Pith. sign in

REVIEW 3 major objections 5 minor 19 references

Quasi-invertible transformations—variable changes with a one-sided inverse—carry relativistic hydrodynamics' characteristic decomposition into the conserved variables used by codes, simplifying known eigenvectors and adding a composition wa

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-03 21:45 UTC pith:ZI743JUW

load-bearing objection A clean, mostly solid methods paper introducing quasi-invertible transformations and a new NSE composition-dependent characteristic decomposition; the suspected D-term flaw does not survive close reading, but the final algebra needs machine verification and the MHD claims should be tempered. the 3 major comments →

arxiv 2511.13836 v2 pith:ZI743JUW submitted 2025-11-17 gr-qc astro-ph.HE

Characteristic Decomposition for Relativistic Numerical Simulations: I. Hydrodynamics

Saul A. Teukolsky This is my paper
classification gr-qc astro-ph.HE
keywords characteristic decompositionquasi-invertible transformationsrelativistic hydrodynamicsconserved variablesRiemann solverselectron fractionnuclear statistical equilibriumhyperbolic systems
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper sets out to make the characteristic decomposition of relativistic hydrodynamics—the eigenvalues and eigenvectors that describe how waves propagate in the fluid—available in the form numerical codes actually use, namely in conserved variables in an arbitrary coordinate frame. Its central claim is that a new class of non-square variable changes, called quasi-invertible transformations, can transport the decomposition from the comoving frame, where it is simple, to the Eulerian frame and then to conserved variables, with the left eigenvectors obtained through a modified transformation rule. The result is a somewhat simpler derivation of the known decomposition, without computer algebra, plus a genuinely new decomposition when the fluid composition (electron fraction in nuclear statistical equilibrium) is evolved as an independent variable. A sympathetic reader takes this as the foundation for the companion paper that derives the full GRMHD decomposition, which would enable full-wave Riemann solvers and rigorous characteristic boundary conditions in relativistic MHD codes.

Core claim

The paper shows that the transformation from the 4-velocity u^a to the Eulerian 3-velocity v^a is not an invertible variable change—its Jacobian is a 4×3 matrix—but it admits a one-sided inverse. The author defines a quasi-invertible transformation by demanding the inverse satisfy both the natural 3-dimensional condition (6.12) and the matrix 'inverse' condition (6.14), which fixes the arbitrary part of the inverse except for a term that the author asserts never contributes in the applications considered. Under this prescription, the right eigenvectors transform by the Jacobian (5.3), while the left eigenvectors obey the new rule (7.13) involving the old principal-part matrix contracted with

What carries the argument

The central object is the quasi-invertible transformation: a change of variables U_old→U whose Jacobian is rectangular but possesses a one-sided inverse satisfying the pair of relations (6.12) and (6.14) that the author imposes. It does the work of converting the comoving characteristic decomposition into the Eulerian and conservative ones: right eigenvectors transform by multiplication with the Jacobian, and left eigenvectors by the replacement rule (7.13) that incorporates the change of time-like congruence from u^a to n^a. The non-uniqueness of the one-sided inverse is the delicate point; the paper fixes it by setting the arbitrary coefficients B and D to zero.

Load-bearing premise

The load-bearing premise is that the arbitrary part of the one-sided inverse of the 4-velocity-to-3-velocity transformation—specifically the D-term—can be set to zero without changing the final left eigenvectors, a step the paper acknowledges is not guaranteed by any theorem.

What would settle it

Compute the full flux Jacobian numerically for a realistic equation of state (with nonzero χ, κ, ζ), diagonalize it, and compare its left eigenvectors with those of §IX or §X; or build the left eigenvectors using a different admissible one-sided inverse (e.g., from differentiating v^i = u^i/(1+u^ju_j)^{1/2}) and check whether [L][X] = 1 still holds. Any deviation would show the quasi-invertible prescription is not complete.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • GRHD codes can adopt full-wave Riemann solvers and characteristic boundary conditions using the §IX eigenvectors, which are valid for any spatial direction and simpler than previous forms.
  • Simulations that evolve electron fraction under nuclear statistical equilibrium get a new six-wave decomposition (§X) that resolves the composition wave at the fluid speed, improving accuracy in neutrino-transport and merger simulations.
  • The quasi-invertible method removes the need for computer algebra when deriving decompositions for relativistic fluids, making the derivation shorter and less error-prone.
  • If the companion paper delivers the promised GRMHD version, the same machinery will provide the first complete characteristic decomposition in conserved variables for magnetized relativistic flows, enabling the most accurate known flux and boundary treatments there too.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • This reader's inference: the one-sided-inverse construction is a general recipe for any hyperbolic system where a constrained set of variables (e.g., a four-vector with unit norm) is replaced by an unconstrained three-vector; the same logic might transfer to other constrained evolution systems beyond fluids.
  • The paper leaves the uniqueness of the inverse formally open; because the D-term is set to zero on the strength of a contraction argument, users of the method in new settings should re-derive the left-eigenvector rule (7.13) for their own system rather than assume it carries over.
  • A natural numerical experiment, not performed here, is a shock-tube test with an approximate Riemann solver built from the §X eigenvectors: if the composition wave is captured more sharply than with the standard five-wave treatment, that would demonstrate the practical value of the new decomposition.
  • The paper's recovery of the known decomposition in simpler form suggests the transformation technique itself could serve as a template for re-deriving characteristic decompositions in other relativistic formulations where the magnetic field transformation is also non-square.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes a new transformation technique, called quasi-invertible transformations, to obtain characteristic decompositions for relativistic fluid systems. After deriving the comoving-frame decomposition, the method is used to transform to Eulerian observers and then to the conserved variables used in numerical relativity codes. The authors recover the known GRHD eigenvectors in what they argue is a simpler form, and present a new decomposition for fluids whose composition is tracked by an electron fraction in nuclear statistical equilibrium. The stated goal is to enable full-wave Riemann solvers and characteristic boundary conditions; a companion paper is promised for GRMHD.

Significance. If the final formulas are correct, this is a useful contribution: the derivation is parameter-free, makes no use of fitted constants, and the hydrodynamics results are benchmarked against the existing literature. The new composition-dependent eigenvectors in Section X are a concrete new result. The quasi-invertible transformation idea is potentially important for the promised GRMHD extension. The main caveat is that the central deliverable — the final 5x5 and 6x6 eigenvector sets — is algebra-heavy and not machine-checked, and the quasi-invertible formalism is presented informally at a few load-bearing points.

major comments (3)
  1. [Sec. VII.B, Eq. (7.8)] The quasi-invertibility relation A^a_old = A^a ∂U/∂U_old appears dimensionally inconsistent as written. The old system uses a 4-velocity u^a (six nominal components with p and epsilon) while the new system uses a 3-velocity v^a (five components with rho and epsilon); the velocity Jacobians in Eqs. (6.7) and (6.15) are 4x3 and 3x4. The paper never defines the constrained 5-dimensional tangent space on which the transformation is an isomorphism. Please state the reduction (e.g., quotient by the u·u = -1 constraint) and prove the quasi-invertibility relations on that space, or give a precise rectangular-matrix calculus for the eigenvalue problem. This is load-bearing because Eq. (7.13) and all subsequent left eigenvectors depend on it.
  2. [Sec. VI.B, Eqs. (6.11)–(6.15)] The paper says immediately above Eq. (6.14) that 'no theorem guarantees that we can do so.' This is a central step: the one-sided inverse is non-unique, with B and D arbitrary. A direct calculation shows that B=0 makes Eq. (6.14) hold and that the D-term annihilates on the physical subspace, but the manuscript does not supply this verification. Please add a short lemma or proof after Eq. (6.15) and explicitly state in Sec. VII.B why the D-term cannot enter Eq. (7.13). Without this, the prescription for the inverse remains ad hoc.
  3. [Secs. IX–X, Eqs. (9.19)–(9.20), (10.14)–(10.17)] The paper asserts that the left and right eigenvector matrices are mutual inverses, but no explicit orthonormality check is shown for the final 5x5 and 6x6 systems. These formulas are the main deliverable and are algebra-heavy, so a single unchecked sign or prefactor would break the [L][X] = I normalization. Please provide an explicit verification, or better, an ancillary notebook or appendix that checks all dot products, including the new Ye block. This is a verification gap rather than a demonstrated error.
minor comments (5)
  1. [Abstract and Ref. [18]] Abstract has 'the the evolution'; Ref. [18] has typo 'Magnetoydrodynamics' and a placeholder arXiv number. Please correct.
  2. [Eq. (8.24)] The eigenvector x3 is rescaled by a factor of chi relative to the direct application of Eq. (8.23). Please state this normalization explicitly, since the scaling affects the later conservative eigenvectors.
  3. [Eq. (9.19)] The conservative right eigenvectors are given in a specific normalization (e.g., R3 has D-component kappa, R± have D-component 1). State this before the equation; otherwise the missing rho factors in the S_i and tau components look like dimensional errors.
  4. [Sec. VII.B] The distinction between the tentative left eigenvector eL and the final L in Eq. (7.13) is not explained in enough detail. A short paragraph on why the normalization (7.12) is the correct one for the non-invertible case would improve readability.
  5. [Sec. VIII.B] The tangential vectors t_(1,2) in the Eulerian frame are introduced only in words. A brief explicit definition (e.g., two orthonormal vectors orthogonal to n^a and s^a) would be helpful.

Circularity Check

0 steps flagged

No significant circularity: the decomposition is derived by explicit transformations from the comoving fluid equations and checked against independent references.

full rationale

The derivation chain is self-contained. The comoving system (4.15) follows directly from the fluid equations and the equation of state; the eigenvalues and eigenvectors (4.20) and (4.26)-(4.35) are obtained by solving the principal symbol. The later sections only transform those eigenvectors through explicit Jacobians: u->v in Eqs. (6.7)/(6.15), (p,epsilon)->(rho,epsilon) in Eqs. (8.22)-(8.23), and primitive-to-conserved in Eqs. (9.5)/(9.15). The one genuinely non-square step is explicitly non-unique, and the paper candidly says 'no theorem guarantees that we can do so' before imposing Eq. (6.14); setting D=0 is justified by the stated annihilating-subspace argument. This is a flagged construction/verification assumption, not a fitted parameter and not an output defined by its own input. The resulting eigenvectors are benchmarked against external, independent references ([2-4], [9], [17]), and no constants are fitted. The only self-citation, [18], is a roadmap to Paper II and is not load-bearing here. The composition-dependent extension in Section X is a direct block-diagonal extension of the same calculation. Therefore no circular step is identified; residual concerns are algebraic verification gaps, not circularity.

Axiom & Free-Parameter Ledger

1 free parameters · 6 axioms · 1 invented entities

No data-fitting parameters, no new physics entities. The only hand-chosen quantity is the non-unique one-sided inverse coefficients (B=0, D=0). The axioms are standard GR/fluid assumptions plus two domain assumptions specific to the simulation context (principal-part decoupling; NSE single-composition variable) and one ad hoc algebraic ansatz (§VI.B) that the entire left-eigenvector construction relies on.

free parameters (1)
  • Inverse-choice coefficients B, D in Eq. (6.11) = B = 0 (required by Eq. 6.14); D = 0 (claimed to never contribute)
    The one-sided inverse ∂v^a/∂u^b is non-unique; the paper pins it down by imposing the matrix-inverse condition (6.14), which forces B=0, and sets D=0 on the assertion that it never acts on contributing directions. The left eigenvectors (7.13) depend on this choice, so the central derivation rests on it.
axioms (6)
  • standard math The system is strongly hyperbolic with a complete eigensystem in the directions of interest
    Invoked in §II to justify invertibility of [X] and [L]; standard for perfect-fluid GRHD, though for GRMHD it is the open question motivating the series.
  • domain assumption Perfect-fluid flow, isentropic away from shocks (ds = 0), giving the energy equation (4.10)–(4.11)
    §IV.A: the comoving decomposition starts from Eqs. (4.7)–(4.13), which assume ds = 0 for the energy equation.
  • domain assumption Gravitational and matter principal parts decouple: T_ab does not couple into the characteristic matrix of the matter sector
    §II, following Ref. [9] §II.B.3; it justifies treating matter characteristics separately from Einstein-equation characteristics.
  • standard math 3+1 metric (3.2)–(3.3) and the observer relationships (Eulerian normal n^a, coordinate t^a = αn^a + β^a) describe the simulation frame
    §III; standard GR background imported from Refs. [13–15].
  • ad hoc to paper The one-sided inverse ∂v^a/∂u^b can be chosen to satisfy both (6.12) and (6.14), with D = 0 never contributing
    §VI.B: the inverse is explicitly non-unique; the paper states "no theorem guarantees we can do so" before imposing (6.14). This is the paper's own construction and is load-bearing for the left eigenvectors.
  • domain assumption In NSE the composition is a single scalar Y_e satisfying ∇_a(ρY_e u^a) = 0, with p = p(ρ, ε, Y_e)
    §X; standard nuclear-statistical-equilibrium modeling assumption for the composition-dependent sector.
invented entities (1)
  • Quasi-invertible transformation no independent evidence
    purpose: Non-square variable transformations (e.g., 4-velocity → 3-velocity) that preserve the principal part via A^a = A^a_old ∂U_old/∂U and allow left-eigenvector transformation via Eq. (7.13)
    Mathematical construction, not a physical entity; its validity is checked only internally (and in the promised Paper II for GRMHD), so there is no independent falsifiable handle in this paper.

pith-pipeline@v1.3.0-alltime-deepseek · 23594 in / 32620 out tokens · 296376 ms · 2026-08-03T21:45:23.341102+00:00 · methodology

0 comments
read the original abstract

The characteristic decomposition for GRMHD is not known in a form useful for current numerical simulations. This prevents us from using the most accurate known computational methods, such as full-wave Riemann solvers. In this paper, we present a new method of finding decompositions. The method is based on transformations from the comoving frame, where the fluid flow is simplest and the decomposition has been known for a long time. The key innovation we introduce is that of quasi-invertible transformations. In this first paper, we introduce these transformations using the simpler example of relativistic hydrodynamics. We recover the known decomposition for relativistic hydrodynamics in somewhat simpler form than previously derived, and without the need for computer algebra. A new result in this paper is the characteristic decomposition when the the evolution tracks the composition of a fluid in nuclear statistical equilibrium. In Paper II of this series, we apply a quasi-invertible transformation to derive the complete characteristic decomposition for GRMHD in the conserved variables used in simulations.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

19 extracted references · 5 linked inside Pith

  1. [1]

    Donat, J

    R. Donat, J. A. Font, J. M. Ib´ a˜ nez, and A. Marquina, A flux-split algorithm applied to relativistic flows, J. Comp. Phys.146, 58 (1998)

    1998

  2. [2]

    Banyuls, J

    F. Banyuls, J. A. Font, J. M. Ib´ a˜ nez, J. M. Mart ´ ı, and J. A. Miralles, Numerical {3 + 1 } general relativistic hy- drodynamics: A local characteristic approach, Astrophys. J.476, 221 (1997)

    1997

  3. [3]

    J. M. Ib´ a˜ nez, M. A. Aloy, J. A. Font, J. M. Mart ´ ı, J. A. Miralles, and J. A. Pons, Riemann solvers in general relativistic hydrodynamics, inGodunov Methods: Theory and Applications, edited by E. F. Toro (Springer US, New York, NY, 2001) pp. 485–496, arXiv:astro-ph/9911034

    Pith/arXiv arXiv 2001

  4. [4]

    J. A. Font, Numerical hydrodynamics and magnetohy- drodynamics in general relativity, Living Rev. Rel.11, 7 (2008)

    2008

  5. [5]

    J. M. Mart ´ ı and E. M¨ uller, Grid-based methods in rela- tivistic hydrodynamics and magnetohydrodynamics, Liv. Rev. Comp. Astrophys.1, 3 (2015)

    2015

  6. [6]

    J. A. Font, M. Miller, W.-M. Suen, and M. Tobias, Three- dimensional numerical general relativistic hydrodynamics: Formulations, methods, and code tests, Phys. Rev. D61, 044011 (2000), arXiv:gr-qc/9811015 [gr-qc]

    Pith/arXiv arXiv 2000

  7. [7]

    Ant´ on,Magnetohidrodin´ amica relativista num´ erica: Aplicaciones en relatividad especial y general, Ph.D

    L. Ant´ on,Magnetohidrodin´ amica relativista num´ erica: Aplicaciones en relatividad especial y general, Ph.D. thesis, Universitat de Val` encia (2008), (in Spanish)

    2008

  8. [8]

    Ant´ on, J

    L. Ant´ on, J. A. Miralles, J. M. Mart ´ ı, J. M. Ib´ a˜ nez, M. A. Aloy, and P. Mimica, Relativistic magnetohydrodynam- ics: Renormalized eigenvectors and full wave decomposi- tion Riemann solver, Astrophys. J. Suppl.188, 1 (2010), arXiv:0912.4692 [astro-ph.IM]

    Pith/arXiv arXiv 2010

  9. [9]

    Schoepe, D

    A. Schoepe, D. Hilditch, and M. Bugner, Revisiting hy- perbolicity of relativistic fluids, Phys. Rev. D97, 123009 (2018), arXiv:1712.09837 [gr-qc]

    Pith/arXiv arXiv 2018

  10. [10]

    Hilditch and A

    D. Hilditch and A. Schoepe, Hyperbolicity of divergence cleaning and vector potential formulations of general rela- tivistic magnetohydrodynamics, Phys. Rev. D99, 104034 (2019)

    2019

  11. [11]

    K. O. Friedrichs, On the laws of relativistic electro- magneto-fluid dynamics, Comm. Pure Appl. Math.27, 749 (1974)

    1974

  12. [12]

    A. M. Anile,Relativistic fluids and magneto-fluids : with applications in astrophysics and plasma physics(Cam- bridge University Press, New York, 1989)

    1989

  13. [13]

    Smarr and J

    L. Smarr and J. York, James W., Kinematical conditions in the construction of spacetime, Phys. Rev. D17, 2529 (1978)

    1978

  14. [14]

    T. W. Baumgarte and S. L. Shapiro,Numerical Relativity: Solving Einstein ’s Equations on the Computer(Cambridge 20 University Press, New York, 2010)

    2010

  15. [15]

    Rezzolla and O

    L. Rezzolla and O. Zanotti,Relativistic Hydrodynamics (Oxford University Press, Oxford, 2013)

    2013

  16. [16]

    Strang,Linear algebra and its applications, 4th ed

    G. Strang,Linear algebra and its applications, 4th ed. (Thomson Brooks/Cole, Belmont, CA, 2006)

    2006

  17. [17]

    Mar ´ ıa Ib´ a˜ nez, I

    J. Mar ´ ıa Ib´ a˜ nez, I. Cordero-Carri´ on, and J. A. Miralles, On numerical relativistic hydrodynamics and barotropic equations of state, Class. Quantum Grav.29, 157001 (2012), arXiv:1206.5972 [astro-ph.SR]

    Pith/arXiv arXiv 2012

  18. [18]

    S. A. Teukolsky, Characteristic decomposition for rel- ativistic numerical simulations: II. Magnetoydrody- namics, arXiv e-prints , arXiv:xxxx.yyyyy (2025), arXiv:xxxx.yyyyy [gr-qc]

    2025

  19. [19]

    J. M. Martin-Garcia, xact: Efficient tensor computer algebra,https://josmar493.dreamhosters.com(2025)

    2025