Can a collisionless relativistic plasma remain smooth for all time?

← All digests

Family 362 · Partial differential equations
Importance 5/5 · Provisional editorial rating
Sources consulted 2026-10-26

1. The question

A collisionless plasma is described by a cloud of charged particles coupled to the electromagnetic field they generate. The field accelerates the particles; the moving particles, in turn, change the field. Even without particle collisions, this feedback makes the evolution nonlinear.

For smooth initial data, the relativistic Vlasov–Maxwell equations have a smooth solution for a short time. The central large-data question is whether that solution can develop a singularity in finite time. Particles never reach the speed of light, but their momenta can become arbitrarily large as their speeds approach that limit. A speed bound alone therefore does not control the evolution.

Family 362 claims global existence and uniqueness of smooth solutions in three spatial dimensions for the one-species model, with no size or symmetry restrictions on admissible initial data. The particle density is initially smooth and compactly supported in both position and momentum; the fields have finite energy and bounded derivatives of every order.

The manuscript’s proposed mechanism is an estimate on the net momentum change along a particle trajectory. It retains cancellation in the force before taking its magnitude, preventing momentum from doubling infinitely often in finite time.

2. Background and essential definitions

Position, momentum and relativistic velocity

Let $f(t,x,p)\ge0$ be the particle density in phase space, with position $x\in\mathbb R^3$ and momentum $p\in\mathbb R^3$. In units where particle mass and the speed of light are one, define

$$q(p)=\sqrt{1+|p|^2},\qquad u(p)=\frac{p}{q(p)}.$$

Here $q$ is particle energy and $u$ is velocity. Always $|u|<1$, even when $|p|$ is very large. Momentum and velocity are therefore different variables.

Let $E(t,x)$ and $B(t,x)$ be the electric and magnetic fields. The Vlasov equation is

$$\partial_t f+u(p)\cdot\nabla_x f +\bigl(E+u(p)\times B\bigr)\cdot\nabla_p f=0.$$

It says that density is transported along particle trajectories, with no collision term. Those trajectories satisfy

$$X'=u(P),\qquad P'=E(t,X)+u(P)\times B(t,X).$$

The force is the Lorentz force. The magnetic part changes momentum direction but does no direct work, since it is perpendicular to velocity.

Maxwell’s equations close the feedback loop

Charge and current densities are

$$\rho(t,x)=\int_{\mathbb R^3}f(t,x,p)\,dp, \qquad j(t,x)=\int_{\mathbb R^3}u(p)f(t,x,p)\,dp.$$

The electromagnetic equations in the manuscript’s normalization are

$$\partial_t E-\nabla\times B=-j, \qquad \partial_t B+\nabla\times E=0,$$

$$\nabla\cdot E=\rho, \qquad \nabla\cdot B=0.$$

There is one particle species and no background charge. The last two equations are constraints that compatible initial data must satisfy and that the evolution preserves.

The conserved total energy has the form

$$\mathcal E=\iint q(p)f(t,x,p)\,dp\,dx +\frac12\int\bigl(|E|^2+|B|^2\bigr)\,dx.$$

This integral bound is powerful but does not directly bound the largest particle momentum or the field at every point. A small amount of density can occupy a high-energy region while the integral stays finite.

Classical versus weak solutions

A classical solution has enough derivatives to satisfy the equations pointwise. The theorem ultimately asserts smoothness on every finite time interval. A weak solution satisfies integrated versions of the equations and may have much lower regularity. Global weak existence does not by itself give global smoothness or uniqueness in the classical class.

3. Development history and previous partial results

Wollman established local classical existence and uniqueness. DiPerna and Lions developed global weak existence for large data; Rein later gave a direct treatment of the relativistic one-species formulation. The remaining classical problem was to prevent loss of regularity.

Glassey and Strauss identified a decisive continuation criterion: a classical solution can be extended as long as the momentum support of the particle distribution stays bounded on finite time intervals. Their work also proved global existence for initially dilute plasmas. Later proofs of the criterion used Fourier methods and physical-space wave-kernel decompositions; Bouchut, Golse and Pallard provide one such approach.

Glassey and Schaeffer obtained global classical solutions in several lower-dimensional reductions and in a symmetric relativistic Vlasov–Poisson setting. Small-data and perturbative results were developed by Rein, Schaeffer, Bigorgne, Wang, Wei and Yang, among others. Some allow noncompact particle tails or large electromagnetic fields, but retain smallness of the particle distribution or proximity to a controlled reference solution.

Luk and Strain sharpened continuation theory by showing that control of momentum projected onto a fixed two-dimensional plane suffices. Other criteria use moments or spatial density bounds. Such results identify sufficient controls, but do not automatically establish those controls for arbitrary data.

A relevant recent predecessor is Xuecheng Wang’s July 2026 two-paper treatment of large cylindrically symmetric data: the iterative smoothing construction and pointwise estimates. It imposes rotational covariance about a fixed axis and uses Sobolev and weighted data assumptions. Family 362 removes symmetry for its own compact-particle data class. The two classes should not be treated as identical, since Wang’s formulation permits localized noncompact particle data.

The retarded-field representation, light-cone energy flux and changes from current phase points to particle labels all have classical antecedents. The new claim rests on combining them into a signed impulse bound strong enough to close the unrestricted large-data continuation argument.

4. What the new manuscript claims

The initial particle density is any

$$f_0\in C_c^\infty(\mathbb R_x^3\times\mathbb R_p^3), \qquad f_0\ge0.$$

The notation $C_c^\infty$ means smooth with compact support. The initial fields satisfy

$$E_0,B_0\in C_b^\infty(\mathbb R^3;\mathbb R^3) \cap L^2(\mathbb R^3;\mathbb R^3),$$

$$\nabla\cdot E_0=\int f_0\,dp, \qquad\nabla\cdot B_0=0.$$

Here $C_b^\infty$ means that the fields and every spatial derivative are bounded. Only the fields themselves are required to be square-integrable; square-integrability of every derivative is not assumed.

For every such datum, the theorem claims a unique global classical solution. For each finite $T$, the functions $f,E,B$ are smooth on $[0,T]$, the particle support stays inside a compact subset of position–momentum space over that interval, and the divergence constraints remain valid.

There is no smallness, spherical symmetry or cylindrical symmetry requirement. Nonzero total charge is allowed: the electric field may have its corresponding Coulomb tail, which is compatible with finite field energy at spatial infinity. The fields are not required to have compact support.

The support bound may grow with $T$. The theorem does not claim a uniform momentum bound for all infinite time, asymptotic scattering, or decay to a particular equilibrium.

5. Why the advance matters

Importance: 5/5, conditional on the claim. Global classical regularity for general large data in the three-dimensional relativistic Vlasov–Maxwell system is a central problem in kinetic PDE. The theorem would remove the size and symmetry restrictions for the stated natural class of smooth compact particle data.

The result would establish deterministic smooth evolution for arbitrarily strong initial configurations within this model. That is substantially stronger than producing weak solutions or constructing special large solutions from prescribed asymptotic behavior.

Its analytic significance is the distinction between total force magnitude and net impulse. Bounding the integral of the absolute force can lose cancellations that are crucial at high momentum. The manuscript proposes to retain enough of the signed vector structure to prevent a finite-time cascade of momentum doublings.

The claim remains specific to the one-species collisionless equations and the stated data. It is not a general regularity theorem for all plasma or fluid models.

6. The main idea

Reduce smooth continuation to momentum support

Along characteristics, the distribution function is transported and particle speeds remain below one. Initial compact spatial support therefore stays within a linearly expanding spatial region on every finite horizon. The harder issue is momentum support.

The Glassey–Strauss continuation criterion reduces the global argument to showing that the largest supported momentum cannot become infinite in finite time. The manuscript proves the local theory and smooth persistence in its precise field class, including a finite-horizon modification that connects it to the Sobolev formulation of the continuation theorem.

This modification retains the Coulomb field and changes divergence-free remainders outside the relevant particle region. Finite propagation makes the removed vacuum Maxwell field irrelevant to the particle dynamics during the chosen horizon. Thus extra global square-integrability of all initial field derivatives is not silently added.

Use the light cone to locate the sources of a force

Electromagnetic influence travels at speed one. The force at a particle is therefore expressed through source particles on its backward light cone, not through an instantaneous interaction at the same time.

The argument follows source trajectories to their intersections with that cone and divides contributions into ranges of momentum, separation distance and relative angles. Energy flux through light cones controls suitable field components. Occupation estimates control how long source particles can remain in a given geometric range.

Direct estimates of the absolute force are still too large in a narrow angular configuration: the source velocity is much more closely aligned with the light ray than the receiving particle’s velocity is. At high receiver energy, the loss is of square-root size relative to the desired estimate.

Preserve cancellation along source trajectories

Instead of taking absolute values term by term, the manuscript integrates the force along source trajectories. An identity combines the part independent of source acceleration with the part involving that acceleration. A term generated by differentiating the cone geometry cancels against the first part.

The remaining expression contains a total derivative and less singular bulk terms. One bulk term involves the receiver’s acceleration, which is itself the Lorentz force. The argument must therefore make its coefficient small enough; integration by parts alone is not the conclusion.

The signed identity is used twice. First, under a provisional impulse estimate, its transverse projection bounds changes of particle direction. That directional control improves occupation estimates when both source and receiver momenta are large: between significant turns, relative motion crosses the relevant region in a controlled way.

Second, the argument applies signed cancellation only in ranges where direct estimates exceed a summable threshold. The selection forces troublesome configurations into the angular window covered by the improved occupation estimate. The resulting receiver-force coefficient gains the square-root factor needed to compensate for the earlier loss.

Time weights handle integration by parts away from interval endpoints. The short endpoint pieces are then restored using the provisional estimate, producing a strict improvement and closing a continuity bootstrap.

Why the final estimate rules out finite-time blowup

The central estimate has the following form. Let $P$ be a dyadic upper scale, at least a fixed multiple of the maximum supported particle energy on a classical interval. Suppose a receiver stays at energy comparable to a dyadic scale $w$ during an interval of length $\Delta t$. Then

$$|\Delta p| \le C_1P\sqrt{\Delta t} +C_2\Delta t\,\frac{P^2\log(2+P)}{w}.$$

The constants depend on the datum and a fixed finite time horizon, but not on the provisional maximum $P$. The norm is taken after integrating the force.

If maximum energy were to pass successively through $2^j$, choose a trajectory making each doubling and take $P$ and $w$ comparable to that level. The estimate gives a lower bound of the form

$$\Delta t_j\ge\frac{c}{\log(2+C2^j)}.$$

These lower bounds have a divergent sum, behaving like the harmonic series in $j$. Infinitely many doublings therefore require infinite total time. The continuation criterion then supplies the claimed global smooth solution.

7. Scope, limitations, and open questions

The theorem concerns three position and three momentum variables, one particle species, and no background charge. It does not state a multispecies extension, a collisional Boltzmann model, or a coupled gravitational system.

Initial compact support of the particle density and the specified smooth finite-energy field class are genuine hypotheses. The result does not automatically cover arbitrary noncompact momentum tails or rough initial data merely because other small-data theories do.

Global classical uniqueness is also distinct from uniqueness among every possible low-regularity weak solution. No such broader uniqueness assertion is made here.

The theorem controls each finite interval, leaving long-time momentum growth, scattering for general large data and extensions to other admissible data classes as separate questions. Its signed-force estimates and global regularity conclusion are reported as stated, without independent proof verification.

8. Sources and further reading

The manuscript’s Theorem 1.1 states the data class and global conclusion; Proposition 2.1 gives the signed increment estimate, and Section 10 supplies the continuation interface.

Primary background includes Bouchut–Golse–Pallard on the Glassey–Strauss criterion, Luk–Strain’s continuation refinement, and Wang’s first and second papers on large cylindrically symmetric data. The manuscript’s bibliography separates weak existence, classical continuation and small-data asymptotics.

Source manuscripts

• M1. OpenAI, Global classical solutions of the three-dimensional relativistic Vlasov–Maxwell system (September 23, 2026).

Sources consulted on 26 October 2026 at repository revision 1581d4c. Download the manuscript citations.