Convergence and all-time validity of Hilbert's power-series solution of the Boltzmann equation: completing Hilbert's 1912 integral-equation foundation of the kinetic theory of gases

In his 1912 paper 'Begründung der kinetischen Gastheorie', David Hilbert recast the initial-value problem for the Boltzmann equation of a monatomic hard-sphere gas in terms of his theory of linear integral equations of the second kind, representing the solution as a formal power series in a small positive parameter $\lambda$ proportional to the mean free path (the Knudsen number), and asserted that this series describes the 'stable' gas state — the state whose distribution function remains finite and continuous for all times — with that state determined for all times by just five initial hydrodynamic fields (density, momentum, energy); in the paper this determination is proved uniquely only within the class of formal power-series solutions in $\lambda$. The paper gives no convergence proof for the series; it announces the numerical evaluation of the viscosity and thermal conductivity from the resulting integral equations without carrying it out, and it closes by mentioning an application to Kirchhoff's law of radiation whose full elaboration appeared in Hilbert's later communications. A century of later work has settled the calculational and physical-scope components — the Chapman–Enskog theory evaluated the transport coefficients, quantum-statistical and polyatomic kinetic theories extended the scope beyond the source's monatomic elastic spheres, and Hilbert himself elaborated the radiation application in follow-up communications — and has partially settled the analytic core: the expansions are asymptotic to a special class of solutions (Grad 1963), are rigorously valid as finite truncations with remainder up to shock formation for well-prepared initial data (Nishida 1978; Caflisch 1980; Guo–Jang–Jiang 2009), and the weak-solution incompressible program was finalized by Golse–Saint-Raymond (2004). The convergence of Hilbert's series and its validity for arbitrary times remain open; recent preprints by Deng–Hani–Ma (2024, 2025) claim a long-time derivation of the fluid equations from hard-sphere dynamics and their scope is under active debate.

Background

The Boltzmann equation describes a dilute gas by a distribution function $f(t,x,v)$ giving the density of molecules at position $x$ and velocity $v$ at time $t$; its right-hand side is a quadratic collision integral $Q(f,f)$ encoding elastic binary collisions. When the gas is close to local thermodynamic equilibrium, $f$ is approximately a local Maxwellian (the Gaussian determined by the local density, mean velocity, and temperature). The Knudsen number $\varepsilon$ is the ratio of the mean free path between collisions to a macroscopic length scale; $\varepsilon \to 0$ is the regime in which continuum (fluid) descriptions should become exact. The central open problem of mathematical kinetic theory, and the kinetic-theory instance of Hilbert's sixth problem (1900, 'to develop mathematically the limiting processes ... which lead from the atomistic view to the laws of motion of continua', as quoted in Deng–Hani–Ma 2025), is to make this passage rigorous.

Hilbert's 1912 paper — the last chapter of his book on linear integral equations, and the founding document of the field now called hydrodynamic limits — approaches this by expanding the solution of the (rescaled) Boltzmann equation as a formal power series in a small positive parameter $\lambda$ proportional to the mean free path, with leading term a local Maxwellian. Each coefficient of the series is determined by solving a linear integral equation of the second kind (an equation of the form $u = g + \lambda K u$ for an unknown function $u$, with $g$ given and $K$ a fixed integral operator) with a symmetric kernel built from the collision operator (for the hard-sphere gas this kernel is weakly singular and explicitly computable). Five independent functions solve the homogeneous problem — the collision invariants $1$, $\xi$, $\eta$, $\zeta$, and $\xi^2+\eta^2+\zeta^2$ (molecular mass, momentum, and energy, which collisions conserve) — so by the Fredholm alternative (the classical solvability criterion for equations of this form) each integral equation is solvable only if its right-hand side is orthogonal to these five functions. These five solvability conditions are precisely the compressible Euler equations of gas dynamics. The transport coefficients (viscosity and thermal conductivity) then emerge from solving particular integral equations of the same type, and Hilbert states that the resulting coefficients can be computed numerically. He further asserts that the 'stable' state of the gas — the one whose distribution remains finite and continuous for all times — is fixed, for all times, by the initial values of only the five hydrodynamic fields, a uniqueness he establishes within the class of formal power-series solutions in $\lambda$, and he sketches the same method for elementary radiation theory, aiming at a rigorous proof of Kirchhoff's law (the equality of a body's emissivity and absorptivity in thermal equilibrium).

What Hilbert's paper does not do: it gives no proof that the power series converges, or that summing (or truncating) it approximates a genuine solution of the Boltzmann equation for arbitrary times; the numerical evaluation of the transport coefficients is announced but not carried out in that paper; the theory covers only monatomic elastic spheres; and the radiation application is only outlined. The later literature sharpened exactly these gaps. Grad (1963) proved that Hilbert-type expansions are asymptotic only to a very special class of solutions: an arbitrary initial distribution first decays rapidly (the initial layer) to a 'special Hilbert distribution function' — the content of the 'Hilbert paradox' emphasized by Uhlenbeck (1980): five moments cannot encode a general initial distribution. Caflisch (1980) rigorously justified finite truncations of the Hilbert expansion, with a remainder bound, up to the first appearance of shocks and only for special (well-prepared) initial values, i.e. data already compatible with the expansion and carrying no initial layer; Guo–Jang–Jiang (2009) established the local-in-time version. On the weak-solution side, Bardos–Golse–Levermore (1991) formulated and Golse–Saint-Raymond (2004) finalized the derivation of the incompressible Navier–Stokes–Fourier equations. Saint-Raymond (2014) stresses that the meaning of the formal series remains unsettled ('even the first two orders of the Chapman–Enskog expansion ... still lack for a rigorous derivation' in the setting of renormalized solutions, the global weak solutions of the Boltzmann equation available for general data). Gorban–Karlin (2014) survey exact summations of such series for special models and the obstacles to treating the general case. Slemrod (2018) argues that the Boltzmann-to-Euler limit in general fails and produces van der Waals–Korteweg capillarity corrections (stress terms involving density gradients, as in Korteweg's continuum theory of capillarity), and Bobylev (2018) shows that the higher-order (Burnett-level) corrections in the standard expansions — the terms beyond Navier–Stokes order, named after Burnett — are unstable. The complementary, particle-level leg — from Newtonian hard-sphere dynamics to the Boltzmann equation (Lanford 1975, short times) — has recently been claimed for long times and connected to the fluid equations in the Deng–Hani–Ma preprints (2024, 2025), whose scope is contested (Gao 2025).

Problem Statement

Hilbert (1912) reduced the initial-value problem for the Boltzmann equation of a monatomic gas of elastic spheres to the theory of linear integral equations of the second kind with an explicit symmetric kernel, representing the solution as a formal power series in a small positive parameter $\lambda$ proportional to the mean free path (the Knudsen number), whose leading term is a local Maxwellian and whose solvability conditions at each order are the compressible Euler equations; he asserted that this series describes the 'stable' state of the gas — the state whose distribution function remains finite and continuous for all times — and that this state is determined, for all times, by the initial values of only five hydrodynamic fields (density, three momentum components, energy), a determination he proved uniquely only within the class of formal power-series solutions in $\lambda$.

The open problem attached to this foundation, as identified by the later literature (Grad 1963; Saint-Raymond 2014) rather than by any list in the 1912 source, is to prove or disprove that Hilbert's series converges, for fixed small Knudsen number, to a solution of the Boltzmann equation and that it represents the gas state for arbitrary times. A strictly weaker question is to prove or disprove that the series is rigorously valid as an asymptotic expansion, at the same fixed small Knudsen number, with its remainder controlled in an explicitly stated norm on an explicitly stated admissible class of initial data and for arbitrary times (beyond the finite-time, well-prepared case already settled, Caflisch 1980). Settling the convergence question is what establishes or refutes Hilbert's asserted theorem that the state of a stable gas is uniquely determined for all time by the five initial hydrodynamic fields: a positive resolution confirms that the five initial fields determine the state for all times, while a negative resolution refutes the claim that Hilbert's series supplies that representation. A result that settles only the weaker asymptotic-expansion question does not by itself settle the uniqueness claim. The 1912 paper itself contains no convergence proof and no concluding open-problem list; it announced the numerical evaluation of the transport coefficients (viscosity and thermal conductivity) from certain integral equations, and it mentioned the application to Kirchhoff's law of radiation as a result just found, with the full elaboration appearing in the later radiation communications. Those announced components, together with the extension of kinetic theory beyond monatomic elastic spheres to polyatomic molecules, non-spherical interaction laws, and quantum-statistical gases — a development of later twentieth-century literature rather than a goal of the 1912 source — and the finite-time, well-prepared-data portion of the convergence question, have been addressed by later work (see Background); the unresolved core is the convergence and all-time validity of Hilbert's series for the nonlinear Boltzmann equation in the general sense stated above.

References

1. D. Hilbert, 'Begründung der kinetischen Gastheorie', Mathematische Annalen 72 (1912), no. 4, pp. 562–577. DOI: 10.1007/BF01456676. URL: https://doi.org/10.1007/bf01456676
2. D. Hilbert, 'Mathematical problems', Bulletin of the American Mathematical Society 8 (1902), no. 10, pp. 437–479. DOI: 10.1090/S0002-9904-1902-00923-3. URL: https://doi.org/10.1090/S0002-9904-1902-00923-3
3. D. Hilbert, 'Begründung der elementaren Strahlungstheorie', in Gesammelte Abhandlungen, Band III, Springer, Berlin, 1935, pp. 217–230. DOI: 10.1007/978-3-662-38452-7_15. URL: https://doi.org/10.1007/978-3-662-38452-7_15
4. D. Hilbert, 'Bemerkungen zur Begründung der elementaren Strahlungstheorie', in Gesammelte Abhandlungen, Band III, Springer, Berlin, 1935, pp. 231–237. DOI: 10.1007/978-3-662-38452-7_16. URL: https://doi.org/10.1007/978-3-662-38452-7_16
5. D. Hilbert, 'Zur Begründung der elementaren Strahlungstheorie. Dritte Mitteilung', in Gesammelte Abhandlungen, Band III, Springer, Berlin, 1935, pp. 238–257. DOI: 10.1007/978-3-662-38452-7_17. URL: https://doi.org/10.1007/978-3-662-38452-7_17
6. H. Grad, 'Asymptotic theory of the Boltzmann equation', The Physics of Fluids 6 (1963), no. 2, pp. 147–181. DOI: 10.1063/1.1706716. URL: https://doi.org/10.1063/1.1706716
7. S. Chapman, 'VI. On the law of distribution of molecular velocities, and on the theory of viscosity and thermal conduction, in a non-uniform simple monatomic gas', Philosophical Transactions of the Royal Society of London, Series A 216 (1916), pp. 279–348. DOI: 10.1098/rsta.1916.0006. URL: https://doi.org/10.1098/rsta.1916.0006
8. C. Cercignani, 'The Hilbert and Chapman–Enskog theories', in Mathematical Methods in Kinetic Theory, Plenum Press, New York, 1969, pp. 101–125. DOI: 10.1007/978-1-4899-5409-1_5. URL: https://doi.org/10.1007/978-1-4899-5409-1_5
9. E. A. Uehling and G. E. Uhlenbeck, 'Transport phenomena in Einstein-Bose and Fermi-Dirac gases. I', Physical Review 43 (1933), pp. 552–561. DOI: 10.1103/PhysRev.43.552. URL: https://doi.org/10.1103/PhysRev.43.552
10. H. Moraal, 'Quantum kinetic theory of polyatomic gases', Physics Reports 17 (1975), no. 5, pp. 225–306. DOI: 10.1016/0370-1573(75)90042-3. URL: https://doi.org/10.1016/0370-1573(75)90042-3
11. G. E. Uhlenbeck, 'Some notes on the relation between fluid mechanics and statistical physics', Annual Review of Fluid Mechanics 12 (1980), pp. 1–10. DOI: 10.1146/annurev.fl.12.010180.000245. URL: https://doi.org/10.1146/annurev.fl.12.010180.000245
12. T. Nishida and K. Imai, 'Global solutions to the initial value problem for the nonlinear Boltzmann equation', Publications of the Research Institute for Mathematical Sciences 12 (1976), no. 1, pp. 229–239. DOI: 10.2977/prims/1195190965. URL: https://doi.org/10.2977/prims/1195190965
13. T. Nishida, 'Fluid dynamical limit of the nonlinear Boltzmann equation to the level of the compressible Euler equation', Communications in Mathematical Physics 61 (1978), no. 2, pp. 119–148. DOI: 10.1007/BF01609490. URL: https://doi.org/10.1007/BF01609490
14. R. E. Caflisch, 'The fluid dynamic limit of the nonlinear Boltzmann equation', Communications on Pure and Applied Mathematics 33 (1980), no. 5, pp. 651–666. DOI: 10.1002/cpa.3160330506. URL: https://doi.org/10.1002/cpa.3160330506
15. Y. Guo, J. Jang and N. Jiang, 'Local Hilbert expansion for the Boltzmann equation', Kinetic and Related Models 2 (2009), no. 1, pp. 205–214. DOI: 10.3934/krm.2009.2.205. URL: https://doi.org/10.3934/krm.2009.2.205
16. C. Bardos, F. Golse and C. D. Levermore, 'Fluid dynamic limits of kinetic equations. I. Formal derivations', Journal of Statistical Physics 63 (1991), pp. 323–344. DOI: 10.1007/BF01026608. URL: https://doi.org/10.1007/BF01026608
17. F. Golse and L. Saint-Raymond, 'The Navier–Stokes limit of the Boltzmann equation for bounded collision kernels', Inventiones mathematicae 155 (2004), no. 1, pp. 81–161. DOI: 10.1007/s00222-003-0316-5. URL: https://doi.org/10.1007/s00222-003-0316-5
18. L. Saint-Raymond, 'A mathematical PDE perspective on the Chapman–Enskog expansion', Bulletin of the American Mathematical Society 51 (2014), no. 2, pp. 247–275. DOI: 10.1090/S0273-0979-2013-01440-X. URL: https://doi.org/10.1090/S0273-0979-2013-01440-X
19. A. N. Gorban and I. V. Karlin, 'Hilbert's 6th problem: exact and approximate hydrodynamic manifolds for kinetic equations', Bulletin of the American Mathematical Society 51 (2014), no. 2, pp. 187–246. DOI: 10.1090/S0273-0979-2013-01439-3. URL: https://doi.org/10.1090/S0273-0979-2013-01439-3
20. M. Slemrod, 'Hilbert's sixth problem and the failure of the Boltzmann to Euler limit', Philosophical Transactions of the Royal Society A 376 (2018), no. 2118, 20170222. DOI: 10.1098/rsta.2017.0222. URL: https://doi.org/10.1098/rsta.2017.0222
21. A. V. Bobylev, 'Boltzmann equation and hydrodynamics beyond Navier–Stokes', Philosophical Transactions of the Royal Society A 376 (2018), no. 2118, 20170227. DOI: 10.1098/rsta.2017.0227. URL: https://doi.org/10.1098/rsta.2017.0227
22. O. E. Lanford III, 'Time evolution of large classical systems', in J. Moser (ed.), Dynamical Systems, Theory and Applications, Lecture Notes in Physics, vol. 38, Springer, Berlin, 1975, pp. 1–111. DOI: 10.1007/3-540-07171-7_1. URL: https://doi.org/10.1007/3-540-07171-7_1
23. S. Génieys, 'Energy-transport model for a nondegenerate semiconductor: Convergence of the Hilbert expansion in the linearized case', Asymptotic Analysis 17 (1998), no. 4, pp. 279–308. DOI: 10.3233/ASY-1998-315. URL: https://doi.org/10.3233/ASY-1998-315
24. C. F. Delale, 'The Hilbert expansion to the Boltzmann equation for steady flow', Journal of Statistical Physics 28 (1982), no. 3, pp. 589–602. DOI: 10.1007/BF01008326. URL: https://doi.org/10.1007/BF01008326
25. A. Schirrmacher, 'Experimenting theory: The proofs of Kirchhoff's radiation law before and after Planck', Historical Studies in the Physical and Biological Sciences 33 (2003), no. 2, pp. 299–335. DOI: 10.1525/hsps.2003.33.2.299. URL: https://doi.org/10.1525/hsps.2003.33.2.299
26. S. Wang and Y. Xu, 'The fluid-dynamical limit of the Boltzmann equation for a polyatomic gas model with soft potential', Journal of Mathematical Physics 67 (2026), no. 3, 033101. DOI: 10.1063/5.0303609. URL: https://doi.org/10.1063/5.0303609
27. Yu Deng, Zaher Hani and Xiao Ma, 'Long time derivation of the Boltzmann equation from hard sphere dynamics', arXiv:2408.07818 (2024). URL: https://arxiv.org/abs/2408.07818
28. Yu Deng, Zaher Hani and Xiao Ma, 'Hilbert's sixth problem: derivation of fluid equations via Boltzmann's kinetic theory', arXiv:2503.01800 (2025). URL: https://arxiv.org/abs/2503.01800
29. Shan Gao, 'Comment on "Hilbert's Sixth Problem: Derivation of Fluid Equations via Boltzmann's Kinetic Theory" by Deng, Hani, and Ma', arXiv:2504.06297 (2025). URL: https://arxiv.org/abs/2504.06297