Riemann solvers and Alfven waves in black hole magnetospheres
- Brian Punsly^{1, 2}Email authorView ORCID ID profile,
- Dinshaw Balsara^{3},
- Jinho Kim^{3} and
- Sudip Garain^{3}
DOI: 10.1186/s40668-016-0018-1
© Punsly et al. 2016
Received: 23 April 2016
Accepted: 17 August 2016
Published: 13 September 2016
Abstract
In the magnetosphere of a rotating black hole, an inner Alfven critical surface (IACS) must be crossed by inflowing plasma. Inside the IACS, Alfven waves are inward directed toward the black hole. The majority of the proper volume of the active region of spacetime (the ergosphere) is inside of the IACS. The charge and the totally transverse momentum flux (the momentum flux transverse to both the wave normal and the unperturbed magnetic field) are both determined exclusively by the Alfven polarization. Thus, it is important for numerical simulations of black hole magnetospheres to minimize the dissipation of Alfven waves. Elements of the dissipated wave emerge in adjacent cells regardless of the IACS, there is no mechanism to prevent Alfvenic information from crossing outward. Thus, numerical dissipation can affect how simulated magnetospheres attain the substantial Goldreich-Julian charge density associated with the rotating magnetic field. In order to help minimize dissipation of Alfven waves in relativistic numerical simulations we have formulated a one-dimensional Riemann solver, called HLLI, which incorporates the Alfven discontinuity and the contact discontinuity. We have also formulated a multidimensional Riemann solver, called MuSIC, that enables low dissipation propagation of Alfven waves in multiple dimensions. The importance of higher order schemes in lowering the numerical dissipation of Alfven waves is also catalogued.
Keywords
black hole physics magnetohydrodynamics (MHD) galaxies: jets galaxies: active accretion accretion disks1 Introduction
In this century, there has been great progress in 3-D magnetohydrodynamic (MHD) simulations of black hole magnetospheres (De Villiers et al. 2003; Komissarov 2004; Komissarov 2005; Hawley and Krolik 2006; Fragile et al. 2007; McKinney and Blandford 2009; McKinney et al. 2012). To varying degrees, each of these simulations require knowledge of the 1-D characteristics of the MHD system in order to time evolve the magnetosphere. Specifically, the polarization properties of the waves determine the changes in the fields that can be propagated at the appropriate speed along a particular characteristic direction. In single fluid ideal MHD there are three plasma modes in the system, the fast mode, the Alfven (or intermediate) mode and the slow mode. Black hole magnetospheres have the property that all plasma must pass progressively through the slow, Alfven and fast critical surfaces before reaching the event horizon (Punsly 2008). As each critical surface is crossed, the unique information associated with each wave mode is unable to be communicated upstream to an outgoing wind or jet. The event horizon wind system has no boundary conditions at its terminus, there are asymptotic infinities both at the event horizon and at large radial coordinate (Punsly 2008). There are only lateral boundary conditions imposed by accreting gas. Thus, the wind system itself and the lateral boundary conditions determine 3-D single fluid perfect MHD wind solutions. Furthermore, due to the paired wind nature of the event horizon wind system (an ingoing accretion inner wind and the outgoing outer wind or jet), plasma is always drained off of the field lines and auxiliary physics (mass floors) must be injected by hand in order to keep numerical simulations from crashing at low density. Mass floors are a source of MHD waves and are generally chosen to enhance dissipation. Consequently, there are many unique aspects to the application of MHD that can influence the final steady state of the wind system. Describing the evolution of the event horizon magnetosphere with single fluid MHD is wrought with non trivial subtleties.
These subtleties relate to the numerical determination of the field line angular velocity as viewed from asymptotic infinity, \(\varOmega _{F}\). This is of primary interest since the Poynting flux of the wind scales as \(\varOmega _{F}^{2}\) (Punsly 2008). First, contrary to previous claims of early simulations, newer simulations indicate that \(\varOmega _{F}\) can be altered significantly by the auxiliary method of injecting plasma (McKinney et al. 2012; Beskin and Zheltoukhov 2103). We consider the unique role of the oblique Alfven wave in this process. A unique component of the momentum flux is propagated along the Alfven characteristics and this momentum flux is a component of the MHD equations written in conservative form. It is also the only isolated discontinuity that propagates a physical charge. Black hole magnetospheres that support an outgoing relativistic jet, rotate and have a Goldreich-Julian charge density. The Alfven critical surface for the inflow (IACS) is quite far from the event horizon. For rapidly rotating black holes and the most recent \(\varOmega _{F}\) values from numerical simulations, the proper distance is 2 to 3 black hole radii from the event horizon. For proper evolution of the magnetic field rotation rate and the induced charge density, one must be able to simulate the role of Alfven waves with high fidelity both globally and inside of the IACS. Thusly motivated, we discuss in this article new numerical methods that are designed to accurately characterize the Alfven wave numerically in the rarefied environment of black hole magnetosphere.
An accurate depiction of the time evolution of a black hole magnetosphere and the global considerations germane to the IACS are intimately related to minimizing the numerical dissipation of Alfven waves. The IACS is a one-way surface as far as t he propagation of Alfven waves is concerned. In other words, at the IACS and within it, all Alfven waves should propagate inwards and only inwards. The propagation of waves in a higher order Godunov code is modulated by the Riemann solver. It is, therefore desirable if the Riemann solver can mimic this one-way propagation property for Alfven waves. Alas, whether a Riemann solver does so or not, depends crucially on the design of the Riemann solver. Some Riemann solvers which retain the substructure associated with Alfven waves within the Riemann fan, can indeed represent such a one-way propagation of Alfven waves. Other Riemann solvers, like the HLL Riemann solver, do not retain the substructure associated with the Alfven waves. At extraordinarily high resolutions, any well-designed code will of course minimize this dissipation. However, the present generation of simulations have all been done with low or modest resolutions. Furthermore, they have mostly used the HLL Riemann solver which, we argue, applies a maximal, and deleterious, dissipation to Alfven waves. To appreciate this point, realize that the HLL Riemann solver is based on a wave model that has just two extremal waves. These two extremal waves determine the ends of the Riemann fan in one-dimension. The speed of these extremal waves is usually set to the extremal signal speeds in the physical problem. For a relativistic MHD (RMHD) simulation of highly magnetized event horizon magnetospheres, these extremal signal speeds are usually set to approximately the speed of light propagating in either direction at a zone interface where the Riemann solver is applied. The HLL Riemann solver does not incorporate any further sub-structure associated with intermediate waves. Consequently, the HLL Riemann solver maximizes the dissipation of Alfven waves even near the IACS. This is the very location where the dissipation of these waves has to be minimized. Introducing the intermediate waves in the Riemann fan reduces the dissipation, but that effect is not incorporated in the HLL Riemann solver.
There is another issue that increases the dissipation of Alfven waves, in more than one dimension. Riemann solvers applied to black hole magnetospheres have been treated as 1-D in each direction. However, a true multi-dimensional Riemann solver has a strongly interacting region in which the numerical fluxes in orthogonal directions become intertwined (Balsara 2012). Careful treatment of the strongly-interacting region results in far less dissipation (as we show for RMHD in Section 4). The increase in computational complexity associated with a multidimensional Riemann solver is handily offset by larger timesteps and greater code robustness. The accuracy of the numerical depiction of the role of the Alfven waves near a black hole during jet production is facilitated by a true multidimensional scheme that incorporates the strongly interacting region.
Our ultimate goal is to understand the detailed time evolution of black hole magnetospheres. This is subtle because one must try to understand in each time step how the mass floor is affecting the time evolution. Thus, the transient structure is essential to monitor in order to see how transients associated with plasma injection are altering the time evolution of the system. This requires an inherently very stable numerical scheme (‘well-balanced’, which we describe below). Furthermore, we note that the simulations of Komissarov (2004) and Komissarov (2005) did tend to a steady state and this made it possible to carry out convergence testing for those simulations. However, several simulations seemed to never reach a steady state and, therefore, one cannot carry out convergence studies for them. In particular, an unexpected finding of Krolik et al. (2005) was that the event horizon magnetospheres in numerical simulations are very unsteady and appear to be more like a cauldron of strong MHD waves rather than a force-free structure. ‘For example, although the funnel region is magnetically-dominated, it is not in general in a state of force-free equilibrium. Indeed, the very large fluctuations that continually occur in the outflow show that it is never in any state of equilibrium, force-free or otherwise.’ This also appears to be the case in the simulations of McKinney et al. (2012) based on the supporting online movies. This suggests that wind formation in event horizon magnetospheres might be subject to large numerically induced transients which would mask the kinds of effects that we are looking for. Alternatively, if these large transients are integral to jet formation it indicates a dynamic in which strong waves from the lateral boundaries created by the accretion flow scatter off the event horizon magnetosphere producing strong gusts of energy in a jetted outflow.
It is important to separate these potential physical effects from numerical effects. However, schemes that are based on higher order Godunov methods, especially those that are based on the HLL Riemann solver, are notorious for not achieving steady state even when the physical problem admits such a steady state! This was first observed when higher order Godunov schemes with Riemann solvers were first applied to metreological simulations (which simulate wind flow in the earth’s atmosphere) or to the shallow water equations (which simulate lake and ocean circulation) (Parés 2006). Unless the numerical scheme has a special property called well-balancing, it usually does not find a stationary state even when one exists. Instead, the simulation generates ‘numerical storms’ - high velocity flows that are entirely a numerical artefact. The issue of well-balancing has recently become more compelling within the context of astrophysical flows with the work of Kappeli and Mishra (2014, 2016). Within the context of Type IIb supernova simulations, it has been found that the proto-neutron stars refuse to reach steady state if the scheme is not well-balanced, i.e., the numerical method has to be specially engineered to that it can find a steady-state proto-neutron star solution if one exists. Kappeli and Mishra have also explored time-dependent simulations that are not in steady state. Their very interesting result is that even for simulations that have no reason to be in steady state, the inclusion of well-balancing provides a substantial improvement in the accuracy of the simulation. This result has bearing on the black hole magnetospheres problem because it suggests that even when the simulations are far from steady state they might indeed be helped by well-balancing.
In order to verify that one has a well-balanced scheme, one has to know what the steady states of the system are and make sure that the simulated system is driven towards that steady state. This does not mean that every system must reach a steady state; it only means that when such a steady state exists, the numerical code is equipped to find it. For a scheme to be truly well-balanced, it needs to have two attributes. First, the reconstruction procedure has to be modified so that any potential steady state solution is subtracted off from the reconstructed solution. Second, the scheme must use a Riemann solver that can capture intermediate waves - specifically the contact discontinuity in classical Euler flow. We admit, that in certain circumstances it may not be possible to identify the steady state solution that has to be subtracted off. In such situations, the numerical scheme should at least be well-balanced up to second order. In practical terms, this means that a modest resolution simulation will not find the steady state solution to machine accuracy. However, it will nevertheless find the steady state solution with accuracies that are proportional to the size of the mesh squared. Being well-balanced up to second order is a weaker notion of well-balancing compared to being truly well-balanced. In order for a scheme to be well-balanced up to second order, it is imperative that the Riemann solver should at least capture stationary contact discontinuities in a self-gravitating situation involving Euler flow. By extension, any MHD code that is capable of capturing stationary equilibria up to second order should at least be based on a Riemann solver that captures contact discontinuities and Alfven waves. The discussion in this paragraph has made it clear that in order to disentangle a flow that is chaotic because of the physics of the situation from a flow that is chaotic because of simple numerical effects, one must pay attention to the form of the Riemann solver.
In summary, for the purposes of exploring the time evolution of event horizon magnetospheres, we require a well-balanced scheme to second order that will eliminate recurrent large transients. This is facilitated by preserving the Alfven wave as discussed above, but also requires that the inclusion of the contact discontinuity in the Riemann fan. Furthermore, without the contact discontinuity, the huge density gradient between accretion flow and the evacuated funnel of the event horizon magnetosphere means that plasma flows into the funnel by numerical diffusion. This is far from ideal if we want to explore the role of mass injection on the final solutions. To improve this situation, one desires a Riemann solver that at least minimizes the dissipation at contact discontinuities. (For more helpful information on numerical schemes and Riemann solvers of relevance to astrophysics, please visit http://www.nd.edu/~dbalsara/Numerical-PDE-Course.)
Designing low-dissipation Riemann solvers for RMHD is a challenging enterprise. An exact RMHD Riemann solver exists (Giacomazzo and Rezzolla 2006; Giacomazzo and Rezzolla 2007), but it is not practicable for use in numerical codes. HLLC Riemann solvers for RMHD exist (Mignone and Bodo 2006; Honkkila and Janhunen 2007; Kim and Balsara 2014). They enable a stationary contact discontinuity to be captured on a mesh. However, they dissipate stationary Alfven waves just like an HLL Riemann solver. HLLD Riemann solvers for RMHD (Mignone et al. 2009), do exist, which enable stationary contact discontinuities as well as stationary Alfvenic discontinuities to be captured on a computational mesh. Unfortunately, the method is iterative, which makes it computationally very expensive. Furthermore, when an iteration fails to converge, the method becomes brittle. With the emergence of the HLLI Riemann solver (Dumbser and Balsara 2016), it has become possible to capture stationary contact discontinuities as well as stationary Alfvenic discontinuities using a Riemann solver that is non-iterative and computationally inexpensive. The first goal of this paper is to document this capability in the astrophysical literature.
RMHD simulations also have to maintain the divergence-free structure of the magnetic field. This necessitates the use of a Yee-type mesh where the magnetic field components are specified at the faces of the mesh and the electric fields are to be evaluated at the edges of the mesh. It was claimed by Gardiner and Stone (2005), Gardiner and Stone (2008), Beckwith and Stone (2011), White and Stone (2015) that stabilizing the evolution of the magnetic field requires that one should always double the dissipation in the electric field at every timestep. Unfortunately, that approach has been used in the RMHD literature with the result that the already excessive dissipation of the HLL Riemann solver is increased even further in simulations. Such explorations ignore recent advances in multidimensional Riemann solver technology (Balsara 2010; Balsara 2012; Balsara 2014; Balsara 2015; Balsara et al. 2014; Balsara and Dumbser 2015; Balsara et al. 2016a). In a recent paper, Balsara and Kim (2016), showed that an exact analogue of the HLLI Riemann solver in multidimensions can be designed for RMHD. Their work is based on the original paper by Balsara et al. (2016b). Such multidimensional Riemann solvers go under the name of MuSIC Riemann solvers. Here the MuSIC acronym stands for Multidimensional, Self-similar strongly-Interacting Riemann solver that is based on Consistency with the conservation law. By introducing substructure associated with the multidimensional propagation of Alfven waves, the MuSIC Riemann solver reduces the dissipation of Alfven waves that propagate at any angle with respect to the mesh. The second goal of this paper is to catalogue the advantages of the MuSIC Riemann solver in reducing the dissipation involved in the multidimensional propagation of Alfven waves in RMHD.
In this article, we claim that modern 5-wave Riemann solvers can now be implemented that can allow a systematic assessment of the issues related to determination of \(\varOmega _{F}\). This would require the formulation and simulation of simple magnetospheres, lateral boundary conditions and plasma injection mechanisms. Proper time evolution of the 3-D magnetosphere requires two main aspects of the solver, low dissipation and well balancing. We have motivated both these issues in the previous paragraphs. We demonstrate that the new Riemann solvers described here are capable of delivering on these goals in the following.
Riemann solvers offer one way of reducing numerical dissipation and Riemann solvers that preserve essential features of the flow are certainly central to many aspects of jet simulation. Recent advances in higher order schemes has made it possible to go beyond the traditional second order Godunov scheme that is commonplace in computational astrophysics. The third goal of this paper is to show that higher order schemes for RMHD do exist (Balsara and Kim 2016; Del Zanna et al. 2007; Zanotti and Dumbser 2016), which make it possible to go beyond second order of accuracy. We show that the combination of higher order schemes and appropriate Riemann solvers can go a long way towards enabling almost dissipation-free propagation of Alfven waves.
The paper is organized as follows. In Section 2, we discuss dissipation inside the IACS in Riemann solvers in general terms. It is shown that the HLLI Riemann solver provides the theoretical minimum dissipation that is consistent with a stable numerical scheme. In Section 3, we demonstrate by explicit examples that the HLLI RMHD Riemann solver preserves the Alfven wave with high accuracy and respects the contact discontinuity. In Section 4, we incorporate the important aspects of multi-dimensionality with the MuSIC RMHD Riemann solver. This is truly a multi-dimensional scheme and we demonstrate that its ability to resolve the strongly interacting region substantially reduces Alfvenic dissipation compared to higher dimensional schemes that utilize 1-D Riemann solvers in each direction. In our final discussion section, we describe how a numerical scheme that utilizes the MuSIC Riemann solver would be suitable for specialized simulations that would shed light on the causal physics of the time evolution of event horizon magnetospheres and help define the full panoply of physically allowed and disallowed solutions.
2 Dissipation in Riemann solvers
In this section, we have only given a flavor of the dissipation characteristics of the HLLI Riemann solver and how it offers a substantial improvement over the HLL Riemann solver. The reader who is interested in details should please read Dumbser and Balsara (2016). The eigenvectors for RMHD that were used in this paper can all be obtained from Balsara (2001) or Antón et al. (2010).
3 One dimensional Riemann solvers in RMHD
The Introduction has shown that it is very desirable to have Riemann solvers that can capture stationary, isolated contact discontinuities as well as stationary, isolated Alfven waves. Indeed the first of the one-dimensional Riemann solvers for RMHD by Komissarov (1999) and Balsara (2001) were Roe-type Riemann solvers. Because such Riemann solvers retain the entire set of eigenvectors for the RMHD system, they can indeed capture stationary, isolated contact discontinuities as well as stationary, isolated Alfven waves on a mesh. There has been a recent effort by Antón et al. (2010) to revive the use of Roe-type Riemann solvers in RMHD simulations, but the effort has met with limited success owing to the exorbitant computational cost of such Riemann solvers, especially for RMHD. The Roe-type Riemann solvers also have an inherent deficiency. This has to do with their loss of positivity of density and pressure in certain circumstances (Einfeldt 1988; Einfeldt et al. 1991). It is easy to find mentions in the early literature on RMHD (Komissarov 1999, 2004, 2005), showing that the early RMHD simulation codes struggled with positivity issues and, therefore, reverted to the use of the HLL Riemann solver. This issue is relevant to the extremely low density environment of event horizon magnetospheres.
HLLD Riemann solvers. Mignone et al. (2009), also enable a code to capture stationary, isolated contact discontinuities as well as stationary, isolated Alfven waves. But they have their own set of attendant problems, as discussed in the Introduction. HLLC Riemann solvers (Mignone and Bodo 2006; Honkkila and Janhunen 2007; Kim and Balsara 2014), represent a compromise position where they enable a code to capture stationary, isolated contact discontinuities, but not Alfven waves. The HLLI Riemann solver of Dumbser and Balsara (2016) is built on top of an HLL Riemann solver, so it inherits all the beneficial positivity properties of the HLL Riemann solver. However, it introduces sub-structure in the Riemann fan. Typically, that substructure includes the contribution from eigenvectors of the contact discontinuity and eigenvectors associated with Alfven waves. The eigenvectors for the fast and slow magnetosonic waves are very expensive to evaluate computationally, and their evaluation is avoided in the HLLI Riemann solver. As a result, the HLLI Riemann solver enables a code to capture stationary, isolated contact discontinuities as well as stationary, isolated Alfven waves at a very low computational cost. Unlike HLLC and HLLD, the HLLI Riemann solver does not require an iterative solution, thereby ensuring that it has even lower computational cost. In the next few paragraphs we demonstrate this facet of the HLLI Riemann solver for RMHD.
4 Multidimensional propagation of Alfven waves
Figures 2 and 3 showed that the one-dimensional HLLI Riemann solver can dramatically reduce the dissipation compared to the HLL Riemann solver. Multidimensional treatment of Alfven waves on a computational mesh requires a multidimensional Riemann solver. In Balsara (2004), we were able to formulate a test problem that measures the ability of a multidimensional MHD code to propagate Alfven waves with the least amount of dissipation. It has been shown that this test problem is very important in benchmarking the dissipation characteristics of multidimensional codes for classical MHD. The analogous problem which benchmarks the low dissipation propagation of Alfven waves in RMHD has been recently presented in Balsara and Kim (2016). It consists of torsional Alfven waves propagating obliquely to the mesh lines of a two-dimensional mesh. The mesh has \(120\times 120\) zones. We do not repeat the problem description. Instead, we show the results and intercompare with older methods that involve dissipation doubling from Gardiner and Stone (2005, 2008).
\(\pmb{L_{1}}\) error in the z -momentum density and the z -component of the magnetic field as a function of mesh resolution at the second order
Zones | \(\boldsymbol{L_{1}}\) error z -momentum density | Accuracy | \(\boldsymbol{L_{1}}\) error z -magnetic field | Accuracy |
---|---|---|---|---|
60 × 60 | 2.7693 × 10^{−1} | 1.1971 × 10^{−1} | ||
120 × 120 | 7.5660 × 10^{−2} | 1.87 | 3.3246 × 10^{−2} | 1.85 |
240 × 240 | 1.6083 × 10^{−2} | 2.23 | 7.1989 × 10^{−3} | 2.21 |
\(\pmb{L_{1}}\) error in the z -momentum density and the z -component of the magnetic field as a function of mesh resolution at the third order
Zones | \(\boldsymbol{L_{1}}\) error z -momentum density | Accuracy | \(\boldsymbol{L_{1}}\) error z -magnetic field | Accuracy |
---|---|---|---|---|
60 × 60 | 2.1295 × 10^{−1} | 9.2981 × 10^{−2} | ||
120 × 120 | 3.7593 × 10^{−2} | 2.50 | 1.6720 × 10^{−2} | 2.48 |
240 × 240 | 5.1524 × 10^{−3} | 2.87 | 2.2025 × 10^{−3} | 2.92 |
5 Discussion and future prospects
In this article, we describe subtle points uniquely associated with numerical simulations of event horizon magnetospheres. All inflows must pass through an IACS, thereby rendering the interior region out of Alfven wave communication with the out-flowing wind. Thus, the IACS is a charge horizon (a one-way membrane) for the global flow since the Alfven wave is the only wave that carries a physical charge (Punsly 2004). The IACS causally excises most of the active region of space-time (the ergosphere) that would be the putative element to enforce rapid rotation of the system. The implication is that simulations of event horizon magnetospheres should be performed with numerical schemes that represent the Alfvenic properties of the system with high fidelity.
It was demonstrated that an improvement of numerical accuracy can be attained by utilizing numerical schemes based on HLLI and MuSIC Riemann solvers rather than schemes based on 1-D HLL solvers. In particular, for 1-D RMHD solvers, the HLLI Riemann solver provides the theoretical minimum dissipation of the Alfven wave and contact discontinuities that can still ensure a numerically stable scheme. In Section 4, we discussed the multidimensional extension of HLLI RMHD Riemann solver; the RMHD MuSIC Riemann solver. It is in higher dimensions that we see an even larger improvement over schemes based on 1-D HLL solvers. The MuSIC Riemann solver was shown to significantly reduce the dissipation of Alfven waves in large part to its ability to resolve the strongly interacting region that is typically ignored in schemes based on 1-D HLL Riemann solvers. We also show that very high order schemes might be free of the excessive dissipation that arises from a lower quality Riemann solver. However, that transition occurs only when schemes of fourth and higher order of accuracy are used.
- 1.
At the risk of being repetitive, the paired wind systems evolves outward and inward towards two asymptotic infinities as opposed to having a causal MHD boundary at one terminus. Thus, unlike other MHD wind problems there is a more complex critical point structure. Most specifically, all inflows pass through the inner Alfven critical surface. Thereby causally disconnecting the outflowing wind from Alfven radiation emanating from the majority of the rapidly rotating ergospheric plasma. The Alfven wave is the only wave that propagates a physical charge and thus should be involved in the establishment of the Goldreich-Julian charge density or equivalently, the field line rotation rate, \(\varOmega _{F}\). Thus, reducing the numerical dissipation of the Alfven wave by implementing the HLLI or MuSIC Riemann solvers would seem to help in this regard.
- 2.
The paired wind system constantly drains itself of plasma in the MHD limit. Thus, plasma injection by means of a mass floor is required. This process will dissipate MHD waves generated by the MHD system and inject new MHD waves. The process is nontrivial and has been shown to modify \(\varOmega _{F}\) significantly in certain 3-D simulations. Since, in principle, it can modify \(\varOmega _{F}\) and mass injection perturbs the Alfven waves generated in the system, a Riemann solver lowers Alfven dissipation might shed light on the nature of the transients that occur as different plasma injection scenarios are explored.
- 3.
Another related issue is the large numerical diffusion of plasma from the bounding accretion disk into the event horizon magnetosphere. Diffusion is substantially reduced if the Riemann solver respects the contact discontinuity. Since the source of plasma injection might be important to the establishment of \(\varOmega _{F}\) this is an important issue as well. Riemann solvers, such as the HLLI or MUSIC Riemann solvers, which treat contact discontinuities explicitly can help in this regard.
In this article, we presented both theoretical and numerical arguments that support the notion that numerical schemes based on the multi-dimensional MuSIC Riemann solver can potentially provide an improvement over existing methods of modeling 3-D event horizon magnetospheres. In particular, the MuSIC Riemann solver is both well suited for the low density environment endemic to the event horizon magnetosphere and it provides reduced dissipation of the Alfven and contact discontinuities. It is also computationally efficient which offsets the cost of improving the computational accuracy.
In Balsara and Kim (2016) we provide a subluminal scheme for RMHD as well as a discussion of the MuSIC Riemann solver. The reader who wishes to get further information can also visit the second author’s website http://www.nd.edu/~dbalsara/Numerical-PDE-Course.
Declarations
Acknowledgements
DB acknowledges support via NSF grants NSF-ACI-1307369, NSF-DMS-1361197 and NSF-ACI-1533850. DB also acknowledges support via NASA grant NASA-NNX 12A088G. Several simulations were performed on a cluster at UND that is run by the Center for Research Computing. BP thanks ICRANet for support.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Authors’ Affiliations
References
- Antón, L, Miralles, JA, Martí, JM, et al.: Relativistic magnetohydrodynamics: renormalized eigenvectors and full wave decomposition Riemann solver. Astrophys. J. Suppl. Ser. 188, 1-31 (2010) ADSView ArticleGoogle Scholar
- Balsara, D: Total variation diminishing scheme for relativistic magnetohydrodynamics. Astrophys. J. Suppl. Ser. 132, 83-101 (2001) ADSView ArticleGoogle Scholar
- Balsara, DS: Schemes for magnetohydrodynamics with divergence-free reconstruction. Astrophys. J. Suppl. Ser. 151, 149-184 (2004) ADSView ArticleGoogle Scholar
- Balsara, DS: Multidimensional extension of the HLLE Riemann solver; application to Euler and magnetohydrodynamical flows. J. Comput. Phys. 229, 1970-1983 (2010) ADSMathSciNetView ArticleMATHGoogle Scholar
- Balsara, DS: A two-dimensional HLLC Riemann solver for conservation laws: application to Euler and magnetohydrodynamic flows. J. Comput. Phys. 231, 7476-7503 (2012) ADSMathSciNetView ArticleMATHGoogle Scholar
- Balsara, DS: Multidimensional Riemann problem with self-similar internal structure. Part I - application to hyperbolic conservation laws on structured meshes. J. Comput. Phys. 277, 163-200 (2014) ADSMathSciNetView ArticleGoogle Scholar
- Balsara, DS: Three dimensional HLL Riemann solver for conservation laws on structured meshes; application to Euler and magnetohydrodynamic flows. J. Comput. Phys. 295, 1-23 (2015) ADSMathSciNetView ArticleGoogle Scholar
- Balsara, DS, Dumbser, M: Multidimensional Riemann problem with self-similar internal structure. Part II - application to hyperbolic conservation laws on unstructured meshes. J. Comput. Phys. 287, 269-292 (2015) ADSMathSciNetView ArticleGoogle Scholar
- Balsara, DS, Dumbser, M, Abgrall, R: Multidimensional HLL and HLLC Riemann solvers for unstructured meshes - with application to Euler and MHD flows. J. Comput. Phys. 261, 172-208 (2014) ADSMathSciNetView ArticleGoogle Scholar
- Balsara, DS, Kim, J: A subluminal relativistic magnetohydrodynamics scheme with ADER-WENO predictor and multidimensional Riemann solver-based corrector. J. Comput. Phys. 312, 357-384 (2016) ADSMathSciNetView ArticleGoogle Scholar
- Balsara, DS, Nkonga, B, Dumbser, M, Munz, CD: (2016b, in preparation)
- Balsara, DS, Vides, J, Gurski, K, et al.: A two-dimensional Riemann solver with self-similar sub-structure - alternative formulation based on least squares projection. J. Comput. Phys. 304, 138-161 (2016a) ADSMathSciNetView ArticleGoogle Scholar
- Beckwith, K, Stone, JM: A second-order Godunov method for multi-dimensional relativistic magnetohydrodynamics. Astrophys. J. Suppl. Ser. 193, 6-35 (2011) ADSView ArticleGoogle Scholar
- Beskin, VS, Zheltoukhov, AA: On the structure of the magnetic field near a black hole in active galactic nuclei. Astron. Lett. 39, 215-220 (2103) View ArticleGoogle Scholar
- De Villiers, J-P, Hawley, JF, Krolik, JH: Magnetically driven accretion flows in the Kerr metric. I. Models and overall structure. Astrophys. J. 599, 1238-1253 (2003) ADSView ArticleGoogle Scholar
- Del Zanna, L, Zanotti, O, Bucciantini, N, Londrillo, P: ECHO: a Eulerian conservative high-order scheme for general relativistic magnetohydrodynamics and magnetodynamics. Astron. Astrophys. 473, 11-30 (2007) ADSView ArticleGoogle Scholar
- Dumbser, M, Balsara, D: A new efficient formulation of the HLLEM Riemann solver for general conservative and non-conservative hyperbolic systems. J. Comput. Phys. 304, 275-319 (2016) ADSMathSciNetView ArticleGoogle Scholar
- Einfeldt, B: On Godunov-type methods for gas dynamics. SIAM J. Numer. Anal. 25, 294-318 (1988) ADSMathSciNetView ArticleMATHGoogle Scholar
- Einfeldt, B, Munz, C-D, Roe, P, Sjogreen, B: On Godunov-type methods near low densities. J. Comput. Phys. 92, 273-295 (1991) ADSMathSciNetView ArticleMATHGoogle Scholar
- Etienne, Z, Paschalidis, V, Haas, R, Mösta, P, Shapiro, S: IllinoisGRMHD: an open-source, user-friendly GRMHD code for dynamical spacetimes. Class. Quantum Gravity 32, 175009 (2015) ADSMathSciNetView ArticleMATHGoogle Scholar
- Fragile, PC, Blaes, OM, Anninos, P, Salmonson, JD: Global general relativistic magnetohydrodynamic simulation of a tilted black hole accretion disk. Astrophys. J. 668, 417-429 (2007) ADSView ArticleGoogle Scholar
- Gammie, CF, McKinney, JC, Toth, G: A numerical scheme for general relativistic magnetohydrodynamics. Astrophys. J. 589, 444-457 (2003) ADSView ArticleGoogle Scholar
- Gardiner, TA, Stone, JM: An unsplit Godunov method for ideal MHD via constrained transport. J. Comput. Phys. 205, 509-539 (2005) ADSMathSciNetView ArticleMATHGoogle Scholar
- Gardiner, TA, Stone, JM: An unsplit Godunov method for ideal MHD via constrained transport in three dimensions. J. Comput. Phys. 227, 4123-4141 (2008) ADSMathSciNetView ArticleMATHGoogle Scholar
- Giacomazzo, B, Rezzolla, L: The exact solution of the Riemann problem in relativistic magnetohydrodynamics. J. Fluid Mech. 562, 223-259 (2006) ADSMathSciNetView ArticleMATHGoogle Scholar
- Giacomazzo, B, Rezzolla, L: WhiskyMHD: a new numerical code for general relativistic magnetohydrodynamics. Class. Quantum Gravity 24(12), S235-S258 (2007) ADSMathSciNetView ArticleMATHGoogle Scholar
- Hawley, J, Krolik, K: Magnetically driven jets in the Kerr metric. Astrophys. J. 641, 103-116 (2006) ADSView ArticleGoogle Scholar
- Honkkila, V, Janhunen, P: HLLC solver for ideal relativistic MHD. J. Comput. Phys. 223, 643-656 (2007) ADSMathSciNetView ArticleMATHGoogle Scholar
- Kappeli, R, Mishra, S: Well-balanced schemes for the Euler equations with gravitation. J. Comput. Phys. 259, 199-219 (2014) ADSMathSciNetView ArticleGoogle Scholar
- Kappeli, R, Mishra, S: A well-balanced finite volume scheme for the Euler equations with gravitation. The exact preservation of hydrostatic equilibrium with arbitrary entropy stratification. Astron. Astrophys. 587, A94-A110 (2016) ADSView ArticleGoogle Scholar
- Kim, J, Balsara, DS: A stable HLLC Riemann solver for relativistic magnetohydrodynamics. J. Comput. Phys. 270, 634-639 (2014) ADSMathSciNetView ArticleGoogle Scholar
- Komissarov, S: A Godunov-type scheme for relativistic magnetohydrodynamics. Mon. Not. R. Astron. Soc. 303, 343-366 (1999) ADSView ArticleGoogle Scholar
- Komissarov, S: General relativistic magnetohydrodynamic simulations of monopole magnetospheres of black holes. Mon. Not. R. Astron. Soc. 350, 1431-1436 (2004) ADSView ArticleGoogle Scholar
- Komissarov, S: Observations of the Blandford-Znajek process and the magnetohydrodynamic Penrose process in computer simulations of black hole magnetospheres. Mon. Not. R. Astron. Soc. 359, 801-808 (2005) ADSView ArticleGoogle Scholar
- Krolik, K, Hawley, J, Hirose, S: Magnetically driven accretion flows in the Kerr metric. IV. Dynamical properties of the inner disk. Astrophys. J. 622, 1008-1023 (2005) ADSView ArticleGoogle Scholar
- McKinney, J, Blandford, R: Stability of relativistic jets from rotating, accreting black holes via fully three-dimensional magnetohydrodynamic simulations. Mon. Not. R. Astron. Soc. Lett. 394, 126-130 (2009) ADSView ArticleGoogle Scholar
- McKinney, J, Tchekhovskoy, A, Blandford, R: General relativistic magnetohydrodynamic simulations of magnetically choked accretion flows around black holes. Mon. Not. R. Astron. Soc. 423, 3083-3117 (2012) ADSView ArticleGoogle Scholar
- Mignone, A, Bodo, G: An HLLC Riemann solver for relativistic flows - II. Magnetohydrodynamics. Mon. Not. R. Astron. Soc. 368, 1040-1054 (2006) ADSView ArticleGoogle Scholar
- Mignone, A, Ugliano, M, Bodo, G: A five-wave Harten-Lax-van Leer Riemann solver for relativistic magnetohydrodynamics. Mon. Not. R. Astron. Soc. 393, 1141-1156 (2009) ADSView ArticleGoogle Scholar
- Parés, C: Numerical methods for nonconservative hyperbolic systems: a theoretical framework. SIAM J. Numer. Anal. 44, 300-321 (2006) MathSciNetView ArticleMATHGoogle Scholar
- Punsly, B: Fast-wave polarization charge horizons, and the time evolution of force-free magnetospheres. Astrophys. J. 612, L41-L44 (2004) ADSView ArticleGoogle Scholar
- Punsly, B: Black Hole Gravitohydromagnetics, 2nd edn. Springer, New York (2008) MATHGoogle Scholar
- White, CJ, Stone, JM: GRMHD in Athena++ using advanced Riemann solvers and staggered-mesh constrained transport (2015). arXiv:1511.00943
- Zanotti, O, Dumbser, M: Efficient conservative ADER schemes based on WENO reconstruction and space-time predictor in primitive variables. Comput. Astrophys. Cosmol. 3, 1 (2016). doi:10.1186/s40668-015-0014-x ADSView ArticleGoogle Scholar