Matthew W. Choptuik
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Matthew W. Choptuik has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Benson Way
Departament de Física Quàntica i Astrofísica Institut de Ciències del Cosmos Universitat de Barcelona Martí i Franquès 1 E-08028 Barcelona Spain
6 shared publications - Jorge E. Santos
University of Cambridge
4 shared publications - Michael P. Kinach
University of British Columbia
4 shared publications - Gray D. Reid
Department of Physics and Astronomy, University of British Columbia, Vancouver, British Columbia, V6T 1Z1 Canada
3 shared publications - Jason F. Ventrella
Center for Computation and Technology, Louisiana State University, Baton Rouge, Louisiana 70803, USA
3 shared publications - Evgeny Sorkin
Max-Planck Institute for Gravitational Physics Albert Einstein Institute Am Muehlenberg 1 14476 Golm Germany
2 shared publications
Latest Publications
- Open Access
- 18 March 2026
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 7 (2)
Addressing the gravitational collapse of a massless scalar field with Physics-Informed Neural Networks
The gravitational collapse of a massless scalar field remains a demanding benchmark for numerical methods in numerical relativity, as it exhibits critical behavior at the boundary between dispersion and black hole formation. In this work, we revisit this problem by relying on Physics-Informed Neural Networks (PINNs) as flexible solvers for partial differential equations, thereby providing a comparative assessment of several recent neural architectures. Building on the Einstein-massless-Klein-Gordon formulation in polar-areal coordinates, we consider four initial-value problems encompassing subcritical, critical, and supercritical regimes and use high-resolution finite-difference simulations as reference solutions. Our study is primarily comparative: we evaluate several state-of-the-art deep learning architectures, including vanilla and high-precision PINNs, sinusoidal-feature and quadratic-residual variants, and Kolmogorov-Arnold Networks, all trained under a common loss design that encodes the field equations, boundary conditions, and causal time-space enforcement, together with a novel adaptive spacetime sampling. Within this framework, we also introduce ModPINN, a modest modification of standard PINNs that augments standard multilayer perceptrons with coordinate embeddings, quadratic layers, and other common ingredients in recent literature. This study shows that deep-learning-based methods can reproduce finite-difference solutions for the scalar field and the spacetime metric with competitive accuracy using significantly fewer collocation points than more traditional methodologies. While no single architecture dominates in all regimes, ModPINN achieves particularly stable and accurate solutions near criticality, indicating that suitably designed embeddings and adaptive sampling can enhance the robustness of PINNs for challenging gravitational-collapse scenarios.
- 18 March 2026
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 7 (2)
Open AccessAddressing the gravitational collapse of a massless scalar field with Physics-Informed Neural Networks
The gravitational collapse of a massless scalar field remains a demanding benchmark for numerical methods in numerical relativity, as it exhibits critical behavior at the boundary between dispersion and black hole formation. In this work, we revisit this problem by relying on Physics-Informed Neural Networks (PINNs) as flexible solvers for partial differential equations, thereby providing a comparative assessment of several recent neural architectures. Building on the Einstein-massless-Klein-Gordon formulation in polar-areal coordinates, we consider four initial-value problems encompassing subcritical, critical, and supercritical regimes and use high-resolution finite-difference simulations as reference solutions. Our study is primarily comparative: we evaluate several state-of-the-art deep learning architectures, including vanilla and high-precision PINNs, sinusoidal-feature and quadratic-residual variants, and Kolmogorov-Arnold Networks, all trained under a common loss design that encodes the field equations, boundary conditions, and causal time-space enforcement, together with a novel adaptive spacetime sampling. Within this framework, we also introduce ModPINN, a modest modification of standard PINNs that augments standard multilayer perceptrons with coordinate embeddings, quadratic layers, and other common ingredients in recent literature. This study shows that deep-learning-based methods can reproduce finite-difference solutions for the scalar field and the spacetime metric with competitive accuracy using significantly fewer collocation points than more traditional methodologies. While no single architecture dominates in all regimes, ModPINN achieves particularly stable and accurate solutions near criticality, indicating that suitably designed embeddings and adaptive sampling can enhance the robustness of PINNs for challenging gravitational-collapse scenarios.
- Open Access
- 28 October 2024
- Published by American Physical Society (APS) in Physical Review D
- Vol. 110 (7)
Dynamics of gauged -balls in three spatial dimensionsWe investigate the dynamics of U ( 1 ) gauged Q -balls using fully three-dimensional numerical simulations. We consider two different scenarios: first, the classical stability of gauged Q -balls with respect to generic three-dimensional perturbations, and second, the behavior of gauged Q -balls during head-on and off-axis collisions at relativistic velocities. With regard to stability, we find that there exist gauged Q -ball configurations which are classically stable in both logarithmic and polynomial scalar field models. With regard to relativistic collisions, we find that the dynamics can depend on many different parameters such as the collision velocity, relative phase, relative charge, and impact parameter of the colliding Q -balls. Published by the American Physical Society 2024
- 28 October 2024
- Published by American Physical Society (APS) in Physical Review D
- Vol. 110 (7)
Open AccessDynamics of gauged -balls in three spatial dimensionsWe investigate the dynamics of U ( 1 ) gauged Q -balls using fully three-dimensional numerical simulations. We consider two different scenarios: first, the classical stability of gauged Q -balls with respect to generic three-dimensional perturbations, and second, the behavior of gauged Q -balls during head-on and off-axis collisions at relativistic velocities. With regard to stability, we find that there exist gauged Q -ball configurations which are classically stable in both logarithmic and polynomial scalar field models. With regard to relativistic collisions, we find that the dynamics can depend on many different parameters such as the collision velocity, relative phase, relative charge, and impact parameter of the colliding Q -balls. Published by the American Physical Society 2024
- Open Access
- 9 July 2024
- Published by American Physical Society (APS) in Physical Review D
- Vol. 110 (1) , 015012
Relativistic head-on collisions of gauged -ballsWe investigate the collision dynamics of gauged -balls by performing high-resolution numerical simulations in axisymmetry. Focusing on the case of relativistic head-on collisions, we consider the effects of the initial velocity, relative phase, relative charge, and electromagnetic coupling strength on the outcome of the collision. We find that the collision dynamics can depend strongly on these parameters; most notably, electromagnetic effects can significantly alter the outcome of the collision when the gauge coupling is large. When the gauge coupling is small, we find that the dynamics generally resemble those of ordinary (nongauged) -balls.- 9 July 2024
- Published by American Physical Society (APS) in Physical Review D
- Vol. 110 (1) , 015012
Open AccessRelativistic head-on collisions of gauged -ballsWe investigate the collision dynamics of gauged -balls by performing high-resolution numerical simulations in axisymmetry. Focusing on the case of relativistic head-on collisions, we consider the effects of the initial velocity, relative phase, relative charge, and electromagnetic coupling strength on the outcome of the collision. We find that the collision dynamics can depend strongly on these parameters; most notably, electromagnetic effects can significantly alter the outcome of the collision when the gauge coupling is large. When the gauge coupling is small, we find that the dynamics generally resemble those of ordinary (nongauged) -balls. - 4 January 2024
- Published by American Physical Society (APS) in Physical Review D
- Vol. 109 (2) , 024008
Stability of nonminimally coupled topological-defect boson stars
As shown by Marunovic and Murkovic [ Classical Quantum Gravity 31, 045010 (2014)], nonminimal d-stars, composite structures consisting of a boson star and a global monopole nonminimally coupled to the general relativistic field, can have extremely high gravitational compactness. In a previous paper we demonstrated that these ground-state stationary solutions are sometimes additionally characterized by shells of bosonic matter located far from the center of symmetry [ Phys. Rev. D 93, 044022 (2016)]. In order to investigate the question of stability posed by Marunovic and Murkovic, we investigate the stability of several families of d-stars using both numerical simulations and linear perturbation theory. For all families investigated, we find that the most highly compact solutions, along with those solutions exhibiting shells of bosonic matter, are unstable to radial perturbations and are therefore poor candidates for astrophysically relevant black hole mimickers or other highly compact stable objects.
- 4 January 2024
- Published by American Physical Society (APS) in Physical Review D
- Vol. 109 (2) , 024008
Stability of nonminimally coupled topological-defect boson stars
As shown by Marunovic and Murkovic [ Classical Quantum Gravity 31, 045010 (2014)], nonminimal d-stars, composite structures consisting of a boson star and a global monopole nonminimally coupled to the general relativistic field, can have extremely high gravitational compactness. In a previous paper we demonstrated that these ground-state stationary solutions are sometimes additionally characterized by shells of bosonic matter located far from the center of symmetry [ Phys. Rev. D 93, 044022 (2016)]. In order to investigate the question of stability posed by Marunovic and Murkovic, we investigate the stability of several families of d-stars using both numerical simulations and linear perturbation theory. For all families investigated, we find that the most highly compact solutions, along with those solutions exhibiting shells of bosonic matter, are unstable to radial perturbations and are therefore poor candidates for astrophysically relevant black hole mimickers or other highly compact stable objects.
- 27 December 2023
- Published by American Physical Society (APS) in Physical Review D
- Vol. 108 (12) , 124070
Reference metric approach to the Z4 system
The hyperbolic formulations of numerical relativity due to Baumgarte, Shapiro, Shibata, and Nakamura (BSSN) and Nagy Ortiz and Reula (NOR), among others, achieve stability through the effective embedding of general relativity within the larger Z4 system. In doing so, various elliptic constraints are promoted to dynamical degrees of freedom, permitting the advection of constraint violating modes. Here we demonstrate that it is possible to achieve equivalent performance through a modification of fully covariant and conformal Z4 (FCCZ4) wherein constraint violations are coupled to a reference metric completely independently of the physical metric. We show that this approach works in the presence of black holes and holds up robustly in a variety of spherically symmetric simulations including the critical collapse of a scalar field. We then demonstrate that our formulation is strongly hyperbolic through the use of a pseudodifferential first order reduction and compare its hyperbolicity properties to those of FCCZ4 and generalized BSSN (GBSSN). Our present approach makes use of a static Lorentzian reference metric and does not appear to provide significant advantages over FCCZ4. However, we speculate that dynamical specification of the reference metric may provide a means of exerting greater control over constraint violations than what is provided by current BSSN-type formulations.
- 27 December 2023
- Published by American Physical Society (APS) in Physical Review D
- Vol. 108 (12) , 124070
Reference metric approach to the Z4 system
The hyperbolic formulations of numerical relativity due to Baumgarte, Shapiro, Shibata, and Nakamura (BSSN) and Nagy Ortiz and Reula (NOR), among others, achieve stability through the effective embedding of general relativity within the larger Z4 system. In doing so, various elliptic constraints are promoted to dynamical degrees of freedom, permitting the advection of constraint violating modes. Here we demonstrate that it is possible to achieve equivalent performance through a modification of fully covariant and conformal Z4 (FCCZ4) wherein constraint violations are coupled to a reference metric completely independently of the physical metric. We show that this approach works in the presence of black holes and holds up robustly in a variety of spherically symmetric simulations including the critical collapse of a scalar field. We then demonstrate that our formulation is strongly hyperbolic through the use of a pseudodifferential first order reduction and compare its hyperbolicity properties to those of FCCZ4 and generalized BSSN (GBSSN). Our present approach makes use of a static Lorentzian reference metric and does not appear to provide significant advantages over FCCZ4. However, we speculate that dynamical specification of the reference metric may provide a means of exerting greater control over constraint violations than what is provided by current BSSN-type formulations.
- 13 November 2023
- Published by American Physical Society (APS) in Physical Review D
- Vol. 108 (10) , 104021
Universality in the critical collapse of the Einstein-Maxwell system
We report on critical phenomena in the gravitational collapse of the electromagnetic field in axisymmetry using cylindrical coordinates. We perform detailed numerical simulations of four families of dipole and quadrupole initial data fine-tuned to the onset of black hole formation. It has been previously observed that families that bifurcate into two on-axis critical solutions exhibit distinct growth characteristics from those that collapse at the center of symmetry. In contrast, our results indicate similar growth characteristics and periodicity across all families of initial data, including those examined in earlier works. More precisely, for all families investigated, we observe power-law scaling for the maximum of the electromagnetic field invariant () with . We find evidence of approximate discrete self-similarity in near-critical time evolutions with a log-scale echoing period of across all families of initial data. Our methodology, while reproducing the results of prior studies up to a point, provides new insights into the later stages of critical searches and we propose a mechanism to explain the observed differences between our work and the previous calculations.- 13 November 2023
- Published by American Physical Society (APS) in Physical Review D
- Vol. 108 (10) , 104021
Universality in the critical collapse of the Einstein-Maxwell system
We report on critical phenomena in the gravitational collapse of the electromagnetic field in axisymmetry using cylindrical coordinates. We perform detailed numerical simulations of four families of dipole and quadrupole initial data fine-tuned to the onset of black hole formation. It has been previously observed that families that bifurcate into two on-axis critical solutions exhibit distinct growth characteristics from those that collapse at the center of symmetry. In contrast, our results indicate similar growth characteristics and periodicity across all families of initial data, including those examined in earlier works. More precisely, for all families investigated, we observe power-law scaling for the maximum of the electromagnetic field invariant () with . We find evidence of approximate discrete self-similarity in near-critical time evolutions with a log-scale echoing period of across all families of initial data. Our methodology, while reproducing the results of prior studies up to a point, provides new insights into the later stages of critical searches and we propose a mechanism to explain the observed differences between our work and the previous calculations.- Open Access
A Stable and Scalable Method for Solving Initial Value PDEs with Neural Networks
Unlike conventional grid and mesh based methods for solving partial differential equations (PDEs), neural networks have the potential to break the curse of dimensionality, providing approximate solutions to problems where using classical solvers is difficult or impossible. While global minimization of the PDE residual over the network parameters works well for boundary value problems, catastrophic forgetting impairs the applicability of this approach to initial value problems (IVPs). In an alternative local-in-time approach, the optimization problem can be converted into an ordinary differential equation (ODE) on the network parameters and the solution propagated forward in time; however, we demonstrate that current methods based on this approach suffer from two key issues. First, following the ODE produces an uncontrolled growth in the conditioning of the problem, ultimately leading to unacceptably large numerical errors. Second, as the ODE methods scale cubically with the number of model parameters, they are restricted to small neural networks, significantly limiting their ability to represent intricate PDE initial conditions and solutions. Building on these insights, we develop Neural IVP, an ODE based IVP solver which prevents the network from getting ill-conditioned and runs in time linear in the number of parameters, enabling us to evolve the dynamics of challenging PDEs with neural networks.
Open AccessA Stable and Scalable Method for Solving Initial Value PDEs with Neural Networks
Unlike conventional grid and mesh based methods for solving partial differential equations (PDEs), neural networks have the potential to break the curse of dimensionality, providing approximate solutions to problems where using classical solvers is difficult or impossible. While global minimization of the PDE residual over the network parameters works well for boundary value problems, catastrophic forgetting impairs the applicability of this approach to initial value problems (IVPs). In an alternative local-in-time approach, the optimization problem can be converted into an ordinary differential equation (ODE) on the network parameters and the solution propagated forward in time; however, we demonstrate that current methods based on this approach suffer from two key issues. First, following the ODE produces an uncontrolled growth in the conditioning of the problem, ultimately leading to unacceptably large numerical errors. Second, as the ODE methods scale cubically with the number of model parameters, they are restricted to small neural networks, significantly limiting their ability to represent intricate PDE initial conditions and solutions. Building on these insights, we develop Neural IVP, an ODE based IVP solver which prevents the network from getting ill-conditioned and runs in time linear in the number of parameters, enabling us to evolve the dynamics of challenging PDEs with neural networks.
- Open AccessDynamical evolution of $U(1)$ gauged Q-balls in axisymmetryWe study the dynamics of $U(1)$ gauged Q-balls using fully non-linear numerical evolutions in axisymmetry. Focusing on two models with logarithmic and polynomial scalar field potentials, we numerically evolve perturbed gauged Q-ball configurations in order to assess their stability and determine the fate of unstable configurations. Our simulations suggest that there exist both stable and unstable branches of solutions with respect to axisymmetric perturbations. For solutions belonging to the stable branch, the gauged Q-balls respond to the perturbations by oscillating continuously or weakly radiating before returning to the initial configuration. For the unstable branch, the solutions are eventually destroyed and can evolve in several ways, such as dispersal of the fields to infinity or fragmentation into smaller gauged Q-balls. In some cases, we observe the formation of ring-like structures which we call "gauged Q-rings". We also investigate the stability of gauged Q-balls when the gauge coupling is small, finding that the behaviour of these configurations closely resembles that of ordinary (non-gauged) Q-balls.Open AccessDynamical evolution of $U(1)$ gauged Q-balls in axisymmetryWe study the dynamics of $U(1)$ gauged Q-balls using fully non-linear numerical evolutions in axisymmetry. Focusing on two models with logarithmic and polynomial scalar field potentials, we numerically evolve perturbed gauged Q-ball configurations in order to assess their stability and determine the fate of unstable configurations. Our simulations suggest that there exist both stable and unstable branches of solutions with respect to axisymmetric perturbations. For solutions belonging to the stable branch, the gauged Q-balls respond to the perturbations by oscillating continuously or weakly radiating before returning to the initial configuration. For the unstable branch, the solutions are eventually destroyed and can evolve in several ways, such as dispersal of the fields to infinity or fragmentation into smaller gauged Q-balls. In some cases, we observe the formation of ring-like structures which we call "gauged Q-rings". We also investigate the stability of gauged Q-balls when the gauge coupling is small, finding that the behaviour of these configurations closely resembles that of ordinary (non-gauged) Q-balls.
- Open Access
- 21 February 2023
- Published by American Physical Society (APS) in Physical Review D
- Vol. 107 (3) , 035022
Dynamical evolution of gauged -balls in axisymmetryWe study the dynamics of gauged -balls using fully nonlinear numerical evolutions in axisymmetry. Focusing on two models with logarithmic and polynomial scalar field potentials, we numerically evolve perturbed gauged -ball configurations in order to assess their stability and determine the fate of unstable configurations. Our simulations suggest that there exist both stable and unstable branches of solutions with respect to axisymmetric perturbations. For solutions belonging to the stable branch, the gauged -balls respond to the perturbations by oscillating continuously or weakly radiating before returning to the initial configuration. For the unstable branch, the solutions are eventually destroyed and can evolve in several ways, such as dispersal of the fields to infinity or fragmentation into smaller gauged -balls. In some cases, we observe the formation of ringlike structures which we call “gauged -rings.” We also investigate the stability of gauged -balls when the gauge coupling is small, finding that the behavior of these configurations closely resembles that of ordinary (nongauged) -balls.- 21 February 2023
- Published by American Physical Society (APS) in Physical Review D
- Vol. 107 (3) , 035022
Open AccessDynamical evolution of gauged -balls in axisymmetryWe study the dynamics of gauged -balls using fully nonlinear numerical evolutions in axisymmetry. Focusing on two models with logarithmic and polynomial scalar field potentials, we numerically evolve perturbed gauged -ball configurations in order to assess their stability and determine the fate of unstable configurations. Our simulations suggest that there exist both stable and unstable branches of solutions with respect to axisymmetric perturbations. For solutions belonging to the stable branch, the gauged -balls respond to the perturbations by oscillating continuously or weakly radiating before returning to the initial configuration. For the unstable branch, the solutions are eventually destroyed and can evolve in several ways, such as dispersal of the fields to infinity or fragmentation into smaller gauged -balls. In some cases, we observe the formation of ringlike structures which we call “gauged -rings.” We also investigate the stability of gauged -balls when the gauge coupling is small, finding that the behavior of these configurations closely resembles that of ordinary (nongauged) -balls. - 26 September 2019
- Published by American Physical Society (APS) in Physical Review Letters
- Vol. 123 (13) , 131101
Multioscillating Boson Stars
We propose that stable boson stars generically fall within an infinite-parameter family of solutions that oscillate on any number of noncommensurate frequencies. We numerically construct two-frequency solutions and explore their parameter space. These solutions merge with the standard boson star family in the limit where the nondominating frequencies are turned off. We find that, for a fixed energy, these two-frequency solutions can differ considerably in size from standard boson stars.
- 26 September 2019
- Published by American Physical Society (APS) in Physical Review Letters
- Vol. 123 (13) , 131101
Multioscillating Boson Stars
We propose that stable boson stars generically fall within an infinite-parameter family of solutions that oscillate on any number of noncommensurate frequencies. We numerically construct two-frequency solutions and explore their parameter space. These solutions merge with the standard boson star family in the limit where the nondominating frequencies are turned off. We find that, for a fixed energy, these two-frequency solutions can differ considerably in size from standard boson stars.