arXiv subjects

Luke F. Roberts

Publications and source records attributed to Luke F. Roberts.

At least 19 recordsLinked to original sources

arXiv2026

Multigroup Thermal Radiation Transport with Tensor Trains

We investigate the application of tensor-train (TT) algorithms to multigroup thermal radiation transport (i.e., photon radiation transport). The TT framework enables simulations at discretizations that might otherwise be computationally infeasible on conventional hardware. We show that solutions to certain multigroup problems possess an intrinsic low-rank structure, which the TT representation leverages effectively. This enables us to solve problems where the discretized solution size exceeds a trillion parameters on a single node. The solver is evaluated on a range of test problems with varying levels of complexity, consistently achieving compression factors greater than $100 \times$ and speedups exceeding $2 \times$. We also investigate alternative TT topologies by analyzing the low-rank structure of the merged spatio-spectral core to assess the potential for greater compression. This analysis suggests that compression gains could increase by factors as large as $7$. Our results indicate that the low-rank structure of the merged spatio-spectral core captures the spatio-spectral complexity of the solution, largely driven by the opacity structure of the medium. Beyond identifying opportunities for improved compression, this analysis highlights the types of errors that may arise in angle-integrated quantities when exploiting this low-rank structure.

astro-ph.IM↗

arXiv2025

Thermal Radiation Transport with Tensor Trains

We present a novel tensor network algorithm to solve the time-dependent, gray thermal radiation transport equation. The method invokes a tensor train (TT) decomposition for the specific intensity. The efficiency of this approach is dictated by the rank of the decomposition. When the solution is "low-rank," the memory footprint of the specific intensity solution vector may be significantly compressed. The algorithm, following a step-then-truncate approach of a traditional discrete ordinates method, operates directly on the compressed state vector thereby enabling large speedups for low-rank solutions. To achieve these speedups we rely on a recently developed rounding approach based on the Gram-SVD. We detail how familiar SN algorithms for (gray) thermal transport can be mapped to this TT framework and present several numerical examples testing both the optically thick and thin regimes. The TT framework finds low rank structure and supplies up to $\simeq$60$\times$ speedups and $\simeq$1000$\times$ compressions for problems demanding large angle counts, thereby enabling previously intractable SN calculations and supplying a promising avenue to mitigate ray effects.

astro-ph.IM↗

arXiv2024

Proto-Neutron Star Convection and the Neutrino-Driven Wind: Implications for the $ν$p-Process

Recent studies of the neutrino-driven wind from proto-neutron stars have indicated that the wind is likely proton-rich for much of its lifetime, and the high flux of neutrinos can induce $ν$p-process nucleosynthesis allowing for the formation of heavy elements. It has also been shown that gravito-acoustic waves, generated by convection within the proto-neutron star, can significantly alter the dynamics and nucleosynthesis in the wind. Therefore, we present a study of the effects of convection-driven waves on the nucleosynthesis in proton-rich neutrino-driven winds, focusing on the $ν$p-process. We find that wave effects can strongly impact $ν$p-process nucleosynthesis even at wave luminosities a factor of $10^{-5}$ smaller than the total neutrino luminosity. The momentum flux of the waves accelerates the wind, reducing the net neutrino heating and the persistent neutron abundance created by p($\barν_e,e^+$), which impedes $ν$p-process nucleosynthesis. However, this effect is generally counteracted by the effects of waves on seed nucleus formation, as the acceleration of the wind and the heating that occurs as these waves shock both favor a more $α$-rich environment with very little heavy seed-nucleus formation. Overall, higher wave luminosities correlate (albeit non-monotonically) with heavier element $ν$p-processing, up to $A\approx 200$ in some cases. At very high wave luminosities ($\gtrsim 10^{-3}L_ν$), early shock heating by the waves disrupts $α$ recombination, and drives a suppressed, fast-outflow r-process proceeding up to $A\approx 200$. This occurs despite an assumed neutrino spectrum that predicts a proton-rich wind with equilibrium $Y_e=0.6$.

astro-ph.HE↗

arXiv2023

Old Data, New Forensics: The First Second of SN 1987A Neutrino Emission

The next Milky Way supernova will be an epochal event in multi-messenger astronomy, critical to tests of supernovae, neutrinos, and new physics. Realizing this potential depends on having realistic simulations of core collapse. We investigate the neutrino predictions of nearly all modern models (1-, 2-, and 3-d) over the first $\simeq$1 s, making the first detailed comparisons of these models to each other and to the SN 1987A neutrino data. Even with different methods and inputs, the models generally agree with each other. However, even considering the low neutrino counts, the models generally disagree with data. What can cause this? We show that neither neutrino oscillations nor different progenitor masses appear to be a sufficient solution. We outline urgently needed work.

astro-ph.HE↗

arXiv2023

Proto-Neutron Star Convection and the Neutrino-Driven Wind: Implications for the r-Process

The neutrino-driven wind from proto-neutron stars is a proposed site for r-process nucleosynthesis, although most previous work has found that a wind heated only by neutrinos cannot produce the third r-process peak. However, several groups have noted that introducing a secondary heating source within the wind can change the hydrodynamic conditions sufficiently for a strong r-process to proceed. One possible secondary heating source is gravito-acoustic waves, generated by convection inside the proto-neutron star. As these waves propagate into the wind, they can both accelerate the wind and shock and deposit energy into the wind. Additionally, the acceleration of the wind by these waves can reduce the total number of neutrino captures and thereby reduce the final electron fraction of the wind. In neutron rich conditions, all of these effects can make conditions more favorable for r-process nucleosynthesis. Here, we present a systematic investigation of the impact of these convection-generated gravito-acoustic waves within the wind on potential nucleosynthesis. We find that wave effects in the wind can generate conditions favorable for a strong r-process, even when the energy flux in the waves is a factor of $10^{-4}$ smaller than the total neutrino energy flux and the wind is marginally neutron-rich. Nevertheless, this depends strongly on the radius at which the wave become non-linear and form shocks. We also find that both entropy production after shock formation and the acceleration of the wind due to stresses produced by the waves prior to shock formation impact the structure and nucleosynthesis of these waves.

astro-ph.HE↗

arXiv2022

Parthenon -- a performance portable block-structured adaptive mesh refinement framework

On the path to exascale the landscape of computer device architectures and corresponding programming models has become much more diverse. While various low-level performance portable programming models are available, support at the application level lacks behind. To address this issue, we present the performance portable block-structured adaptive mesh refinement (AMR) framework Parthenon, derived from the well-tested and widely used Athena++ astrophysical magnetohydrodynamics code, but generalized to serve as the foundation for a variety of downstream multi-physics codes. Parthenon adopts the Kokkos programming model, and provides various levels of abstractions from multi-dimensional variables, to packages defining and separating components, to launching of parallel compute kernels. Parthenon allocates all data in device memory to reduce data movement, supports the logical packing of variables and mesh blocks to reduce kernel launch overhead, and employs one-sided, asynchronous MPI calls to reduce communication overhead in multi-node simulations. Using a hydrodynamics miniapp, we demonstrate weak and strong scaling on various architectures including AMD and NVIDIA GPUs, Intel and AMD x86 CPUs, IBM Power9 CPUs, as well as Fujitsu A64FX CPUs. At the largest scale on Frontier (the first TOP500 exascale machine), the miniapp reaches a total of $1.7\times10^{13}$ zone-cycles/s on 9,216 nodes (73,728 logical GPUs) at ~92% weak scaling parallel efficiency (starting from a single node). In combination with being an open, collaborative project, this makes Parthenon an ideal framework to target exascale simulations in which the downstream developers can focus on their specific application rather than on the complexity of handling massively-parallel, device-accelerated AMR.

cs.DC↗

arXiv2022

Angular Momentum Transport in Proto-Neutron Stars and the Fate of Neutron Star Merger Remnants

Both the core collapse of rotating massive stars, and the coalescence of neutron star (NS) binaries, result in the formation of a hot, differentially rotating NS remnant. The timescales over which differential rotation is removed by internal angular-momentum transport processes (`viscosity') has key implications for the remnant's long-term stability and the NS equation-of-state (EOS). Guided by a non-rotating model of a cooling proto-NS, we estimate the dominant sources of viscosity using an externally imposed angular velocity profile $Ω(r)$. Although the magnetorotational instability provides the dominant source of effective viscosity at large radii, convection and/or the Spruit-Tayler dynamo dominate in the core of merger remnants where $dΩ/dr \geq 0$. Furthermore, the viscous timescale in the remnant core is sufficiently short that solid body rotation will be enforced faster than matter is accreted from rotationally-supported outer layers. Guided by these results, we develop a toy model for how the merger remnant core grows in mass and angular momentum due to accretion. We find that merger remnants with sufficiently massive and slowly rotating initial cores may collapse to black holes via envelope accretion, even when the total remnant mass is less than the usually considered threshold $\approx 1.2 M_{\rm TOV}$ for forming a stable solid-body rotating NS remnant (where $M_{\rm TOV}$ is the maximum non-rotating NS mass supported by the EOS). This qualitatively new picture of the post-merger remnant evolution and stability criterion has important implications for the expected electromagnetic counterparts from binary NS mergers and for multi-messenger constraints on the NS EOS.

astro-ph.HE↗

arXiv2022

Wave heating from proto-neutron star convection and the core-collapse supernova explosion mechanism

Our understanding of the core-collapse supernova explosion mechanism is incomplete. While the favoured scenario is delayed revival of the stalled shock by neutrino heating, it is difficult to reliably compute explosion outcomes and energies, which depend sensitively on the complex radiation hydrodynamics of the post-shock region. The dynamics of the (non-)explosion depend sensitively on how energy is transported from inside and near the proto-neutron star (PNS) to material just behind the supernova shock. Although most of the PNS energy is lost in the form of neutrinos, hydrodynamic and hydromagnetic waves can also carry energy from the PNS to the shock. We show that gravity waves excited by core PNS convection can couple with outgoing acoustic waves that present an appreciable source of energy and pressure in the post-shock region. Using one-dimensional simulations, we estimate the gravity wave energy flux excited by PNS convection and the fraction of this energy transmitted upward to the post-shock region as acoustic waves. We find wave energy fluxes near $10^{51}\,\mathrm{erg}\,\mathrm{s}^{-1}$ are likely to persist for $\sim1\,\mathrm{s}$ post-bounce. The wave pressure on the shock may exceed $10\%$ of the thermal pressure, potentially contributing to shock revival and, subsequently, a successful and energetic explosion. We also discuss how future simulations can better capture the effects of waves, and more accurately quantify wave heating rates.

astro-ph.HE↗

arXiv2022

HARM3D+NUC: A new method for simulating the post-merger phase of binary neutron star mergers with GRMHD, tabulated EOS and neutrino leakage

The first binary neutron star merger has already been detected in gravitational waves. The signal was accompanied by an electromagnetic counterpart including a kilonova component powered by the decay of radioactive nuclei, as well as a short $γ$-ray burst. In order to understand the radioactively-powered signal, it is necessary to simulate the outflows and their nucleosynthesis from the post-merger disk. Simulating the disk and predicting the composition of the outflows requires general relativistic magnetohydrodynamical (GRMHD) simulations that include a realistic, finite-temperature equation of state (EOS) and self-consistently calculating the impact of neutrinos. In this work, we detail the implementation of a finite-temperature EOS and the treatment of neutrinos in the GRMHD code HARM3D+NUC, based on HARM3D. We include formal tests of both the finite-temperature EOS and the neutrino leakage scheme. We further test the code by showing that, given conditions similar to those of published remnant disks following neutron star mergers, it reproduces both recombination of free nucleons to a neutron-rich composition and excitation of a thermal wind.

astro-ph.HE↗

arXiv2022

Machine learning moment closure models for the radiative transfer equation I: directly learning a gradient based closure

In this paper, we take a data-driven approach and apply machine learning to the moment closure problem for radiative transfer equation in slab geometry. Instead of learning the unclosed high order moment, we propose to directly learn the gradient of the high order moment using neural networks. This new approach is consistent with the exact closure we derive for the free streaming limit and also provides a natural output normalization. A variety of benchmark tests, including the variable scattering problem, the Gaussian source problem with both periodic and reflecting boundaries, and the two-material problem, show both good accuracy and generalizability of our machine learning closure model.

math.NA↗

arXiv2021

Machine learning moment closure models for the radiative transfer equation III: enforcing hyperbolicity and physical characteristic speeds

This is the third paper in a series in which we develop machine learning (ML) moment closure models for the radiative transfer equation (RTE). In our previous work \cite{huang2021gradient}, we proposed an approach to learn the gradient of the unclosed high order moment, which performs much better than learning the moment itself and the conventional $P_N$ closure. However, while the ML moment closure has better accuracy, it is not able to guarantee hyperbolicity and has issues with long time stability. In our second paper \cite{huang2021hyperbolic}, we identified a symmetrizer which leads to conditions that enforce that the gradient based ML closure is symmetrizable hyperbolic and stable over long time. The limitation of this approach is that in practice the highest moment can only be related to four, or fewer, lower moments. In this paper, we propose a new method to enforce the hyperbolicity of the ML closure model. Motivated by the observation that the coefficient matrix of the closure system is a lower Hessenberg matrix, we relate its eigenvalues to the roots of an associated polynomial. We design two new neural network architectures based on this relation. The ML closure model resulting from the first neural network is weakly hyperbolic and guarantees the physical characteristic speeds, i.e., the eigenvalues are bounded by the speed of light. The second model is strictly hyperbolic and does not guarantee the boundedness of the eigenvalues. Several benchmark tests including the Gaussian source problem and the two-material problem show the good accuracy, stability and generalizability of our hyperbolic ML closure model.

math.NA↗

arXiv2021

Machine learning moment closure models for the radiative transfer equation II: enforcing global hyperbolicity in gradient based closures

This is the second paper in a series in which we develop machine learning (ML) moment closure models for the radiative transfer equation (RTE). In our previous work \cite{huang2021gradient}, we proposed an approach to directly learn the gradient of the unclosed high order moment, which performs much better than learning the moment itself and the conventional $P_N$ closure. However, the ML moment closure model in \cite{huang2021gradient} is not able to guarantee hyperbolicity and long time stability. We propose in this paper a method to enforce the global hyperbolicity of the ML closure model. The main idea is to seek a symmetrizer (a symmetric positive definite matrix) for the closure system, and derive constraints such that the system is globally symmetrizable hyperbolic. It is shown that the new ML closure system inherits the dissipativeness of the RTE and preserves the correct diffusion limit as the Knunsden number goes to zero. Several benchmark tests including the Gaussian source problem and the two-material problem show the good accuracy, long time stability and generalizability of our globally hyperbolic ML closure model.

math.NA↗

arXiv2020

Exciting Prospects for Detecting Late-Time Neutrinos from Core-Collapse Supernovae

The importance of detecting neutrinos from a Milky Way core-collapse supernova is well known. An under-studied phase is proto-neutron star cooling. For SN 1987A, this seemingly began at about 2 s, and is thus probed by only 6 of the 19 events (and only the $\barν_e$ flavor) in the Kamiokande-II and IMB detectors. With the higher statistics expected for present and near-future detectors, it should be possible to measure detailed neutrino signals out to very late times. We present the first comprehensive study of neutrino detection during the proto-neutron star cooling phase, considering a variety of outcomes, using all flavors, and employing detailed detector physics. For our nominal model, the event yields (at 10 kpc) after 10 s -- the approximate duration of the SN 1987A signal -- far exceed the entire SN 1987A yield, with $\simeq$250 $\barν_e$ events (to 50 s) in Super-Kamiokande, $\simeq$110 $ν_e$ events (to 40 s) in DUNE, and $\simeq$10 $ν_μ, ν_τ, \barν_μ, \barν_τ$ events (to 20 s) in JUNO. These data would allow unprecedented probes of the proto-neutron star, including the onset of neutrino transparency and hence its transition to a neutron star. If a black hole forms, even at very late times, this can be clearly identified. But will the detectors fulfill their potential for this perhaps once-ever opportunity for an all-flavor, high-statistics detection of a core collapse? Maybe. Further work is urgently needed, especially for DUNE to thoroughly investigate and improve its MeV capabilities.

astro-ph.HE↗

arXiv2020

Reaction Rate Sensitivity of the Production of $γ$-ray Emitting Isotopes in Core-Collapse Supernova

Radioactive isotopes produced in core-collapse supernovae (CCSNe) provide useful insights into the underlying processes driving the collapse mechanism and the origins of elemental abundances. Their study generates a confluence of major physics research, including experimental measurements of nuclear reaction rates, astrophysical modeling, and $γ$-ray observations. Here we identify the key nuclear reaction rates to the nucleosynthesis of observable radioactive isotopes in explosive silicon-burning during CCSNe. Using the nuclear reaction network calculator SkyNet and current REACLIB reaction rates, we evolve temperature-density-time profiles of the innermost $0.45~M_\odot$ ejecta from the core collapse and explosion of a $12~M_\odot$ star. Individually varying 3403 reaction rates by factors of 100, we identify 141 reactions which cause significant differences in the isotopes of interest, namely, $^{43}$K, $^{47}$Ca, $^{44,47}$Sc, $^{44}$Ti, $^{48,51}$Cr, $^{48,49}$V, $^{52,53}$Mn, $^{55,59}$Fe, $^{56,57}$Co, and $^{56,57,59}$Ni. For each of these reactions, we present a novel method to extract the temperature range pertinent to the nucleosynthesis of the relevant isotope; the resulting temperatures lie within the range $T = 0.47$ to $6.15~$GK. Limiting the variations to within $1σ$ of STARLIB reaction rate uncertainties further reduces the identified reactions to 48 key rates, which can be used to guide future experimental research. Complete results are presented in tabular form.

astro-ph.SR↗

arXiv2019

Binary Neutron Star Mergers: Mass Ejection, Electromagnetic Counterparts and Nucleosynthesis

We present a systematic numerical relativity study of the mass ejection and the associated electromagnetic transients and nucleosynthesis from binary neutron star (NS) mergers. We find that a few $10^{-3}\, M_\odot$ of material are ejected dynamically during the mergers. The amount and the properties of these outflow depend on binary parameters and on the NS equation of state (EOS). A small fraction of these ejecta, typically ${\sim}10^{-6}\, M_\odot$, is accelerated by shocks formed shortly after merger to velocities larger than $0.6\, {\rm c}$ and produces bright radio flares on timescales of weeks, months, or years after merger. Their observation could constrain the strength with which the NSs bounce after merger and, consequently, the EOS of matter at extreme densities. The dynamical ejecta robustly produce second and third $r$-process peak nuclei with relative isotopic abundances close to solar. The production of light $r$-process elements is instead sensitive to the binary mass ratio and the neutrino radiation treatment. Accretion disks of up to ${\sim}0.2\, M_\odot$ are formed after merger, depending on the lifetime of the remnant. In most cases, neutrino- and viscously-driven winds from these disks dominate the overall outflow. Finally, we generate synthetic kilonova light curves and find that kilonovae depend on the merger outcome and could be used to constrain the NS EOS.

astro-ph.HE↗

arXiv2018

Viscous-Dynamical Ejecta from Binary Neutron Star Merger

General-relativistic simulations of binary neutron star mergers with viscosity reveal a new outflow mechanism operating in unequal mass binaries on dynamical timescales and enabled by turbulent viscosity. These "viscous-dynamical" ejecta are launched during merger due to the thermalization of mass exchange streams between the secondary and the primary neutron star. They are characterized by asymptotic velocities extending up to ${\sim} 0.8\, c$, and have masses that depend on the efficiency of the viscous mechanism. Depending on the unknown strength of the effective viscosity arising from magnetohydrodynamics instabilities operating during merger, the overall mass of the dynamical ejecta could be enhanced by a factor of a few and the mass of the fast tail of the ejecta having asymptotic velocities $\geq 0.6\, c$ by up to four orders of magnitude. The radioactive decay of the expanding viscous-dynamical ejecta could produce bright kilonova transients with signatures of free neutron decay in the first hour and enhanced near infrared flux on a timescale of a few days. The synchrotron remnant produced by the interaction between the ejecta and the interstellar medium could also be significantly enhanced by viscosity. Such remnant could be detected in the case of GW170817 as a rebrightening of the radio signal in the next months to years.

astro-ph.HE↗

arXiv2018

FRIB and the GW170817 Kilonova

In July 2018 an FRIB Theory Alliance program was held on the implications of GW170817 and its associated kilonova for r-process nucleosynthesis. Topics of discussion included the astrophysical and nuclear physics uncertainties in the interpretation of the GW170817 kilonova, what we can learn about the astrophysical site or sites of the r process from this event, and the advances in nuclear experiment and theory most crucial to pursue in light of the new data. Here we compile a selection of scientific contributions to the workshop, broadly representative of progress in r-process studies since the GW170817 event.

astro-ph.HE↗

arXiv2018

A large scale dynamo and magnetoturbulence in rapidly rotating core-collapse supernovae

Magnetohydrodynamic (MHD) turbulence is of key importance in many high-energy astrophysical systems, where MHD instabilities can amplify local magnetic field over very short time scales. Specifically, the magnetorotational instability (MRI) and dynamo action have been suggested as a mechanism to grow magnetar-strength magnetic field ($\ge10^{15} G$) and magnetorotationally power the explosion of a rotating massive star. Such stars are progenitor candidates for type Ic-bl hypernova explosions and make up all supernovae connected to long gamma-ray bursts (GRBs). The MRI has been studied with local high-resolution shearing box simulations in 3D or with global 2D simulations, but it is an open question whether MRI-driven turbulence can result in the creation of a large-scale ordered and dynamically relevant field. Here we report results from global 3D general-relativistic magnetohydrodynamic (GRMHD) turbulence simulations and show that MRI-driven MHD turbulence in rapidly rotating protoneutron stars produces an inverse cascade of energy. We find a large-scale ordered toroidal field that is consistent with the formation of bipolar magnetorotationally driven outflows. Our results demonstrate that rapidly rotating massive stars are plausible progenitors for both type Ic-bl supernovae and long GRBs, present a viable formation scenario for magnetars, and may account for potentially magnetar-powered superluminous supernovae.

astro-ph.HE↗