-
Black Box Probabilistic Numerics
Authors:
Onur Teymur,
Christopher N. Foley,
Philip G. Breen,
Toni Karvonen,
Chris. J. Oates
Abstract:
Probabilistic numerics casts numerical tasks, such the numerical solution of differential equations, as inference problems to be solved. One approach is to model the unknown quantity of interest as a random variable, and to constrain this variable using data generated during the course of a traditional numerical method. However, data may be nonlinearly related to the quantity of interest, renderin…
▽ More
Probabilistic numerics casts numerical tasks, such the numerical solution of differential equations, as inference problems to be solved. One approach is to model the unknown quantity of interest as a random variable, and to constrain this variable using data generated during the course of a traditional numerical method. However, data may be nonlinearly related to the quantity of interest, rendering the proper conditioning of random variables difficult and limiting the range of numerical tasks that can be addressed. Instead, this paper proposes to construct probabilistic numerical methods based only on the final output from a traditional method. A convergent sequence of approximations to the quantity of interest constitute a dataset, from which the limiting quantity of interest can be extrapolated, in a probabilistic analogue of Richardson's deferred approach to the limit. This black box approach (1) massively expands the range of tasks to which probabilistic numerics can be applied, (2) inherits the features and performance of state-of-the-art numerical methods, and (3) enables provably higher orders of convergence to be achieved. Applications are presented for nonlinear ordinary and partial differential equations, as well as for eigenvalue problems-a setting for which no probabilistic numerical methods have yet been developed.
△ Less
Submitted 28 October, 2021; v1 submitted 15 June, 2021;
originally announced June 2021.
-
The Kinematic Richness of Star Clusters - II. Stability of Spherical Anisotropic Models with Rotation
Authors:
Philip G. Breen,
Simon Rozier,
Douglas C. Heggie,
Anna Lisa Varri
Abstract:
We study the bar instability in collisionless, rotating, anisotropic, stellar systems, using N-body simulations and also the matrix technique for calculation of modes with the perturbed collisionless Boltzmann equation. These methods are applied to spherical systems with an initial Plummer density distribution, but modified kinematically in two ways: the velocity distribution is tangentially aniso…
▽ More
We study the bar instability in collisionless, rotating, anisotropic, stellar systems, using N-body simulations and also the matrix technique for calculation of modes with the perturbed collisionless Boltzmann equation. These methods are applied to spherical systems with an initial Plummer density distribution, but modified kinematically in two ways: the velocity distribution is tangentially anisotropic, using results of Dejonghe, and the system is set in rotation by reversing the velocities of a fraction of stars in various regions of phase space, a la Lynden-Bell. The aim of the N-body simulations is first to survey the parameter space, and, using those results, to identify regions of phase space (by radius and orbital inclination) which have the most important influence on the bar instability. The matrix method is then used to identify the resonant interactions in the system which have the greatest effect on the growth rate of a bar. Complementary series of N-body simulations examine these processes in relation to the evolving frequency distribution and the pattern speed. Finally, the results are synthesised with an existing theoretical framework, and used to consider the old question of constructing a stability criterion.
△ Less
Submitted 8 February, 2021;
originally announced February 2021.
-
$l=1$: Weinberg's weakly damped mode in an $N$-body model of a spherical stellar system
Authors:
Douglas C. Heggie,
Philip G. Breen,
Anna Lisa Varri
Abstract:
Spherical stellar systems such as King models, in which the distribution function is a decreasing function of energy and depends on no other invariant, are stable in the sense of collisionless dynamics. But Weinberg showed, by a clever application of the matrix method of linear stability, that they may be nearly unstable, in the sense of possessing {\sl weakly} damped modes of oscillation. He also…
▽ More
Spherical stellar systems such as King models, in which the distribution function is a decreasing function of energy and depends on no other invariant, are stable in the sense of collisionless dynamics. But Weinberg showed, by a clever application of the matrix method of linear stability, that they may be nearly unstable, in the sense of possessing {\sl weakly} damped modes of oscillation. He also demonstrated the presence of such a mode in an $N$-body model by endowing it with initial conditions generated from his perturbative solution. In the present paper we provide evidence for the presence of this same mode in $N$-body simulations of the King $W_0 = 5$ model, in which the initial conditions are generated by the usual Monte Carlo sampling of the King distribution function. It is shown that the oscillation of the density centre correlates with variations in the structure of the system out to a radius of about 1 virial radius, but anticorrelates with variations beyond that radius. Though the oscillations appear to be continually reexcited (presumably by the motions of the particles) we show by calculation of power spectra that Weinberg's estimate of the period (strictly, $2π$ divided by the real part of the eigenfrequency) lies within the range where the power is largest. In addition, however, the power spectrum displays another very prominent feature at shorter periods, around 5 crossing times.
△ Less
Submitted 2 March, 2021; v1 submitted 3 February, 2020;
originally announced February 2020.
-
Newton vs the machine: solving the chaotic three-body problem using deep neural networks
Authors:
Philip G. Breen,
Christopher N. Foley,
Tjarda Boekholt,
Simon Portegies Zwart
Abstract:
Since its formulation by Sir Isaac Newton, the problem of solving the equations of motion for three bodies under their own gravitational force has remained practically unsolved. Currently, the solution for a given initialization can only be found by performing laborious iterative calculations that have unpredictable and potentially infinite computational cost, due to the system's chaotic nature. W…
▽ More
Since its formulation by Sir Isaac Newton, the problem of solving the equations of motion for three bodies under their own gravitational force has remained practically unsolved. Currently, the solution for a given initialization can only be found by performing laborious iterative calculations that have unpredictable and potentially infinite computational cost, due to the system's chaotic nature. We show that an ensemble of solutions obtained using an arbitrarily precise numerical integrator can be used to train a deep artificial neural network (ANN) that, over a bounded time interval, provides accurate solutions at fixed computational cost and up to 100 million times faster than a state-of-the-art solver. Our results provide evidence that, for computationally challenging regions of phase-space, a trained ANN can replace existing numerical solvers, enabling fast and scalable simulations of many-body systems to shed light on outstanding phenomena such as the formation of black-hole binary systems or the origin of the core collapse in dense star clusters.
△ Less
Submitted 16 October, 2019;
originally announced October 2019.
-
A Bulirsch-Stoer algorithm using Gaussian processes
Authors:
Philip G. Breen,
Christopher N. Foley
Abstract:
In this paper, we treat the problem of evaluating the asymptotic error in a numerical integration scheme as one with inherent uncertainty. Adding to the growing field of probabilistic numerics, we show that Gaussian process regression (GPR) can be embedded into a numerical integration scheme to allow for (i) robust selection of the adaptive step-size parameter and; (ii) uncertainty quantification…
▽ More
In this paper, we treat the problem of evaluating the asymptotic error in a numerical integration scheme as one with inherent uncertainty. Adding to the growing field of probabilistic numerics, we show that Gaussian process regression (GPR) can be embedded into a numerical integration scheme to allow for (i) robust selection of the adaptive step-size parameter and; (ii) uncertainty quantification in predictions of putatively converged numerical solutions. We present two examples of our approach using Richardson's extrapolation technique and the Bulirsch-Stoer algorithm. In scenarios where the error-surface is smooth and bounded, our proposed approach can match the results of the traditional polynomial (parametric) extrapolation methods. In scenarios where the error surface is not well approximated by a finite-order polynomial, e.g. in the vicinity of a pole or in the assessment of a chaotic system, traditional methods can fail, however, the non-parametric GPR approach demonstrates the potential to continue to furnish reasonable solutions in these situations.
△ Less
Submitted 23 May, 2019;
originally announced May 2019.
-
Mapping the stability of stellar rotating spheres via linear response theory
Authors:
Simon Rozier,
Jean-Baptiste Fouvry,
Philip G. Breen,
Anna Lisa Varri,
Christophe Pichon,
Douglas C. Heggie
Abstract:
Rotation is ubiquitous in the Universe, and recent kinematic surveys have shown that early type galaxies and globular clusters are no exception. Yet the linear response of spheroidal rotating stellar systems has seldom been studied. This paper takes a step in this direction by considering the behaviour of spherically symmetric systems with differential rotation. Specifically, the stability of seve…
▽ More
Rotation is ubiquitous in the Universe, and recent kinematic surveys have shown that early type galaxies and globular clusters are no exception. Yet the linear response of spheroidal rotating stellar systems has seldom been studied. This paper takes a step in this direction by considering the behaviour of spherically symmetric systems with differential rotation. Specifically, the stability of several sequences of Plummer spheres is investigated, in which the total angular momentum, as well as the degree and flavour of anisotropy in the velocity space are varied. To that end, the response matrix method is customised to spherical rotating equilibria. The shapes, pattern speeds and growth rates of the systems' unstable modes are computed. Detailed comparisons to appropriate N-body measurements are also presented. The marginal stability boundary is charted in the parameter space of velocity anisotropy and rotation rate. When rotation is introduced, two sequences of growing modes are identified corresponding to radially and tangentially-biased anisotropic spheres respectively. For radially anisotropic spheres, growing modes occur on two intersecting surfaces (in the parameter space of anisotropy and rotation), which correspond to fast and slow modes, depending on the net rotation rate. Generalised, approximate stability criteria are finally presented.
△ Less
Submitted 25 February, 2019;
originally announced February 2019.
-
Light element variations in globular clusters via nucleosynthesis in black hole accretion discs
Authors:
Philip G. Breen
Abstract:
Ancient globular clusters contain multiple stellar populations identified by variations in light elements (e.g., C, N, O, Na). Although many scenarios have been suggested to explain this phenomenon, all are faced with challenges when compared with all the observational evidence. In this Letter, we propose a new scenario in which light element variations originate from nucleosynthesis in accretion…
▽ More
Ancient globular clusters contain multiple stellar populations identified by variations in light elements (e.g., C, N, O, Na). Although many scenarios have been suggested to explain this phenomenon, all are faced with challenges when compared with all the observational evidence. In this Letter, we propose a new scenario in which light element variations originate from nucleosynthesis in accretion discs around black holes. Since the black holes form after a few $Myrs$, the cluster is expected to still be embedded in a gas rich environment. Through a simplified accretion model, we show that the correct light element anti-correlations can be produced. Assuming a Kroupa stellar initial mass function (IMF), each black hole would only have to process ${\approx}300M_{\odot}$ of material in order to explain multiple populations; over a period of $3Myr$ this corresponds to $ \sim10^{-4} M_{\odot}yr^{-1}$ (similar to the estimated accretion rate for the X-ray binary SS 433).
△ Less
Submitted 5 September, 2018; v1 submitted 24 April, 2018;
originally announced April 2018.
-
The Kinematic Richness of Star Clusters I. Isolated Spherical Models with Primordial Anisotropy
Authors:
Philip G. Breen,
Anna Lisa Varri,
Douglas C. Heggie
Abstract:
We investigate the dynamical evolution of isolated equal-mass star cluster models by means of direct N-body simulations, primarily focusing on the effects of the presence of primordial anisotropy in the velocity space. We found evidence of the existence of a monotonic relationship between the moment of core collapse and the amount and flavour of anisotropy in the stellar system. Specifically, equi…
▽ More
We investigate the dynamical evolution of isolated equal-mass star cluster models by means of direct N-body simulations, primarily focusing on the effects of the presence of primordial anisotropy in the velocity space. We found evidence of the existence of a monotonic relationship between the moment of core collapse and the amount and flavour of anisotropy in the stellar system. Specifically, equilibria characterised by the same initial structural properties (Plummer density profile) and with different degrees of tangentially-biased (radially-biased) anisotropy, reach core collapse earlier (later) than isotropic models. We interpret this result in light of an accelerated (delayed) phase of the early evolution of collisional stellar systems "anisotropic-response"), which we have characterised both in terms of the evolution of the velocity moments and of a fluid model of two-body relaxation. For the case of the most tangentially anisotropic model the initial phase of evolution involves a catastrophic collapse of the inner part of the system which continues until an isotropic velocity distribution is reached. This study represents a first step towards a comprehensive investigation of the role played by kinematic richness in the long-term dynamical evolution of collisional systems.
△ Less
Submitted 10 July, 2017;
originally announced July 2017.
-
Stellar envelopes of globular clusters embedded in dark mini-haloes
Authors:
Jorge Peñarrubia,
Anna Lisa Varri,
Philip G. Breen,
Annette M. N. Ferguson,
Rubén Sánchez-Janssen
Abstract:
We show that hard encounters in the central regions of globular clusters embedded in dark matter (DM) haloes necessarily lead to the formation of gravitationally-bound stellar envelopes that extend far beyond the nominal tidal radius of the system. Using statistical arguments and numerical techniques we derive the equilibrium distribution function of stars ejected from the centre of a non-divergen…
▽ More
We show that hard encounters in the central regions of globular clusters embedded in dark matter (DM) haloes necessarily lead to the formation of gravitationally-bound stellar envelopes that extend far beyond the nominal tidal radius of the system. Using statistical arguments and numerical techniques we derive the equilibrium distribution function of stars ejected from the centre of a non-divergent spherical potential. Independently of the velocity distribution with which stars are ejected, GC envelopes have density profiles that approach asymptotically $ρ\sim r^{-4}$ at large distances and become isothermal towards the centre. Adding a DM halo component leaves two clear-cut observational signatures: (i) a flattening, or slightly increase of the projected velocity dispersion profile at large distances, and (ii) an outer surface density profile that is systematically shallower than in models with no dark matter.
△ Less
Submitted 8 June, 2017;
originally announced June 2017.
-
On black hole sub-systems in idealized nuclear star clusters
Authors:
Philip G. Breen,
Douglas C. Heggie
Abstract:
Recent observational evidence, numerical simulations and theoretical arguments seem to indicate that stellar mass black holes (BH) persist in stellar systems such as globular star clusters for much longer than previously thought. Previously, theory implied that the BH would segregate into a compact system with short dynamical time scales, and that the BH would escape long before the present. But s…
▽ More
Recent observational evidence, numerical simulations and theoretical arguments seem to indicate that stellar mass black holes (BH) persist in stellar systems such as globular star clusters for much longer than previously thought. Previously, theory implied that the BH would segregate into a compact system with short dynamical time scales, and that the BH would escape long before the present. But stellar systems can exist in a state of balanced evolution, where the energy generated in the core is regulated by the process of two-body relaxation in the bulk of the system. If the system has a centrally concentrated BH subsystem and there is no massive central BH, then the energy is generated by three-body encounters in the core of this subsystem. Therefore, in balanced evolution, the evolution of the BH subsystem is regulated by the much longer time scales of the host system. In the present paper the implications of these results for idealized nuclear star clusters (NSC) are discussed. Though previous theory implied that BH would be almost absent from many NSC -- those with relatively short dynamical time scales -- it is argued here that, based on the results of idealized models, many such NSC could still be host to substantial BH subsystems.
△ Less
Submitted 21 August, 2013;
originally announced August 2013.
-
Dynamical evolution of black hole sub-systems in idealised star clusters
Authors:
Philip G. Breen,
Douglas C. Heggie
Abstract:
In this paper, globular star clusters which contain a sub-system of stellar-mass black holes (BH) are investigated. This is done by considering two-component models, as these are the simplest approximation of more realistic multi-mass systems, where one component represents the BH population and the other represents all the other stars. These systems are found to undergo a long phase of evolution…
▽ More
In this paper, globular star clusters which contain a sub-system of stellar-mass black holes (BH) are investigated. This is done by considering two-component models, as these are the simplest approximation of more realistic multi-mass systems, where one component represents the BH population and the other represents all the other stars. These systems are found to undergo a long phase of evolution where the centre of the system is dominated by a dense BH sub-system. After mass segregation has driven most of the BH into a compact sub-system, the evolution of the BH sub-system is found to be influenced by the cluster in which it is contained. The BH sub-system evolves in such a way as to satisfy the energy demands of the whole cluster, just as the core of a one component system must satisfy the energy demands of the whole cluster. The BH sub-system is found to exist for a significant amount of time. It takes approximately 10t_{rh,i}, where t_{rh,i} is the initial half-mass relaxation time, from the formation of the compact BH sub-system up until the time when 90% of the sub-system total mass is lost (which is of order 10^{3} times the half-mass relaxation time of the BH sub-system at its time of formation). Based on theoretical arguments the rate of mass loss from the BH sub-system (\dot{M}_2) is predicted to be -(beta*zeta*M)/(alpha*t_{rh}), where M is the total mass, t_{rh} is the half-mass relaxation time, and alpha, beta, zeta are three dimensionless parameters (see Section 2 for details). An interesting consequence of this is that the rate of mass loss from the BH sub-system is approximately independent of the stellar mass ratio (m_2/m_1) and the total mass ratio (M_2/M_1) (in the range m_2/m_1 >~ 10 and M_2/M_1 ~ 10^{-2}, where m_1, m_2 are the masses of individual low-mass and high-mass particles respectively, and M_1, M_2 are the corresponding total masses).
△ Less
Submitted 11 April, 2013;
originally announced April 2013.
-
Gravothermal oscillations in multi-component models of star clusters
Authors:
Philip G. Breen,
Douglas C. Heggie
Abstract:
In this paper, gravothermal oscillations are investigated in multi-component star clusters which have power law initial mass functions (IMF). For the power law IMFs, the minimum masses ($m_{min}$) were fixed and three different maximum stellar masses ($m_{max}$) were used along with different power-law exponents ($α$) ranging from 0 to -2.35 (Salpeter). The critical number of stars at which gravot…
▽ More
In this paper, gravothermal oscillations are investigated in multi-component star clusters which have power law initial mass functions (IMF). For the power law IMFs, the minimum masses ($m_{min}$) were fixed and three different maximum stellar masses ($m_{max}$) were used along with different power-law exponents ($α$) ranging from 0 to -2.35 (Salpeter). The critical number of stars at which gravothermal oscillations first appear with increasing $N$ was found using the multi-component gas code SPEDI. The total mass ($M_{tot}$) is seen to give an approximate stability condition for power law IMFs with fixed values of $m_{max}$ and $m_{min}$ independent of $α$. The value $M_{tot}/m_{max} \simeq 12000$ is shown to give an approximate stability condition which is also independent of $m_{max}$, though the critical value is somewhat higher for the steepest IMF that was studied. For appropriately chosen cases, direct N-body runs were carried out in order to check the results obtained from SPEDI. Finally, evidence of the gravothermal nature of the oscillations found in the N-body runs is presented.
△ Less
Submitted 11 July, 2012;
originally announced July 2012.
-
Gravothermal oscillations in two-component models of star clusters
Authors:
Philip G. Breen,
Douglas C. Heggie
Abstract:
In this paper, gravothermal oscillations are investigated in two-component clusters with a range of different stellar mass ratios and total component mass ratios. The critical number of stars at which gravothermal oscillations first appeared is found using a gas code. The nature of the oscillations is investigated and it is shown that the oscillations can be understood by focusing on the behaviour…
▽ More
In this paper, gravothermal oscillations are investigated in two-component clusters with a range of different stellar mass ratios and total component mass ratios. The critical number of stars at which gravothermal oscillations first appeared is found using a gas code. The nature of the oscillations is investigated and it is shown that the oscillations can be understood by focusing on the behaviour of the heavier component, because of mass segregation. It is argued that, during each oscillation, the re-collapse of the cluster begins at larger radii while the core is still expanding. This re-collapse can halt and reverse a gravothermally driven expansion. This material outside the core contracts because it is losing energy both to the cool expanding core and to the material at larger radii. The core collapse times for each model are also found and discussed. For an appropriately chosen case, direct N -body runs were carried out, in order to check the results obtained from the gas model, including evidence of the gravothermal nature of the oscillations and the temperature inversion that drives the expansion.
△ Less
Submitted 24 October, 2011;
originally announced October 2011.
