The equations at the heart of the problem are balance laws: systems in which the time rate of change of conserved quantities, such as mass, momentum and energy, equals the divergence of a flux plus a source term. In astrophysics, the source term is often gravity, expressed through a gravitational potential. These equations possess non-trivial steady states in which the flux divergence exactly cancels the source, leaving nothing to evolve. Standard numerical schemes, however, do not satisfy a discrete version of this subtle equilibrium. They approximate the steady state with an error on the order of their truncation error, and that error piles up at every time step. In multi-dimensional simulations, the resolution needed to keep the pile-up from swamping the physics of interest can become prohibitively expensive.

The roots of the solution go back decades. Early work in the late 1970s and 1980s by Liu, van Leer and others suggested replacing the usual piecewise polynomial representation of the solution with a piecewise steady one that fulfils the equilibrium balance exactly or approximately. The term well-balanced itself was coined by Cargo and LeRoux in 1994, who built a scheme for the Euler equations with gravity that preserved a discrete form of hydrostatic equilibrium exactly, independent of resolution. Since then, an entire literature has flourished, particularly for the shallow water equations used in environmental applications, where the analogous property is known as the exact C-property.

To see why ordinary schemes fail, the review offers a deceptively simple example: the linear advection–reaction equation, which models radioactive material transported at constant speed while decaying at a fixed rate. Its steady states are exponential profiles in which advection and decay cancel exactly. When a standard first-order finite volume scheme—based on piecewise constant reconstruction, a Rusanov numerical flux and a midpoint-rule source discretization—is used to evolve such a steady state, spurious deviations appear immediately. The culprit is twofold: piecewise constant profiles cannot faithfully represent an exponentially varying equilibrium, introducing truncation errors of order of the cell size, and the flux divergence and source term are discretized independently, making an exact discrete balance unlikely.

The remedy is a piecewise steady reconstruction. Instead of fitting a polynomial within each cell, the scheme fits an equilibrium profile that satisfies the steady-state equation and matches the cell average. When this reconstruction is combined with an unsplit source discretization, the unperturbed steady state is preserved down to machine precision—roughly ten to the minus sixteen in double-precision arithmetic. Remarkably, the review shows that the same reconstruction does not help a fractional-step (split) scheme, because splitting the advection and reaction subproblems breaks the delicate coupling needed for the balance. The lesson generalizes: well-balancing requires the reconstruction and the source discretization to conspire, not merely to be individually accurate.

The general recipe for arbitrarily high-order well-balanced finite volume schemes follows from this insight. The solution is decomposed into an equilibrium part and a perturbation part, where the equilibrium part fulfils the steady-state balance by construction. Within each cell, the given cell average is first projected onto an average consistent with the steady states of interest, then a local equilibrium profile is matched to that projected average. The perturbation part is obtained by extrapolating the equilibrium profile to neighbouring cells, sensing how far they deviate, and feeding the resulting cell-averaged perturbations to any standard high-order reconstruction procedure. The source term is discretized by adding and subtracting the equilibrium contribution: the equilibrium part is integrated exactly via the fundamental theorem of calculus, while only the perturbation part requires numerical quadrature. At a steady state, the perturbation vanishes and the scheme reduces to an exact telescoping balance.

A crucial practical feature is the graceful fallback. If no local equilibrium profile can be found that matches a given cell average—because the solution may be arbitrarily far from equilibrium—the equilibrium reconstruction is simply set to zero, and the scheme automatically reduces to a standard high-order finite volume method without any loss of accuracy or robustness. This guarantee matters in real simulations, where shocks, contacts and turbulent regions coexist with near-equilibrium layers. The review also demonstrates the framework on the Euler equations in spherical symmetry, where a geometric source term becomes singular near the origin. A well-balanced treatment preserves a resting fluid with uniform density and pressure exactly, while a naive discretization produces spurious velocity fluctuations near the centre.

For the Euler equations with gravity, the review classifies well-balanced schemes into three broad families, reflecting progressively weaker assumptions about the equilibria they preserve. The first assumes the steady state is known a priori; representative schemes include the α-β method of Berberich and collaborators, which builds the reconstruction from known dimensionless density and pressure functions, and high-order discontinuous Galerkin schemes of Li, Xing and Wu that handle arbitrary known hydrostatic equilibria with positivity preservation. The second family exploits barotropic relations, in which density is a function of pressure alone; schemes by Käppeli and Mishra use thermodynamic potentials to reconstruct isothermal, isentropic and polytropic hydrostatic states exactly, for any equation of state. The third and most ambitious family builds equilibria from a consistent discretization of the underlying hydrostatic equation itself, integrating the equilibrium pressure numerically within each cell; these discretely well-balanced schemes of Käppeli and Mishra, later generalized to arbitrary order by Berberich and colleagues, avoid any thermal-stratification assumption but are hardest to extend to multiple dimensions, where the direct line-integration of the hydrostatic profile becomes path-dependent.

The numerical evidence is striking. On one-dimensional hydrostatic atmospheres—both isothermal and isentropic—well-balanced schemes maintain the discrete equilibrium to machine precision over a hundred sound crossing times, while standard schemes accumulate truncation errors that generate spurious flow. For tiny-amplitude waves, with perturbations as small as ten to the minus eight, the well-balanced scheme commits errors orders of magnitude smaller than the standard scheme. In three-dimensional simulations of a polytropic model star held together by self-gravity, both barotropic and discretely well-balanced second-order schemes preserve the hydrostatic configuration to machine precision over twenty sound crossing times, whereas the unbalanced scheme shows errors of several percent. For small pressure perturbations, the well-balanced schemes are roughly three orders of magnitude more accurate, and on the coarsest grid they outperform the unbalanced scheme on the finest grid tested. For large perturbations that steepen into shocks, all schemes perform comparably, confirming that the piecewise steady reconstruction does not diminish shock-capturing robustness.

The implications reach across the field. Stellar structure and evolution calculations, simulations of convection in stellar interiors, wave propagation in stratified atmospheres, and even general-relativistic hydrodynamics on curved spacetime all involve flows near equilibrium states that standard schemes struggle to resolve affordably. The review also situates well-balanced methods within the broader family of structure-preserving schemes—methods that maintain discrete forms of physical properties such as entropy admissibility, positivity of density and pressure, solenoidal magnetic fields, and total energy conservation in self-gravitating flows. Far from being a luxury, such structure preservation is often a necessity: near discontinuities, truncation errors are inevitably of order one, and the resolution needed to overcome them by brute force may be unaffordable. As computational astrophysics pushes toward multi-dimensional, long-duration simulations of stars and their violent deaths, well-balanced methods are emerging not as a niche trick but as an essential foundation for faithfully simulating the quiet before the storm.

Subject of Research: Well-balanced numerical methods for balance laws in computational astrophysics

Article Title: Well-balanced methods for computational astrophysics

Article References: Käppeli, R. (2022). Well-balanced methods for computational astrophysics. Living Reviews in Computational Astrophysics, 8(1), Article 2. https://doi.org/10.1007/s41115-022-00014-6

Image Credits: AI Generated

DOI: 10.1007/s41115-022-00014-6

Keywords: computational astrophysics, well-balanced schemes, hydrostatic equilibrium, finite volume methods, Euler equations, numerical methods, gravity, stellar simulations, shock capturing, structure-preserving methods, Well-balanced, methods

The research, carried out by Andrés Asensio Ramos of the Instituto de Astrofísica de Canarias and the Universidad de La Laguna in Tenerife, addresses a long-standing weakness in the traditional approach. In conventional MOMFBD, the algorithm performs what statisticians call a joint maximum likelihood estimation: it simultaneously estimates both the unknown object being imaged and the unknown atmospheric aberrations that distorted each frame. This joint strategy has powered pipelines at major facilities such as the Swedish 1-meter Solar Telescope and the GREGOR solar telescope for years, but it carries inherent instabilities that can silently degrade scientific results.

The heart of the problem lies in how the algorithm allocates power between signal and noise. When object and aberrations are estimated together, the optimizer can mistakenly attribute noise fluctuations in the data to high-order wavefront aberrations, fabricating atmospheric complexity that was never there. It is a bit like trying to transcribe a whispered conversation in a loud café while simultaneously modeling the background chatter: with no principled separation, the transcript can absorb café noise as words. For solar physics, where quantitative measurements of contrast and intensity feed directly into inferences about magnetic fields and plasma dynamics, such errors matter enormously.

Asensio Ramos’s solution reframes the estimation problem through a Bayesian lens. Rather than treating the observed object as an unknown to be pinned down exactly, the new approach marginalizes over it, integrating out the object’s uncertainty instead of estimating it jointly. This marginal estimator, built on ideas previously explored in adaptive optics for retinal imaging and astronomical adaptive-optics systems, produces what Bayesian practitioners describe as a form of automatic regularization. By acknowledging that the true object is not perfectly known, the method naturally prevents the reconstruction from funneling noise power into spurious high-order aberrations.

Technically, the procedure models the object with a Gaussian prior characterized by a power spectral density, or PSD, which encodes expectations about how image power is distributed across spatial frequencies. In the joint approach, the hyperparameters governing this prior act as delicate tuning knobs whose values strongly influence the final image contrast; set them slightly wrong and the restored Sun appears either unnaturally smooth or artificially grainy. The marginal formulation changes this sensitivity dramatically. Because the object uncertainty is folded into the likelihood analytically, the resulting estimator is far more forgiving, and those PSD hyperparameters can be optimized directly from the data rather than hand-tuned by an experienced pipeline operator.

This robustness yields what the author calls a plug-and-play deployment. Observatories currently depend on specialists who understand the subtle interplay between prior settings and image quality, and transferring pipelines between instruments or observing conditions often demands painstaking recalibration. With the marginal estimator, the need for that expert manual tuning largely disappears, lowering the barrier for smaller facilities and for scientists who want reproducible reconstructions without encyclopedic knowledge of the algorithm’s interior. In an era when next-generation instruments at the Daniel K. Inouye Solar Telescope are producing torrents of data, anything that reduces human intervention per dataset carries real practical weight.

Perhaps the most striking feature of the contribution is its mathematical minimalism. Implementing the marginal estimator requires adding only a log-determinant term to the traditional merit function that the optimizer minimizes. That single addition encapsulates the entire Bayesian marginalization, penalizing solutions in ways that account for the uncertainty of the object being reconstructed. Existing blind deconvolution pipelines can adopt the improvement with minimal modification, which dramatically shortens the path from theoretical insight to operational practice at working observatories.

To make adoption even easier, the method has been fully integrated into torchmfbd, an open-source package built on the PyTorch deep learning framework that Asensio Ramos and colleagues introduced in 2025 in Astronomy & Astrophysics. The package provides flexible multi-object, multi-frame blind deconvolution with automatic differentiation, allowing modern gradient-based optimizers such as Adam to drive the reconstruction. Scripts for reproducing the paper’s results are publicly available on GitHub, putting the technique within reach of the entire solar physics community, from large institutional pipelines to individual researchers analyzing archival data.

The significance extends beyond a single algorithmic tweak. High-resolution ground-based solar imaging underpins much of modern solar physics: measurements of penumbral fine structure, chromospheric magnetic concentrations, and the small-scale dynamo all depend on image restoration pipelines that faithfully preserve photometric contrast. Synthetic observations generated by advanced magnetohydrodynamic codes, such as MANCHA3D, are increasingly compared against restored observations to test our understanding of the solar atmosphere, and any systematic bias introduced during deconvolution propagates directly into those physical inferences. A marginal estimator that controls contrast more reliably therefore strengthens the entire chain from photons on a detector to statements about magnetic fields on the Sun.

The work also illustrates a broader trend in astronomy: the migration of rigorous Bayesian machinery into operational pipelines once dominated by ad hoc regularization. Marginal blind deconvolution has already proven valuable in adaptive-optics imaging of the retina and of night-sky targets, and this study extends the framework to the multi-object, multi-frame setting that solar observations uniquely demand, where wide fields of view must be patched together from many simultaneously recorded channels and frames. By combining accessibility, statistical soundness, and immediate software availability, the marginal approach may well become the default for the next generation of solar image restoration, ensuring that as telescopes grow larger and cameras faster, our pictures of the Sun stay as honest as they are spectacular.

Subject of Research: Bayesian marginal blind deconvolution for high-resolution ground-based solar imaging

Article Title: Marginal Multi-Object Multi-Frame Blind Deconvolution

Article References: Asensio Ramos, A. (2026). Marginal Multi-Object Multi-Frame Blind Deconvolution. Solar Physics, 301(10), Article 155. https://doi.org/10.1007/s11207-026-02752-9

Image Credits: AI Generated

DOI: 10.1007/s11207-026-02752-9

Keywords: solar physics, blind deconvolution, MOMFBD, Bayesian inference, image restoration, atmospheric turbulence, adaptive optics, PyTorch, torchmfbd, power spectral density, regularization, solar telescopes

The study, conducted by Francisco A. Guzmán, Elías Castellanos, and Jorge Mastache, tackles the Gross–Pitaevskii–Poisson (GPP) system, a pair of coupled equations that describes a Bose–Einstein condensate evolving under its own gravity. In this picture, the entire dark matter halo of a galaxy is represented by a single macroscopic wavefunction whose squared amplitude gives the mass density. The first equation is a nonlinear Schrödinger equation that includes the gravitational potential and a short-range repulsive interaction between the bosons, while the second is the Newtonian Poisson equation, which computes the gravitational field generated by the condensate’s own density. Solving these equations simultaneously is notoriously difficult because the problem is nonlinear and nonlocal: the gravitational potential depends on the wavefunction everywhere, not just at a single point.

What sets this analysis apart from much of the existing literature is a deliberate choice of bookkeeping. Many previous studies exploit mathematical scaling symmetries of the equations, solving everything in dimensionless units and only converting back to physical quantities at the end. The Mexican team instead kept three quantities explicit throughout: the boson mass, the scattering length that characterizes the strength of repulsive self-interaction between particles, and the total particle number N. That last parameter is crucial, because it directly fixes the total halo mass through the simple relation that the halo mass equals the boson mass multiplied by the particle count. By treating N as an independent control parameter, the researchers built a direct bridge between microscopic particle physics and macroscopic observables such as halo size, density profile, and rotation speed.

Technically, the team discretized the radial equations on a finite grid extending to 100 kiloparsecs and solved the resulting nonlinear eigenvalue problem with a Newton–Raphson iterative scheme, updating the wavefunction, the gravitational potential, and the chemical potential self-consistently until convergence to tolerances between 10 and 8 and 10 to the minus 10. To guard against artifacts of the starting guess, they initialized the solver with four different smooth, localized profiles: Gaussian, exponential, linear-exponential, and hyperbolic-secant shapes. A striking robustness emerged. Whenever different starting profiles converged to the same solution branch, the resulting stationary configurations were numerically indistinguishable, with relative differences below roughly one part in ten thousand for the wavefunction. The choice of initial guess affected only how often the iteration converged, not the physics of the answer.

Scanning a 40-by-40 grid of boson masses between 10 to the minus 23 and 10 to the minus 20 electronvolts and particle numbers between 10 to the 93 and 10 to the 99, the researchers classified every converged solution into three families. Ground states are nodeless, meaning the wavefunction never crosses zero, and they represent the lowest-energy equilibrium of the condensate. Excited states possess one or more radial nodes, producing oscillatory density shells around a central core, much like the excited orbitals of an atom. Unbound configurations fail a binding criterion, meaning the energy per particle exceeds the depth of the self-generated gravitational well, so no localized condensate can exist. The resulting phase diagrams show the ground-state branch occupying a well-defined diagonal band in the mass–particle-number plane: heavier bosons need fewer particles to bind, while lighter ones require vastly larger numbers.

One of the most consequential findings concerns how repulsive self-interactions reshape this map. As the scattering length increases, the ground-state branch systematically migrates toward larger boson masses and smaller particle numbers, because the repulsive pressure helps support the condensate against gravitational collapse. The excited-state and unbound regions, by contrast, shift far more slowly. This sensitivity matters for phenomenology: the location of the ground-state band, the region where realistic galactic halos live, is exquisitely dependent on the strength of the self-interaction, while the rest of the solution space is comparatively stubborn.

From the converged solutions, the team extracted empirical scaling laws for the characteristic halo radius, defined as the sphere enclosing 99 percent of the mass. In the weakly interacting regime, the fitted relation gives a radius proportional to the boson mass raised to the power of minus 2.996 and the particle number to the power of minus 0.998, values astonishingly close to the analytic prediction of minus 3 and minus 1 for non-interacting solitons. Rewritten in terms of halo mass, this recovers the famous mass–radius relation of Schrödinger–Poisson solitons, in which the radius scales inversely with the boson mass squared and the halo mass, providing a clean consistency check of the numerics. When self-interactions dominate, the scaling shifts toward the Thomas–Fermi expectation, with the radius growing as the square root of the scattering length and falling as the boson mass to the power of minus 1.545. Yet the fitted exponents reveal that the solutions occupy an intermediate regime where quantum pressure, self-interaction pressure, and gravity all contribute simultaneously, deviating measurably from the asymptotic Thomas–Fermi limit. A three-parameter fit confirms a clear hierarchy: the boson mass exerts the strongest influence on halo size, the particle number a moderate one, and the scattering length a softer but non-negligible effect.

The astrophysical payoff comes when these equilibrium configurations are confronted with real galaxies. Using ground-state solutions with a boson mass of 10 to the minus 22 electronvolts, the team modeled the rotation curves of two dwarf galaxies from the SPARC catalog, KK98-251 and UGC01281, adding the baryonic contribution from gas and stars to the condensate’s contribution computed from the enclosed mass profile. The results were impressive. For KK98-251, a halo of about 896 million solar masses with a characteristic radius of 5.05 kiloparsecs reproduced the observed kinematics, while UGC01281 required roughly 3.58 billion solar masses spread over 5.88 kiloparsecs. Remarkably, in both cases the solitonic core alone accounted for the dark matter, with no need for the extended interference-supported envelope that cosmological simulations of fuzzy dark matter typically predict around the central condensate.

Here, however, the story collides with cosmological constraints. Observations of the Lyman-alpha forest, the thicket of absorption features imprinted by intergalactic hydrogen on the light of distant quasars, place stringent lower bounds on the boson mass when ultralight bosons constitute all of the dark matter. Depending on the dataset and modeling assumptions, masses below roughly 10 to the minus 21 up to 2 times 10 to the minus 20 electronvolts are disfavored, excluding the canonical fuzzy dark matter scale of 10 to the minus 22 electronvolts. The new study shows that repulsive self-interactions can indeed rescue the equilibrium solutions: for scattering lengths around 10 to the minus 69 meters, ground-state branches extend into the mass range of roughly 2 times 10 to the minus 20 to 10 to the minus 18 electronvolts, comfortably compatible with the Lyman-alpha bounds. But there is a catch. Those high-mass, self-interacting configurations generate circular velocities far too small to match the dwarf-galaxy rotation curves that the lighter bosons fit so well. Within the class of stationary halos examined, the tension between the masses preferred by galactic dynamics and those demanded by the Lyman-alpha forest remains unresolved.

The broader significance of the work lies in its framework rather than any single verdict. By keeping the particle number explicit and mapping the full solution space, the authors have produced phase diagrams and scaling relations that connect fundamental boson properties directly to observable galactic quantities, offering a controlled laboratory for testing Bose–Einstein condensate dark matter models. The authors note that future work should probe the dynamical stability of the ground-state and excited branches, extend the analysis to halos with outer envelopes, and confront a much larger sample of rotation curves. For now, the message is twofold: quantum-wave halos remain a viable and elegant description of dwarf galaxies at the canonical mass scale, but escaping the tightening noose of Lyman-alpha constraints through self-interactions may cost the model the very galaxies it explains best. The universe, it seems, is still auditing every proposal for its missing matter.

Subject of Research: Equilibrium solutions of the Gross–Pitaevskii–Poisson system as a model for self-gravitating Bose–Einstein condensate dark matter halos

Article Title: Equilibrium halo solutions of the Gross–Pitaevskii–Poisson system: the role of the particle number

Article References: Guzmán, F. A., Castellanos, E., & Mastache, J. (2026). Equilibrium halo solutions of the Gross–Pitaevskii–Poisson system: the role of the particle number. The European Physical Journal C, 86(9), Article 1103. https://doi.org/10.1140/epjc/s10052-026-16373-w

Image Credits: AI Generated

DOI: 10.1140/epjc/s10052-026-16373-w

Keywords: dark matter, Bose–Einstein condensate, Gross–Pitaevskii–Poisson system, ultralight bosons, fuzzy dark matter, galaxy rotation curves, solitonic core, Thomas–Fermi limit, Lyman-alpha forest, self-interaction, halo scaling relations, computational astrophysics

The technical ambition of MEGATRON lies in its simultaneous treatment of processes that are usually modelled separately. The simulations track the hydrodynamic motion of gas, the propagation of radiation emitted by young stars, the non-equilibrium chemistry of the cooling and fragmenting gas, and the synthesis and dispersal of heavy elements by supernova explosions. All of these are evolved self-consistently at exceptionally high resolution, following a single young galaxy that will eventually grow into a system comparable in mass to the Milky Way. That choice matters: rather than simulating hundreds of crude galaxies, the team resolved the internal structure of one progenitor deeply enough to capture the small-scale filaments and dense clumps where the first stars actually formed, structures that simpler models smear out entirely.

The simulations begin from pristine gas containing no heavy elements at all, mirroring the composition of the cosmos shortly after the Big Bang, when hydrogen and helium were essentially the only ingredients available. From those initial conditions, the code follows the birth of the first stars, the ultraviolet radiation they pour into their surroundings, the supernovae that end their brief lives, and the dispersal of newly forged carbon, oxygen and iron into the gas that will form subsequent generations of stars. Each stage feeds back into the next: radiation heats and ionises gas and suppresses fragmentation, supernovae drive metal-rich winds out of nascent galaxies, and the changing chemical composition alters how efficiently the gas can cool and form new stars. It is this coupled chain of events that determines the chemical fingerprints later recorded in long-lived low-mass stars.

One of the collaboration’s headline results concerns the smallest dwarf galaxies, the systems least able to hold onto their gas and metals. The published work examines how the first stars can create an iron metallicity plateau in these tiny systems, a regime where the iron abundance of successive stellar generations stops rising and instead flattens out. Understanding such plateaus is crucial for stellar archaeology, because astronomers infer the properties of the first stars from statistical patterns in the abundance ratios of ancient stars. If simplified models mispredict how metals are mixed, expelled and diluted in dwarf galaxies, the inferred properties of the first stellar generations could be systematically wrong. By resolving the interplay of radiation, gas flows and chemical enrichment at high resolution, MEGATRON provides a physically grounded calibration for interpreting those abundance patterns.

The broader conclusion of the first wave of results is a cautionary one for the field: simplified subgrid models may substantially underestimate the influence of stellar radiation and complex chemical processes on the gas surrounding early galaxies. Radiation from massive stars does not merely illuminate the gas; it heats it, ionises it, drives photoevaporative flows and regulates where subsequent stars can form. Chemical enrichment, meanwhile, does not proceed as a smooth, well-mixed process but through inhomogeneous, clumpy channels that leave distinct signatures in different environments. When these effects are modelled at high resolution, structures emerge in the circumgalactic gas that coarser simulations simply do not capture, and predictions for what JWST should observe change accordingly. Improving those predictions is one of the project’s central practical goals.

Dr Martin Rey of the Department of Physics at the University of Bath, a lead contributor to the collaboration, framed the significance of the work in terms of connection. “The James Webb Space Telescope gives us a direct glimpse of the infant cosmos, while stellar archaeology allows us to study the relics of those earliest times in our own Galactic neighbourhood,” he said. “MEGATRON provides a physical bridge between the two.” The phrase captures the project’s core claim: the same physical processes that JWST observes in action in distant young galaxies are the ones that wrote the chemical record now readable in ancient stars much closer to home.

That bridge also speaks to one of astronomy’s most fundamental questions: where the elements that make up today’s Universe came from. Every atom of carbon in DNA, every atom of oxygen breathed on Earth and every particle of iron in the planet’s core was forged inside a star and scattered into space by stellar death. “The elements that make our world and life possible – carbon, oxygen, iron and many others – were forged by stars,” Dr Rey explained. “To understand where those elements came from, we need to understand how the first stars formed and enriched their surroundings. MEGATRON allows us to test these ideas directly by comparing detailed simulations with observations from JWST and the chemical fingerprints preserved in ancient stars.” In this sense, the first stars are not merely a remote curiosity of the early Universe; they are the opening chapter of the chemical history that made rocky planets and life possible.

The timing of the project could hardly be better. JWST is transforming the observational picture of the earliest galaxies, revealing objects that are brighter, more numerous and more chemically evolved than many models had anticipated. At the same time, large-scale stellar surveys are cataloguing hundreds of thousands of ancient stars in the Milky Way and its dwarf satellites, measuring their abundance patterns with ever greater precision. Each dataset is powerful on its own, but the combination is uniquely constraining. “MEGATRON provides a common physical framework for interpreting two of astronomy’s most exciting new datasets: JWST’s view of the earliest galaxies and the stellar fossil record,” said Dr Rey. “Together, these complementary observations allow us to test competing models of the first stars in ways that weren’t previously possible.” A model of the first stars must now survive two independent tests at once: it must reproduce the galaxies JWST sees, and it must reproduce the abundance patterns written into the oldest stars.

The computational demands of this programme are enormous. The project has been awarded 40 million processor hours on the United Kingdom’s national supercomputers, a resource allocation the team compares to running five million laptops in parallel for a full year. Those hours are already being spent developing the next generation of MEGATRON simulations, which will push to higher resolution and incorporate more complete physical models, from improved treatments of radiative transfer to richer chemical networks. The payoff, the researchers argue, will be increasingly direct comparisons between simulated and observed galaxies, and between simulated and measured chemical abundance patterns, tightening the link between theory and data at every stage of cosmic history.

MEGATRON began in 2023 and is scheduled to run until 2030, with the four papers just published representing only the first substantial body of results and further releases expected to follow. If the collaboration’s central thesis holds, the coming years should see the two great archives of the cosmic dawn – the photons collected by JWST from the infant Universe and the chemical fossils preserved in the oldest stars of the Milky Way – converge into a single, mutually consistent account of how the first stars formed, died and seeded the cosmos with the elements from which everything, including us, is ultimately built.

Subject of Research: Cosmological simulations of the first stars and galaxies and their chemical enrichment of the early Universe

Article Title: New simulations connect the first stars to cosmic fingerprints still visible today

Article References: New simulations connect the first stars to cosmic fingerprints still visible today. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: first stars, early galaxies, MEGATRON simulations, James Webb Space Telescope, stellar archaeology, chemical enrichment, cosmic dawn, dwarf galaxies, supernovae, metallicity, University of Bath, computational astrophysics

The stakes could hardly be higher. Gravitational wave detections of merging black holes and neutron stars have made it abundantly clear that some mechanism must shrink stellar orbits by orders of magnitude before the final, feeble gravitational-wave-driven inspiral can begin. For most systems, that mechanism is the common envelope. The concept dates back to 1976, when Bohdan Paczyński proposed it to explain V 471 Tauri, a compact white dwarf and K dwarf binary whose tight orbit could only be understood if the two stars had once shared an envelope. Since then, the common envelope has become the linchpin of compact binary formation theory, invoked to explain cataclysmic variables, binary pulsars, short gamma-ray burst progenitors, and the close white dwarf pairs that will one day merge as Type Ia supernovae. Practically all massive stars are born in multiple systems, and up to seventy percent of them experience binary interaction during their lives, making this phase not an exotic curiosity but a routine chapter in stellar history.

The challenge for theorists is that the common envelope phase is fast, asymmetric, and wildly multi-scale. The review organizes the interaction into three stages: a pre-envelope phase in which unstable mass transfer begins, a dynamical inspiral in which the companion plunges into the giant’s envelope, and a post-inspiral phase in which the system either ejects the envelope or settles toward merger. The dynamical inspiral, the heart of the problem, unfolds on timescales of roughly fifty days for the envelopes of red giant and asymptotic giant branch stars. Yet the cores of those same stars, with sound speeds exceeding a thousand kilometers per second, evolve on dynamical timescales of just twenty seconds. This five-order-of-magnitude temporal gap, combined with spatial scales spanning up to eight orders of magnitude when companions range from planets to black holes, means that fully resolved three-dimensional simulations remain out of reach for the foreseeable future.

For decades, the field relied on parametric shortcuts. The energy formalism, introduced by van den Heuvel and Webbink, balances the binding energy of the envelope against the orbital energy lost during inspiral, mediated by an efficiency parameter called alpha. Population synthesis models that predict gravitational wave event rates and Type Ia supernova progenitor channels depend critically on this parameter, yet its value remains deeply uncertain. Observational calibrations based on post-common-envelope white dwarf binaries have yielded conflicting results, with some studies favoring a constant alpha and others a value that varies with system parameters. A parallel angular momentum formalism, the so-called gamma prescription, suffers from its own lack of predictive power. Compounding the problem, the very definition of envelope binding energy is ambiguous: whether recombination energy released as ionized hydrogen and helium recombine can be tapped to eject the envelope is one of the most contested questions in the field, and the answer may push the effective efficiency above unity.

One-dimensional mechanical models attempted something more physical, integrating an equation of motion for the inspiralling companion under gravity and an assumed drag force while feeding the released energy into a stellar evolution code. The drag itself traces back to classical work on gravitational focusing: Hoyle and Lyttleton’s treatment of accretion onto a moving point mass, refined by Bondi and Hoyle into the famous accretion column picture, and complemented by Bondi’s spherical limit. Dynamical friction, first analyzed by Chandrasekhar for collisionless systems and extended to gaseous media by Dokuchaev and later Ostriker, produces the backward pull that shrinks the orbit. These analytic expressions, involving the Mach number of the companion’s motion and a Coulomb logarithm whose integration limits remain arbitrary, capture the functional dependencies but not the full nonlinear reality of a turbulent, shock-laden envelope.

Three-dimensional hydrodynamic simulations, begun tentatively in the late 1980s and maturing through waves of increasingly capable codes, have transformed the picture. They revealed just how violently non-spherical the interaction is: the envelope deforms into a toroidal shape, spiral shocks issue from the core binary, and shear instabilities shred the outflow. Crucially, they showed that ejecting the envelope using orbital energy alone is extraordinarily difficult. The inspiral tends to stall while a significant fraction of the envelope remains bound, a phenomenon attributed to several conspiring effects: the expansion and dilution of gas around the cores weakens the drag, the gas can be dragged into co-rotation with the binary until the velocity contrast vanishes, and the growing gravitational attraction between the approaching cores demands ever larger forces to sustain the inspiral. Simulations that include recombination energy through realistic equations of state eject substantially more material, though whether that energy thermalizes or radiates away remains contested, making radiation transport an urgent addition to the modeling toolkit.

The numerical techniques themselves form a fascinating landscape. Smoothed particle hydrodynamics, a Lagrangian particle method, excels at conserving angular momentum and avoiding advection errors, making it naturally suited to two orbiting stars embedded in vacuum, but it struggles to resolve shocks and low-density flows without enormous particle counts. Eulerian grid-based finite-volume schemes capture shocks and instabilities with superior accuracy through Riemann solvers, but suffer advection errors on fixed grids and require adaptive mesh refinement to concentrate resolution where it matters. Moving-mesh codes such as arepo, which evolved from cosmological simulation technology, thread the needle by advecting an unstructured Voronoi mesh with the flow, combining Lagrangian flexibility with Godunov-type accuracy. The first three-dimensional magnetohydrodynamic simulations of common envelope evolution, performed with arepo, showed that magnetic fields are amplified enormously by the magnetorotational instability in the accretion flow around the companion, though they remain dynamically subdominant during the inspiral itself.

Equally delicate is the art of setting up these simulations. The core of a giant star cannot be resolved without triggering fatal timestep restrictions, so modelers replace it with a gravitating point particle and reconstruct the envelope in hydrostatic equilibrium using a modified Lane-Emden equation. Gravitational softening must be applied to prevent the point masses from generating unphysical singular forces, yet the choice of softening length measurably alters the inspiral rate and envelope unbinding. Mapping one-dimensional stellar evolution models onto three-dimensional grids introduces noise and spurious velocities that must be damped away over several dynamical timescales before the companion is released. Even the pseudo-vacuum used to fill empty grid cells can contaminate predictions of observable light curves and colors. Energy and angular momentum conservation, monitored to the ten-percent level in early work, remains a critical diagnostic, because the envelope is so loosely bound that numerical errors can masquerade as physical ejection.

What emerges from the review is a field in rapid ascent. The past decade has seen a proliferation of global simulations, wind-tunnel experiments probing drag forces at high resolution, and growing recognition that the pre-envelope and post-inspiral phases, which unfold on thermal timescales of thousands of years, must eventually be coupled to the dynamical calculations. Observational anchors abound: post-common-envelope binaries provide statistical constraints, planetary nebulae bear the morphological fingerprints of ejected envelopes, and luminous red novae such as V1309 Sco, the merger that was caught in pre-outburst survey data, offer direct glimpses of the interaction in action. The ultimate prize is a predictive connection between the parameters of a binary entering the common envelope phase and the properties of the remnant that emerges, whether a tight double white dwarf, an X-ray binary, or a future gravitational wave source. Röpke and De Marco conclude that with rapidly advancing computational power, refined numerical techniques, and a clearer grasp of the relevant physics, one of the last fundamental unsolved problems of stellar astrophysics may finally yield.

Subject of Research: Three-dimensional hydrodynamic simulations of common-envelope evolution in binary stellar systems

Article Title: Simulations of common-envelope evolution in binary stellar systems: physical models and numerical techniques

Article References: Röpke, F. K., & De Marco, O. (2023). Simulations of common-envelope evolution in binary stellar systems: physical models and numerical techniques. Living Reviews in Computational Astrophysics, 9(1), Article 2. https://doi.org/10.1007/s41115-023-00017-x

Image Credits: AI Generated

DOI: 10.1007/s41115-023-00017-x

Keywords: common envelope evolution, binary stars, stellar astrophysics, hydrodynamic simulations, gravitational waves, compact objects, smoothed particle hydrodynamics, adaptive mesh refinement, recombination energy, stellar mergers, computational astrophysics, binary evolution

The central computational difficulty stems from a property astronomers call gravothermal evolution. Unlike ordinary gases, star clusters transport heat through the cumulative effect of countless small-angle gravitational deflections between distant stars—a process known as two-body relaxation. Following this effect directly in a simulation means tracking all pairwise force calculations, which scales roughly with the square of the particle number. Because the relaxation time in a cluster is vastly longer than the dynamical timescale, simulations must maintain extraordinary energy accuracy—errors below one part in one hundred thousand per crossing time—over thousands of crossing times. This combination of precision and duration has driven more than six decades of algorithmic innovation.

Before raw computing power caught up, astronomers relied on statistical approximations rooted in the Fokker–Planck equation, which describes how gravitational encounters diffuse stellar velocities in phase space. By truncating the so-called BBGKY hierarchy of kinetic equations at the two-body correlation level, theorists built gaseous and moment models that treated clusters like heat-conducting spheres. These models revealed the phenomenon of core collapse—the runaway contraction of a cluster’s center driven by its negative heat capacity—and later explained gravothermal oscillations. They also showed, remarkably, that rotation accelerates cluster evolution through an effective viscosity that transports angular momentum outward, a prediction made by anisotropic gaseous models of rotating clusters developed by Einsel and Spurzem and still debated against modern direct simulations today.

The most enduring statistical technique is the Monte Carlo method, descended from the schemes of Hénon and Spitzer. Modern implementations such as the MOCCA code and the Northwestern CMC code build quasi-realistic cluster models in which every star is represented by a particle orbiting in a smooth spherical potential, with random velocity kicks applied to mimic relaxation. Because these codes retain a three-dimensional data structure identical to that of N-body simulations, they can incorporate full single and binary stellar evolution, stellar collisions, and relativistic binary mergers. They have generated vast databases of simulated clusters for comparison with observations and are now central to predicting gravitational-wave source populations. Their limitation, the authors caution, appears when many massive objects crowd a cluster’s core or when the strictly spherical potential assumption breaks down in external tidal fields.

Direct N-body integration remains the gold standard against which all approximate methods are calibrated. The lineage begins with Sebastian von Hoerner’s earliest simulations, which were halted by the intractable problem of tight binaries, and Sverre Aarseth’s subsequent introduction of regularization—mathematically transforming the singular Newtonian two-body problem into a regular harmonic oscillator in four dimensions via the Kustaanheimo–Stiefel transformation. This breakthrough, refined through chain regularization for multiple close encounters, allowed simulations to proceed through the very events that had previously stopped them. The Hermite integration scheme, using accelerations and their time derivatives with hierarchically blocked individual time steps, became the backbone of the nbody6 code family, achieving fourth-order force accuracy and enormous speed gains for spatially structured systems.

Hardware evolution proved equally decisive. When Sugimoto and colleagues realized in 1990 that million-body simulations would take decades on conventional processors, the Tokyo group built GRAPE—special-purpose pipelined chips dedicated solely to gravitational force calculation. These were eventually supplanted by graphics processing units, with CUDA-accelerated kernels written by Keigo Nitadori still powering current codes. The flagship nbody6++gpu combines coarse-grained MPI parallelization across nodes with per-process GPU acceleration, and holds the record for the largest fully astrophysical direct simulation of a globular cluster—a million bodies evolved over twelve billion years with single and binary stellar evolution, stellar collisions, and tidal fields, performed by Long Wang and collaborators.

Relativistic physics has now entered the direct-integration domain. When black holes in simulations approach merger, Post-Newtonian correction terms—expansions in powers of velocity squared over the speed of light squared—must be added. Conservative terms produce periastron shifts, while the dissipative 2.5-order term drains orbital energy as gravitational radiation. Codes such as nbody7 incorporate the full Post-Newtonian hierarchy through algorithmic chain regularization, and model the asymmetric gravitational-wave recoil kicks that can eject merged black holes from their host clusters at hundreds or even thousands of kilometers per second—velocities that critically determine whether intermediate-mass black holes can accumulate in dense clusters.

Reliability remains a subtle question. Because star clusters are chaotically unstable, individual stellar trajectories diverge exponentially from their initial conditions within a fraction of a crossing time, and exact orbit reproduction would demand order-N decimal places of precision. Yet global quantities—Lagrangian radii, velocity dispersions, collapse times—are remarkably robust, as demonstrated by Giersz and Heggie’s pioneering ensemble comparisons and confirmed against Fokker–Planck theory. The practical lesson, the review stresses, is that N-body results should be interpreted statistically, with care when extracting rates of rare individual encounters from single realizations.

New architectures are pushing the frontier further. The hybrid code petar combines a particle-particle particle-tree scheme for long-range forces with a slow-down algorithmic regularizer for hard binaries, achieving excellent scaling on Germany’s Juwels Booster supercomputer for up to eight million particles with high binary fractions. The bifrost code, built on forward symplectic integrators with GPU acceleration, promises accuracy suited to the innermost nuclear star clusters where stars orbit supermassive black holes. Meanwhile stellar evolution prescriptions—natal kicks, remnant masses, black hole spins, and pair-instability supernovae—continue to be refined, with metallicity emerging as a decisive factor in producing the massive black hole seeds that gravitational-wave observatories are now detecting.

What emerges from this sweeping review is a field transformed. Statistical models, direct integration, special-purpose hardware, GPU acceleration, and relativistic extensions have converged to make dense star clusters computable in unprecedented fidelity. As instruments like the James Webb Space Telescope probe proto-globular clusters at redshift six and gravitational-wave detectors catalog mergers forged in cluster cores, the simulations described by Spurzem and Kamlah provide the theoretical bridge connecting the smallest scales of stellar encounters to the grandest structures in the Universe—ensuring that these fossil records of galaxy formation will keep yielding discoveries for decades to come.

Subject of Research: Computational methods for simulating dense, collisional star clusters, including direct N-body integration, Fokker–Planck and Monte Carlo models, stellar evolution, and relativistic dynamics.

Article Title: Computational methods for collisional stellar systems

Article References: Spurzem, R., & Kamlah, A. (2023). Computational methods for collisional stellar systems. Living Reviews in Computational Astrophysics, 9(1), Article 3. https://doi.org/10.1007/s41115-023-00018-w

Image Credits: AI Generated

DOI: 10.1007/s41115-023-00018-w

Keywords: star clusters, N-body simulation, computational astrophysics, globular clusters, two-body relaxation, core collapse, black holes, gravitational waves, Monte Carlo methods, stellar evolution, Fokker–Planck equation, GPU computing

The scale of the problem is difficult to overstate. AGB stars shine with luminosities of 100 to 1000 times that of the Sun and have effective temperatures below 3000 kelvin, giving them radii of several hundred solar radii. Red supergiants are even more extreme: hotter, between 3450 and 4100 kelvin, with radii that can exceed 1000 solar radii. Despite their vast size, their atmospheres are astonishingly thin and dynamic. Convection, the process that transports heat from the interior to the visible surface by bulk motion of hot gas, is not a local phenomenon in these stars. In the Sun, convection manifests as millions of small granules covering the photosphere. In giant envelopes, the pressure scale height is so large that individual convective plumes grow to sizes comparable to the stellar radius itself. Observations with optical interferometers have now directly imaged these giant cells on stars such as Betelgeuse, Antares and the AGB star π1 Gruis, revealing a handful of enormous bright and dark patches that evolve on weekly to yearly timescales.

Understanding such convection demands a fully three-dimensional, time-dependent treatment, because the process is non-local, non-linear, and spans wildly disparate length scales. Traditional one-dimensional stellar evolution models rely on mixing length theory, a prescription invented by Böhm-Vitense in 1958 that compresses all of convection into a single tunable parameter. For the Sun and solar-like stars, calibrated mixing lengths work reasonably well. But in the coolest and most luminous stars, where the convective plumes span a large fraction of the star, turbulence becomes supersonic, and radiative opacity varies by orders of magnitude, the one-dimensional approximation breaks down dramatically. The review emphasizes that only global simulations, often called star-in-a-box models, which encompass the majority of the convective envelope plus the near circumstellar environment, can capture the physics at play.

Two numerical codes anchor this effort. The first, CO5BOLD, developed over decades and first applied to global giant-star simulations by Freytag and collaborators in 2002, solves the coupled equations of compressible hydrodynamics and non-local radiative energy transport on a three-dimensional Cartesian grid embedded in a fixed, spherically symmetric gravitational field. It uses a finite-volume approach with directional splitting and an approximate Riemann solver of Roe type, along with a short-characteristics method for radiative transfer built on Rosseland mean opacity tables that merge high-temperature OPAL data with low-temperature PHOENIX data. The code can be extended to magnetohydrodynamics and can handle rotation through a corotating frame. The second code, Athena++, developed by Stone and colleagues, takes a different route: it employs adaptive mesh refinement, task-based dynamic execution, and a Godunov-type shock-capturing scheme, solving the time-dependent, frequency-integrated radiative transfer equation directly over discrete ordinates, with radiation fully coupled to the gas energy and momentum equations. For red supergiants, Athena++ simulations use spherical polar coordinates covering 70.6 percent of the face-on hemisphere, with an accurate gravitational potential that includes the mass of the simulated envelope itself, a detail that matters because these loosely bound atmospheres are anything but negligible in mass.

What do these simulations actually show? The picture that emerges is spectacularly turbulent. The surfaces of simulated AGB and red supergiant stars are dominated by a few giant convective cells, with wide upflows surrounded by narrow, turbulent downdrafts, evolving on timescales from weeks to years. Non-stationary convection in the deep interior generates sound waves. These waves travel outward, slow down and compress as the temperature and sound speed drop, and their amplitudes grow with the steep decline in density until they steepen into shocks. Just above the visible surface, the resulting shock network is extraordinary: turbulent pressure can exceed twenty times the gas pressure, shock waves propagate with average Mach numbers of 3 to 4, and local peaks reach Mach numbers of 8 to 10. Individual shocks above the photosphere can span 200 to 250 solar radii, a width comparable to the star’s own radius, at temperatures around 2500 kelvin. These shocks intermittently levitate gas to distances where dust can condense, providing the crucial first step in the wind-acceleration mechanism that drives the massive mass-loss rates of 10 to the minus 6 to 10 to the minus 4 solar masses per year observed in AGB stars.

Remarkably, the simulations show that radial pulsations are not imposed on these models but emerge spontaneously from the convective motions themselves, a result first demonstrated for AGB stars by Freytag and colleagues in 2017 and confirmed in later work by Ahmad and coauthors in 2023. The extracted pulsation periods agree with those observed in long-period variable stars. This interplay between convection and pulsation leaves an observable fingerprint in stellar spectra. Spectral lines form over a range of atmospheric depths, and each line carries the imprint of the local velocity field through its shift, width and asymmetry. The tomographic method, originally conceived to study shock waves in pulsating Cepheids and Mira variables, exploits this by sorting spectral lines according to their formation depth and cross-correlating them with observed spectra. In Mira stars, the technique beautifully reproduces the classic Schwarzschild scenario, in which a shock front marching upward through the atmosphere produces a sequence from red-shifted absorption to line doubling and finally blue-shifted emission.

Red supergiants, it turns out, play by different rules. Applied to years of high-resolution spectra of the prototypical red supergiants μ Cephei and Betelgeuse, the tomographic method revealed no such clean shock scenario. Instead, the atmospheric layers exhibit a phase shift between velocity and temperature variations, tracing a hysteresis loop in the temperature-velocity plane. Crucially, synthetic spectra computed from three-dimensional CO5BOLD simulations show the same hysteresis behavior with timescales matching the observed photometric variability. The interpretation is elegant: the loop is the signature of convective turnover, in which hot rising material reaches the upper atmosphere, cools, and falls back, modulated by acoustic waves generated by the turbulent convective flow below. The method even provided insight into Betelgeuse’s famous Great Dimming of early 2020, when the star faded to magnitude 1.6. Tomography identified two successive shocks, in February 2018 and January 2019, that expanded part of the atmosphere and, through increased molecular opacity in the cooler gas, obscured the star, an explanation consistent with, and complementary to, the dust-veil picture from direct imaging.

The Athena++ simulations have yielded equally consequential results on the theoretical side. Goldberg and colleagues in 2022 performed the first three-dimensional calibration of mixing length theory in the red supergiant regime, recovering surprisingly high convective efficiencies with a mixing length parameter of roughly 3 to 4, and showing that radiation pressure provides about one-third of the support against gravity in the envelope. The simulations also revealed an inverse correlation between density and radial velocity in the outer layers, uncharacteristic of ordinary convection, and an atmosphere far more extended than one-dimensional models predict. In these luminous layers, where the opacity locally exceeds the Eddington opacity, the usual correlation between dense sinking material and convection flips entirely: cold, opaque, dense regions move outward. This has direct implications for how red supergiants die. When such a star explodes as a Type II-P supernova, its shock must traverse this wildly inhomogeneous envelope, whose surface undulates over radial ranges of 10 to 20 percent of the stellar radius. The shock therefore breaks out at different radii and different times around the star, smearing the ultraviolet flash of shock breakout over roughly 3 to 6 hours instead of the sub-hour duration predicted by spherical models, and reducing peak luminosities by factors of 2 to 10. This means astronomers cannot use breakout rise times to measure progenitor radii via light-travel-time effects, but they can instead use breakout durations to probe the scale of convective inhomogeneities themselves.

The convergence of three-dimensional radiation-hydrodynamics modeling, interferometric imaging, ALMA observations of molecular envelopes, spectropolarimetric monitoring of surface magnetism, and tomographic spectroscopy is transforming our understanding of the late lives of stars. Convection, once a fudge factor in one-dimensional codes, is now a quantitatively constrained, observable, three-dimensional process that shapes winds, dust formation, pulsation, magnetism and the very first light of supernovae. As future ultraviolet missions such as ULTRASAT catch hundreds of shock breakouts at high cadence, and as ever more powerful simulations push toward even greater realism, the giant convective cells churning on the surfaces of dying stars will continue to yield their secrets, one shock wave, one spectral line, and one hysteresis loop at a time.

Subject of Research: Three-dimensional radiation-hydrodynamics simulations of convection in the atmospheres of asymptotic giant branch and red supergiant stars

Article Title: Signatures of convection in the atmospheres of cool evolved stars

Article References: Chiavassa, A., Kravchenko, K., & Goldberg, J. A. (2024). Signatures of convection in the atmospheres of cool evolved stars. Living Reviews in Computational Astrophysics, 10(1), Article 2. https://doi.org/10.1007/s41115-024-00020-w

Image Credits: AI Generated

DOI: 10.1007/s41115-024-00020-w

Keywords: stellar convection, red supergiants, AGB stars, radiation hydrodynamics, CO5BOLD, Athena++, tomography, Betelgeuse, shock breakout, supernovae, mixing length theory, stellar mass loss

The historical arc of the field is one of steady, incremental extension into adjacent problems. The first synthetic supernova spectrum was presented by David Branch in 1980, who adapted the Schuster-Schwarzschild approach from stellar atmospheres and introduced the Sobolev formalism for line transfer. His tools, which eventually became the widely used SYNOW code, allowed detailed comparisons between white dwarf explosion models and observations of Type I supernovae. In the same year, Timothy Axelrod’s doctoral thesis laid the foundation for nebular-phase modeling, establishing the treatment of non-thermal and non-local thermodynamic equilibrium (NLTE) physics that still sets the standard for many codes today. The explosion of SN 1987A then drew new workers into the field, including Claes Fransson in Stockholm and Leon Lucy in London, whose Monte Carlo techniques would prove transformative.

Around 2005, a wave of development swept through the field, much of it inspired by Lucy’s work on Monte Carlo methods. Three-dimensional codes operating under the assumption of local thermodynamic equilibrium (LTE) appeared, including SEDONA and ARTIS, while one-dimensional NLTE codes such as SUMO, NERO, and CMFGEN were developed for higher-fidelity spectral calculations. The first public Schuster-Schwarzschild code, TARDIS, was released in 2014 and remains popular for rapid line identification. When the first kilonova, AT2017gfo, was discovered in 2017 following the gravitational-wave detection GW170817, the supernova toolkit was ready to be adapted. Metzger and colleagues extended SEDONA to kilonovae as early as 2010, and codes such as SuperNu, POSSIS, and a kilonova version of SUMO followed, opening the door to NLTE modeling of neutron star merger ejecta.

The review emphasizes that astrophysics grows by small extension steps into related areas, a process that takes decades. Sometimes growth in computing power drives these steps, but more often they are application-driven. This has a practical consequence: methods derived under computing constraints orders of magnitude stricter than today’s may carry simplifications that are no longer necessary. Studying the history of a code or methodology therefore reveals where modern computing power can be exploited to improve upon legacy approximations, a point the author urges researchers to keep in mind whenever they adopt an existing model.

Why invest such effort in spectral synthesis? The review identifies three main science drivers. First, identifying which elements produce which line features and determining the elemental abundances in the ejecta. For decades this seemed almost insurmountable, because supernovae blend lines through rapid Doppler expansion, cascade energy over six orders of magnitude, and involve non-thermal effects, asymmetries, molecules, and dust. Today, clear diagnostic methods exist for seventeen elements from hydrogen to nickel, and direct full-ejecta spectral tests of explosion models have been performed for all major supernova classes. Second, understanding the origin of the elements across the periodic table: with AT2017gfo, the upper two-thirds of the table opened for direct analysis, with diagnostic potential already established for elements including strontium, yttrium, tellurium, lanthanum, cerium, neodymium, and tungsten. Third, determining the progenitor systems and explosion mechanisms, since spectra diagnose both the hydrostatic life of the progenitor star and the physics of its death, from neutrino-driven core collapse to accretion disk outflows near black holes.

The physical situation these codes must capture is extreme. The ejecta expand homologously, with faster fragments outrunning slower ones, at characteristic velocities of roughly 5,000 kilometers per second for core-collapse supernovae, 8,000 for thermonuclear supernovae, and 50,000 for kilonovae. Kilonovae carry about one hundred times less mass than supernovae but comparable kinetic energy, hence their higher speeds and their rise and decline over just days. Both classes are powered by radioactivity: supernovae chiefly by the decay of nickel-56 and cobalt-56, kilonovae by a vast array of radioactive r-process nuclides whose energy is carried mainly by leptons rather than gamma rays. Because the ejecta are semi-transparent, with line opacity producing lingering radiative transfer effects for years or even decades, much of the optical and infrared spectra of both supernovae and kilonovae is formed by fluorescence.

Temperature emerges as the most fundamental quantity in the modeling. It governs ionization and excitation, sets the regime of spectral formation, and strongly influences opacity. The governing law is energy conservation, expressed either through the internal energy of the gas or, equivalently in NLTE treatments, through the evolution of thermal kinetic energy balanced by heating and cooling. Codes differ in whether they retain the time-derivative terms: STELLA, SuperNu, CMFGEN, and, since 2022, SUMO can solve the time-dependent energy equation with implicit or semi-implicit finite-difference schemes, typically using time steps of about ten percent, which is well matched to both the homologous expansion timescale and radioactive decay timescales. The review shows that this choice of stepping has a solid physical anchoring for both supernova and kilonova applications across all conceivable epochs.

The treatment of line cooling exposes a subtle but consequential divide between LTE and NLTE approaches. In LTE codes, a parameterized thermalization probability is often assigned to line absorptions, and surprisingly, realistic light curves and spectra require this probability to be near unity, effectively mimicking the fluorescence process that the simplified treatment omits. But when radioactivity dominates the heating, such an inflated absorption coefficient can drive temperature estimates a factor of ten too low, likely explaining why LTE codes yield significantly lower temperatures than NLTE ones from quite early phases. In NLTE, cooling is computed from net electron collision rates, which represent the small difference between two large terms and demand careful formulation. A code comparison for a simple Type Ia test model shows that even state-of-the-art tools produce temperature profiles that vary considerably, underscoring how much work remains to achieve robust temperature determinations.

Solving the NLTE rate equations presents its own numerical challenges. With potentially more than one hundred thousand levels per grid cell, the full system is split into blocks, typically one per ion, and solved iteratively. Matrix storage scales with the square of the system size and inversion time with its cube, so the standard LAPACK routine DGESV, based on LU decomposition, remains efficient only up to a few hundred levels. Superlevels, which bundle groups of levels assuming LTE distributions within them, offer dramatic savings: in one full-composition Type Ia model, 10,605 levels were grouped into 2,338 superlevels, saving a factor of sixteen in storage and up to sixty-four in matrix inversion time. Negative populations, false singularities, and oscillating Newton-Raphson steps all require layered fall-back strategies, and co-solving all ions of an element, as done in recent 3D codes, reduces such pathologies.

Radioactive powering completes the triad of central modeling blocks. High-energy decay particles transfer their energy through ionization, excitation, and heating, with the Bethe stopping-power formula, derived in the early 1930s, describing the continuous energy loss of fast particles. Gamma rays Compton-scatter to create primary electrons of 0.1 to 1 megaelectronvolt, which cascade into secondaries; the distribution of these secondaries, measured experimentally by Opal and colleagues, feeds into steady-state Boltzmann solutions of the degradation spectrum. For kilonovae, time-dependent thermalization effects become important within weeks, and modeling the powering has already allowed constraints to be placed on the composition of AT2017gfo. Looking forward, the review charts a roadmap: better physical treatments, accurate atomic data for trans-iron elements, 3D hydrodynamic explosion models, and high-quality observations from the ultraviolet to the mid-infrared. Eighteen of the thirty lightest elements now have good or moderate diagnostic potential, and with kilonova spectroscopy advancing rapidly, the goal of determining the origin of the elements directly at their production sites is coming within reach.

Subject of Research: Computational spectral synthesis modeling of supernova and kilonova spectra to infer ejecta composition and explosion physics

Article Title: Spectral synthesis techniques for supernovae and kilonovae

Article References: Jerkstrand, A. (2025). Spectral synthesis techniques for supernovae and kilonovae. Living Reviews in Computational Astrophysics, 11(1), Article 1. https://doi.org/10.1007/s41115-025-00022-2

Image Credits: AI Generated

DOI: 10.1007/s41115-025-00022-2

Keywords: supernovae, kilonovae, spectral synthesis, radiative transfer, NLTE modeling, r-process nucleosynthesis, Monte Carlo methods, neutron star mergers, radioactive powering, astrophysical transients, computational astrophysics, origin of the elements

The core obstacle is captured by a single number, the Reynolds number, which measures the ratio of non-linear inertial forces to viscous damping in a flow. When this number exceeds roughly a few thousand, flows become turbulent. Astrophysical flows routinely reach values that are astronomically larger; the turbulent convection zone of the Sun, for instance, is estimated to have a Reynolds number of order ten to the fourteenth. Direct numerical simulations that resolve every scale down to the microscopic dissipation length would require computing resources that scale with the Reynolds number cubed. For solar convection alone, the review notes, a single dynamical time would demand roughly ten to the forty-second floating point operations, far beyond even exascale machines.

Large eddy simulations sidestep this impossibility through a mathematical procedure called spatial filtering. A low-pass filter, defined by a convolution with a kernel such as a box, Gaussian, or sharp cut-off filter, separates the flow into smoothed, large-scale variables and fluctuating components on scales smaller than the filter length, which is usually identified with the grid spacing. Because the equations of fluid dynamics are non-linear, filtering them generates new terms: a subgrid-scale turbulence stress tensor that describes the momentum exchange between resolved and unresolved eddies, a turbulent pressure proportional to the subgrid-scale kinetic energy, and additional fluxes for energy and chemical species. The trace of the stress tensor defines the subgrid-scale turbulence energy, the kinetic energy hidden in eddies too small for the grid to see. Modelling these terms is known as the closure problem, since the filtered equations form an infinite hierarchy of moments that must be truncated.

Several families of closures dominate astrophysical practice. The simplest is the Smagorinsky model, which assumes a local balance between turbulence production and dissipation and yields an eddy viscosity proportional to the square of the grid scale times the local rate of strain. More sophisticated is the one-equation model, which solves a transport equation for the subgrid-scale turbulence energy itself, including production by shear, dissipation, turbulent diffusion, and pressure-dilatation effects. A third approach, structural modelling, reconstructs the unresolved stresses directly from gradients of the resolved velocity and magnetic fields, without tuneable coefficients or extra equations. For magnetohydrodynamics, the structural model also provides a closure for the subgrid-scale electromotive force, the term that governs how unresolved turbulent motions amplify magnetic fields through dynamo action.

A crucial subtlety is that most astrophysical simulation codes already dissipate energy through their numerical schemes. Finite-volume and finite-difference truncation errors behave like diffusion terms, producing an effective numerical viscosity that mimics turbulent viscosity on the grid scale. This observation underlies implicit large eddy simulation, or ILES, in which no explicit model is used and the numerics themselves play the role of the closure. The review explains why ILES often suffices: because inertial-range scaling is largely independent of the dissipation mechanism, statistics such as energy spectra can be reproduced as long as the dynamical range is adequate. However, the so-called bottleneck effect distorts spectra near the grid scale, and explicit subgrid-scale models add only a marginal effect for strongly diffusive second-order schemes. The picture changes for higher-order methods and mesh-free Lagrangian codes, which lack intrinsic diffusion and benefit far more from explicit modelling.

Validating these models without experiments or direct numerical simulations is a distinctive challenge of astrophysics. The standard tools are a priori tests, in which turbulence data from simulations are explicitly filtered and the correlations between modelled and true subgrid-scale terms are measured, and a posteriori comparisons of turbulence statistics between LES and ILES. Such analyses have calibrated the coefficients of the eddy-viscosity and gradient-diffusion closures, revealing, for example, that the turbulent Prandtl number is around ten rather than unity, and that purely linear eddy-diffusivity closures fail badly for magnetic stresses, matching the sign of the energy cascade only about half the time. Non-linear structural closures, by contrast, achieve correlations close to unity for the kinetic and magnetic stress tensors and the electromotive force, especially when a compressibility correction is included.

The payoff of this machinery appears across a remarkable range of cosmic phenomena. In thermonuclear supernovae, where a white dwarf detonates in a deflagration that no grid can resolve, the subgrid-scale turbulence energy sets the effective flame propagation speed, replacing the microscopic laminar burning velocity with a turbulent one. In galaxy simulations, the subgrid-scale turbulence energy can be fed by supernova feedback as an additional source term, and it allows a locally variable, turbulence-regulated star formation efficiency to be computed from the turbulent Mach number and virial parameter, reproducing observed star formation laws without assuming them. In cosmological simulations, filtered equations formulated in co-moving coordinates show that unresolved turbulent pressure can contribute a non-negligible fraction of the support against gravity in galaxy clusters, while shear-improved models cleanly separate genuine turbulence from gravity-driven bulk flows.

Perhaps the most striking recent success concerns magnetic field amplification in binary neutron star mergers. Kelvin-Helmholtz instabilities in the shear layer between the merging cores trigger a small-scale dynamo, and the growth rate of the magnetic field normally increases with resolution because the smallest resolved eddies dominate the amplification. By combining a relativistic generalization of the structural subgrid-scale model with high-order schemes, researchers achieved, for the first time, a numerically converged magnetic field amplification in merger simulations: the large eddy simulation reached the saturated field strength at half the resolution required without the model, a substantial saving in computational cost. The resulting magnetic energy spectrum even displays the expected Kazantsev scaling, and the post-merger field appears universal, insensitive to the initial magnetic configuration of the two stars.

Subgrid-scale modelling also matters for the chemical evolution of galaxies. In mesh-free codes without numerical diffusion, metals expelled by supernovae would never mix with their surroundings without an explicit turbulent diffusivity, and studies show that such mixing significantly reshapes the metallicity distributions of the circumgalactic and warm-hot intergalactic media. Looking forward, the review argues that the main utility of subgrid-scale models lies precisely in this treatment of complex sub-resolution physics, from turbulent burning and star formation to dynamo action and metal transport. While statistical properties of well-resolved turbulence remain largely insensitive to the choice of model, the coupling between resolved and unresolved scales is real and physically motivated closures capture it better than numerical truncation errors alone. As exascale computing opens new regimes, large eddy simulations are poised to remain an indispensable bridge between the eddies we can compute and the turbulent universe we observe.

Subject of Research: The methodology of large eddy simulations and subgrid-scale modelling for turbulent astrophysical flows

Article Title: Large eddy simulations in astrophysics

Article References: Schmidt-Brückner, W. (2025). Large eddy simulations in astrophysics. Living Reviews in Computational Astrophysics, 11(1), Article 2. https://doi.org/10.1007/s41115-025-00023-1

Image Credits: AI Generated

DOI: 10.1007/s41115-025-00023-1

Keywords: large eddy simulation, subgrid-scale modelling, astrophysical turbulence, magnetohydrodynamics, computational astrophysics, turbulent dynamo, type Ia supernovae, star formation, galaxy clusters, neutron star mergers, turbulent mixing, cosmological simulations

The governing equations of fuzzy dark matter are deceptively compact. The model is described by the Schrödinger–Poisson system, in which a complex wave function evolves under the influence of its own self-gravity. Despite their quantum pedigree, these equations admit a fluid interpretation through the Madelung transformation, which recasts the wave function in terms of a density and a velocity field. In this fluid picture, the familiar equations of hydrodynamics acquire an additional term, the quantum pressure, which counteracts gravity on small scales and introduces a characteristic Jeans scale. Structures larger than this scale grow much as they would in the standard cold dark matter scenario, while smaller structures are suppressed, because the uncertainty principle prevents the bosons from being localized into arbitrarily tight clumps. This suppression of small-scale structure is one of the model’s most distinctive and testable predictions, and it manifests as a sharp cutoff in the linear matter power spectrum at a wavenumber that scales with the square root of the boson mass.

Simulating this quantum fluid is dramatically harder than simulating cold dark matter. Wave-based schemes, which evolve the wave function directly, must resolve not only the de Broglie wavelength but also its rapid oscillations in time. The time-step constraints scale with the square of the spatial resolution, a diffusion-like requirement that makes high-resolution simulations extraordinarily expensive. The review details how the computational cost of resolving density granulation within a halo scales as the fourth power of the boson mass parameter and as the 8/3 power of halo mass, meaning that simulations of massive halos with heavier bosons quickly become infeasible. This is the central reason why most genuine wave-based cosmological simulations to date have been restricted to small boxes of a few megaparsecs, or to relatively light boson masses, while heavier candidates and larger volumes remain out of reach.

Among wave-based methods, the pseudo-spectral approach reigns supreme in accuracy. The global Fourier method applies a fast Fourier transform to the wave function, multiplies by the kinetic operator in momentum space, and transforms back, achieving spectral convergence with errors that decay faster than any algebraic rate as resolution increases. The method is unconditionally stable and conserves mass to machine precision, making it the algorithm of choice for codes such as GAMER, AREPO, PyUltraLight and SPoS. Its Achilles heel is that the discrete Fourier transform assumes a uniform grid with periodic boundaries, which is incompatible with adaptive mesh refinement, the technique that makes cosmological simulations affordable by concentrating resolution only where it is needed. This limitation has spurred the development of alternatives, including finite-difference schemes that work on refined grids but converge far more slowly, and a newer local pseudo-spectral technique based on Fourier continuations with Gram polynomials, which achieves eleventh-order accuracy on non-periodic grids and has been implemented in the GAMER code.

Fluid-based methods offer a compelling escape from the resolution bottleneck. Because the density, velocity and phase fields are smooth in the vast low-density regions outside halos, fluid schemes do not need to resolve the de Broglie wavelength there, allowing much larger simulation volumes. Smoothed particle hydrodynamics, implemented in codes such as AX-GADGET, and meshless finite-volume methods in the GIZMO code both exploit this advantage, and the finite-volume formulation conserves mass and momentum to machine precision. But fluid approaches have a fundamental weakness: at density nodes created by destructive wave interference, the velocity and quantum potential diverge, and the phase field becomes discontinuous. These singular points, which correspond to quantized vortices threading the halos, are notoriously difficult for any fluid scheme to capture, and the review documents persistent discrepancies between SPH simulations and wave-based calculations, including disagreements over whether soliton cores and the soliton–halo relation are recovered correctly.

The most promising path forward may be hybrid schemes that combine the strengths of both formulations. In the approach developed by Schwabe and Niemeyer and implemented in the AxioNyx code, collisionless N-body particles carry wave packets on coarse levels, and the full wave function is reconstructed on fine levels using a Gaussian beam method that captures statistically correct interference patterns. A newer fully grid-based hybrid scheme in GAMER, described by Kunkel and collaborators, solves the Hamilton–Jacobi–Madelung fluid equations on coarse levels and switches to the wave formulation on refined levels, with careful phase unwrapping at the fluid–wave interfaces. These hybrid methods, coupled with adaptive mesh refinement, have enabled cosmological zoom-in simulations of individual halos down to the present day, something that pure wave-based schemes could not previously achieve, and they open the door to simulating more massive halos and heavier boson masses than ever before.

The distinctive phenomenology of fuzzy dark matter gives simulations their scientific payoff. Each virialized halo hosts a dense, stable soliton core, the ground-state solution of the halo potential, surrounded by an outer profile that closely resembles the standard Navarro–Frenk–White form. The soliton density profile is redshift-independent and obeys an elegant scaling symmetry, and its properties are linked to the host halo through a debated soliton–halo relation that recent work interprets as thermal equilibrium between the core and the surrounding granules. Beyond the soliton, the halo is permeated by stochastically fluctuating density granules, roughly a quarter of the local de Broglie wavelength in size, generated by constructive and destructive interference. These granules scatter stars and gas, dynamically heating dwarf galaxies, stellar streams and galactic disks, and their effects are now being used to place some of the tightest constraints on the boson mass, with recent analyses of ultrafaint dwarfs suggesting the mass must exceed roughly 2.2 x 10^-21 electronvolts.

Simulations also reveal subtler pitfalls that can silently corrupt results. Spurious halos, seeded by numerical noise along filaments, can form below the half-mode mass in both N-body and genuine fuzzy dark matter simulations, mimicking real objects and potentially biasing constraints derived from the halo mass function. The review shows that the global Fourier method appears largely immune to this artificial fragmentation, while hybrid and N-body schemes produce suspiciously regular clumps along filaments that lack counterparts across different numerical setups. Equally treacherous is insufficient resolution of the de Broglie wavelength in low-density regions, which underestimates flow velocities and delays structure formation, and inadequate resolution within halos, which distorts the quantum pressure and can drive unphysical halo contraction. The review demonstrates that even when a simulated halo’s mass and central soliton profile look correct, the outer profile can still be wrong, a sobering warning that agreement with theory is necessary but not sufficient evidence of numerical accuracy.

To help the community navigate these hazards, the review provides a test bench of standard problems, from Gaussian wave packets and vortex pairs to the Jeans instability, isolated solitons, isolated halos and full cosmological boxes, and it makes publicly available initial condition files for isolated-halo and cosmological simulations so that different codes can be compared on equal footing. Eigenmode methods, which construct halos as superpositions of bound states of a static potential, offer yet another efficient tool for studying interference phenomena without solving the full nonlinear system, while collisionless N-body simulations with fuzzy initial conditions remain valuable for probing large-scale structure and checking convergence. Looking ahead, the review notes that machine learning approaches, such as physics-informed neural networks, may eventually accelerate these calculations, and that the algorithms can be extended to self-interacting, vector and multi-component variants of the model. What remains clear is that simulating fuzzy dark matter has matured into a rich computational discipline in its own right, one whose progress will determine whether this quantum alternative to cold dark matter survives the confrontation with the universe it seeks to explain.

Subject of Research: Numerical simulation methods for fuzzy dark matter, an ultralight bosonic dark matter model exhibiting wave phenomena on galactic scales

Article Title: Fuzzy dark matter simulations

Article References: Schive, H.-Y. (2026). Fuzzy dark matter simulations. Living Reviews in Computational Astrophysics, 12(1), Article 1. https://doi.org/10.1007/s41115-026-00027-5

Image Credits: AI Generated

DOI: 10.1007/s41115-026-00027-5

Keywords: fuzzy dark matter, ultralight bosons, Schrödinger–Poisson equations, soliton cores, density granulation, adaptive mesh refinement, pseudo-spectral methods, quantum pressure, cosmological simulations, halo mass function, vortices, computational astrophysics