye drops below ≈0.48.
  • Quantitative comparisons: Hydrogen-burning lifetimes for 15M⊙, 20M⊙, and 25M⊙ models from different authors are within 10% of each other.Differences in lifetime definitions likely contribute to scatter; Limongi et al. (2000) generally gives the shortest lifetimes and Woosley et al. (2002) the longest.
  • Extreme-mass demonstration: A 1000M⊙, zero-metallicity, zero-mass-loss model reaches the onset of core collapse at very high central temperature and density.Its central-temperature–density trajectory demonstrates MESA star’s capability for purely metal-free stellar evolution at extreme mass.

7.4. Stellar Evolution with Mass Transfer

MESA star models stellar responses to mass loss and accretion, including cataclysmic-variable mass transfer and thermonuclear Type I X-ray bursts on accreting neutron stars. These examples demonstrate detailed resolution and timestep control while identifying lower-rate dynamical events beyond its present scope.

  • Cataclysmic-variable mass transfer: MESA star models mass transfer in compact binaries, including systems with low-mass Roche-lobe-filling donors and accreting white dwarfs.Applications include cataclysmic variables, accretion onto white dwarfs, and responses to sudden mass loss.
  • Cataclysmic-variable mass transfer: 67.4 minutes is the orbital-period minimum reached by modeled cataclysmic-variable tracks after their initial mass-transfer adjustment.All tracks accrete onto a MWD = 0.6M⊙ white dwarf and converge toward the same trajectory.
  • Accreting neutron-star bursts: 9.56 hours is the recurrence time of periodic Type I X-ray bursts in a MESA star simulation of a neutron star accreting pure helium at Ṁ = 3 × 10^-9M⊙yr^-1.The modeled neutron star has Mc = 1.4M⊙ and Rc = 10 km.
  • Accreting neutron-star bursts: ≈1.5×10^-20M⊙ is the achieved mass resolution, while timestep control changes smoothly from almost an hour between bursts to millisecond steps at peak luminosity.The model uses nonzero center boundary conditions so computational variables cover only the relatively small envelope mass.
  • Modeling scope: Lower accretion rates produce dynamical events beyond MESA star’s present scope, although it can follow the preceding longer-timescale hydrostatic evolution.Multidimensional hydrodynamical codes may be needed to follow the details once hydrodynamic effects dominate.

8. Summary and conclusion

MESA provides open source, portable, robust, efficient, thread-safe libraries and a modern stellar-evolution module for a broad range of stellar masses, metallicities, and evolutionary scenarios. Its ongoing testing supports current applications while future versions will add physics, user-driven features, and architectural refinements.

  • Capabilities: MESA provides open source, portable, robust, efficient, thread-safe libraries for stellar astrophysics across a broad range of stellar masses and metallicities.State-of-the-art modules include the equation of state, opacity, nuclear reaction rates and networks, atmosphere boundary conditions, and element diffusion.
  • Capabilities: MESA star solves fully coupled structure and composition equations simultaneously and calculates complete evolutionary tracks without user intervention.It uses adaptive mesh refinement, sophisticated timestep adjustment, mass loss and accretion, and OpenMP parallelism.
  • Validation and scope: Ongoing testing and verification support evolutionary tracks for very low mass objects, gas giant planets, intermediate-mass and pulsating stars, accreting compact objects, and massive stars.The massive-star calculations extend from the pre-main sequence to late times.
  • Future development: Future MESA versions will add new physics modules, features driven by the user community, and architectural refinements.These developments are identified as future capabilities following the ongoing testing and verification process.

A. Manifesto

MESA was developed over six years through the concerted efforts of the lead author and many theoretical and computational astrophysicists. Its public availability is intended to support education, scientific research, and outreach, while guiding its use, stewardship, and future development.

  • A. Manifesto: MESA was developed over six years through the lead author’s concerted efforts and deep involvement from many theoretical and computational astrophysicists.The development involved sustained collaboration across theoretical and computational astrophysics.
  • A. Manifesto: MESA’s public availability is intended to serve education, scientific research, and outreach.The passage presents public access as serving multiple educational and scientific purposes.
  • A. Manifesto: The appendix addresses MESA’s scientific motivation, philosophy and rules of use, stewardship, and future research and education tools.These topics define the framework for MESA’s continued use and development.

A.1. Motivation for a new tool

MESA is motivated by the central role of high-fidelity stellar evolution calculations across astrophysics and by the need for an open, modular, broadly applicable, modern, performant one-dimensional tool. Although some stellar phenomena require three-dimensional treatment, accurate one-dimensional calculations remain scientifically and pedagogically valuable.

  • Motivation: Stellar evolution calculations enable research spanning asteroseismology, nuclear astrophysics, chemical evolution, compact objects, supernovae, populations, hydrodynamics, and stellar activity.High-fidelity, modern calculations are described as critically important across these areas.
  • Design goals: The proposed tool should be open to researchers and support shared updating, fine-tuning, and further development.Openness is intended both to accelerate discovery and distribute development effort.
  • Design goals: It should provide independent reusable modules and flexible, up-to-date microphysics.Modularity and microphysics are separate stated requirements.
  • Design goals: The tool should cover stars across environments and evolutionary phases, including low- and massive stars, binaries, accreting and mass-losing stars, and advanced stages.Wide applicability would enable multi-problem, multi-object physics validation.
  • Design goals: It should use modern numerical methods, defined interfaces, and shared-memory parallelism while retaining scientific and pedagogical value through accurate one-dimensional calculations.The motivation acknowledges that convection, rotation, and magnetism are genuinely three-dimensional research-frontier problems.

A.2. MESA philosophy

MESA is an open community project that invites participation ranging from using releases to testing, debugging, and stewarding code modules. Users are encouraged to expand MESA while following conduct requirements that support attribution, reciprocity, reproducibility, and collaboration.

  • Participation: MESA welcomes researchers, students, and interested amateurs to participate through using releases, testing, debugging, submitting fixes, or stewarding code modules.Participation can range from using MESA for a science project to taking responsibility for continued stewardship of particular modules.
  • Community resource: Users are encouraged to add capabilities to MESA, which is intended to remain a community resource.The project frames user contributions as part of maintaining MESA for the broader community.
  • Code of conduct: Using MESA requires acknowledging the Paxton et al. (2010) publication and website, returning modifications, and notifying the MESA Council about publications.These requirements apply to research, educational, and outreach publications deriving from MESA use.
  • Openness and support: Users must share information needed to reproduce MESA results and help others learn the software as the project progresses.The conduct code describes this as “open know how” matching “open source” and asks users to identify themselves by name, email, and home location.

A.3. Establishment of the MESA council

The MESA Council was established to steward the project and support its shared missions in scientific research, education, outreach, and continued development. Its responsibilities include maintaining MESA infrastructure, engaging users and contributors, and enabling research and education.

  • Council establishment: The MESA Council was established by contributors working toward the project’s shared missions.These missions arose from the need to structure stewardship of MESA for scientific research, education, outreach, and development.
  • Steward MESA: The Council stewards MESA by supporting contributors, maintaining web resources, seeking funding, organizing yearly working groups, documenting development, and sustaining advanced development.
  • Interface with the User Community: The Council interfaces with users by answering questions, integrating new code, maintaining a user registry, and identifying engaged candidates for Council membership.
  • Enable Scientific Research and Education with MESA: The Council enables scientific research and education by promoting MESA, identifying matching science opportunities, facilitating collaborations, and tracking community science.

B. Code testing and verification

MESA uses regular, systematic testing to preserve expected behavior as the complex software evolves. Testing proceeds from module-level checks during compilation to a broad MESA star suite, which can be extended with new research-specific cases.

  • Testing rationale: Regular, systematic testing helps ensure MESA continues functioning as expected when new features are added.The motivation is to prevent unintended consequences for existing features in a large, complex software project.
  • Testing levels: Module-level tests run whenever MESA is compiled from the install scripts and check results against expectations.A second testing level is the MESA star suite, covering evolutionary cases across a broad range of applications.
  • Testing levels: The MESA star test suite covers Roche lobe overflow, low-mass-star He core flash, sub-stellar objects, advanced massive-star burning, and accreting white dwarfs.These evolutionary cases are intended to span a broad range of applications.
  • Test-suite extension: The test suite can be extended with research-specific cases, supported by a template for creating new tests.This allows regular testing of important aspects not covered by the existing set.
Loading 1009.1622v1…