Kinetic theory for a relativistic charged gas: mathematical foundations of the hydrodynamic limit and first-order results within the projection method

1/ Relativistic dissipative hydrodynamics from kinetic theory has long been plagued by causality/instability problems (Eckart, Landau-Lifshitz). Recent BDNK-style first-order theories fix this by exploiting frame and 'representation' freedoms. 2/ This paper gives a microscopic derivation: it generalizes Saint-Raymond's projection method (Chapman–Enskog with the linearized collision operator inverted on the orthogonal complement of its kernel) to the relativistic Boltzmann equation in curved spacetime + EM field. 3/ The natural frame singled out by the projection is the trace-fixed particle frame. The authors recover the Salazar–García-Perciante–Sarbach hyperbolic/causal/stable first-order theory, show transport coefficients are frame-invariant if defined right, and prove the truncated entropy current equals Israel–Stewart's at this order.

Tiny particles bumping into each other in curved space: careful math finds clean rules for how hot fluid flows.

• unflagged_assumption · Section V, Lemma 6
  Lemma 6 explicitly notes K_2 compactness is established only for d=3; yet the rest of the paper proceeds in arbitrary d≥2 as if the spectral framework were available.
• claim_without_derivation · Section IV.E, after Eq. IV.50
  Convergence of the Chapman-Enskog expansion in ε is assumed without proof; authors note absence of results.

• standard_math: Linearized collision operator L is self-adjoint, nonnegative, with finite-dim kernel = collision invariants and a positive spectral gap on L^2(P^+_x, dμ_x).
  Established for d=3 in Dudyński–Ekiel-Jeżewska [32] under stated cross-section bounds; for general d≥2 the compactness of K_2 is not proved here (acknowledged in Lemma 6).
• domain_assumption: Differential cross section satisfies hard-interaction lower bound and Grad-cutoff upper bound (Eq. V.3).
  Excludes Coulomb-like long-range interactions; covers hard spheres/disks and Israel particles.
• domain_assumption: Spacetime (M,g) is time-oriented with smooth g and smooth F; gas is dilute, classical, point-like, binary elastic collisions only.
• ad_hoc_to_paper: Convergence of the Chapman-Enskog series in ε (Eq. IV.49) is assumed formally; no convergence proof.
  Authors explicitly note 'to our knowledge, there are no results on the convergence of the resulting series'.
• standard_math: Hyperbolicity, causality, and stability of the resulting fluid theory under the stated conditions on Γ_1, Γ_2, η/κ.
  Imported from Refs. [17,18]; not re-proved here.

• Comparison with Rocha–Denicol–Noronha modified Chapman-Enskog [21,22] is qualitative: Authors state their method 'bears some similarities' but differs by inverting L rather than using moments; a more explicit technical comparison of derived constitutive equations would strengthen the novelty claim.

This is a mathematically careful theory paper, not an experimental claim. The construction (orthogonal projection onto (ker L)^⊥, kernel = collision invariants ⇒ TFP compatibility conditions) is internally coherent and consistent with cited prior work [4,5,17,18,20–22]. The novel contributions claimed — relativistic projection method, encoding representation freedom by adding on-shell-zero terms to Boltzmann, and identification of Boltzmann entropy with Israel–Stewart at first order — appear to be derivations rather than postulates, with proofs in appendices. Main caveat: the central spectral-gap / compactness result for the linearized collision operator (Lemma 6, K_2 compact) is only established in d=3 via [32]; for general d the analysis is incomplete by the authors' own admission. Also, hyperbolicity/causality/stability are not re-proved here but inherited from [17,18], so this paper is best read as providing kinetic-theoretic foundations for that prior result rather than independently establishing those properties. I have not line-checked every algebraic identity; correctness risk is medium primarily because of length and density of computation, not because of any flagged error.