Modified Zakharov-Kuznetsov Equation

• Modified Zakharov–Kuznetsov equation is a dispersive nonlinear PDE that extends the KdV mechanism to anisotropic, mass-critical settings with solitary wave solutions.
• Analyses cover local and global well-posedness, blow-up concentration, scattering phenomena, and multi-soliton asymptotics across various geometries.
• Recent studies focus on its scaling invariance, conservation laws, and the interplay of focusing–defocusing regimes in determining solution dynamics.

The modified Zakharov–Kuznetsov equation is a dispersive nonlinear evolution equation that extends the modified Korteweg–de Vries mechanism to anisotropic higher-dimensional settings. In its standard real-valued two-dimensional normalization on $\mathbb{R}^2$, it is written

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$

or equivalently $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0$. It is a mass-critical model in two dimensions, supports solitary waves traveling in the distinguished $x$-direction, and has generated parallel theories for local and global well-posedness, blow-up, scattering, analyticity, and multi-soliton asymptotics. The term also appears in closely related semiperiodic, toroidal, complex-valued, and coefficient-retaining variants, so the exact equation depends on domain and normalization (Kinoshita, 2019, Kenig et al., 28 May 2026, Pan-Collantes et al., 2024).

1. Equation and model variants

In the generalized Zakharov–Kuznetsov family on $\mathbb{R}^d$, the unified real-valued model takes the form

$\partial_t u+\partial_{x_1}(\Delta u+u^p)=0,$

and the two-dimensional modified case corresponds to $(d,p)=(2,3)$, namely

$\partial_t u+\partial_{x_1}(\Delta u+u^3)=0.$

In the focusing–defocusing notation often used for the initial-value problem on $\mathbb{R}^2$, the same equation is written

$$v_t+\partial_x(\Delta v)+\sigma\,\partial_x(v^3)=0, \qquad \sigma\in\{+1,-1\},$$

with $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$0 focusing and $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$1 defocusing (Valet, 2020, Bhattacharya et al., 2019).

A standard linear change of variables symmetrizes the anisotropic dispersive operator $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$2 into $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$3. After that transformation, the modified equation is studied in the form

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$4

for a fixed constant $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$5. This symmetrized representation is central in Bourgain-space and Gevrey-space analyses because it replaces the mixed phase $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$6 by the diagonal phase $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$7 (Bhattacharya et al., 2019, Baldasso et al., 2024).

The designation “modified Zakharov–Kuznetsov” is not completely uniform across the literature. On the two-dimensional torus, Kenig, Pavlović, Staffilani, and Velasco introduced the complex-valued equation

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$8

as a two-dimensional analogue of the complex modified KdV equation; in that setting the cubic term is not a total derivative, which materially changes the analysis (Kenig et al., 28 May 2026). In a broader coefficient-retaining traveling-wave setting, the equation

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$9

has also been treated under the mZK label, with the quadratic and cubic advective coefficients kept explicit (Pan-Collantes et al., 2024). By contrast, the rectangle initial-boundary-value problem titled “modified Zakharov-Kuznetsov” in (Castelli et al., 2019) analyzes

$\partial_t u+\partial_{x_1}(\Delta u+u^3)=0$0

so the nomenclature there follows boundary-value usage rather than the standard cubic whole-space normalization.

These variants are related by the common anisotropic dispersive core $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0$1, but they are not interchangeable. Thresholds, conservation laws, and asymptotic behavior depend strongly on whether the setting is $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0$2, $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0$3, $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0$4, a bounded rectangle, or a coefficient-generalized traveling-wave reduction.

2. Scaling, conservation laws, and criticality

For the standard two-dimensional real-valued cubic equation, the natural scaling is

$\partial_t u+\partial_{x_1}(\Delta u+u^3)=0$5

This leaves the $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0$6 norm invariant and yields the critical Sobolev index

$\partial_t u+\partial_{x_1}(\Delta u+u^3)=0$7

so the two-dimensional modified Zakharov–Kuznetsov equation is mass-critical. The same scaling underlies both the global low-regularity theory and the blow-up concentration analysis (Bhattacharya et al., 2019, Bhattacharya, 2020).

For sufficiently regular real-valued solutions, the conserved mass and energy are

$\partial_t u+\partial_{x_1}(\Delta u+u^3)=0$8

and

$\partial_t u+\partial_{x_1}(\Delta u+u^3)=0$9

In the defocusing case $x$0, the quartic contribution is positive and the energy is coercive at the $x$1 level. In the focusing case $x$2, coercivity is recovered below the ground-state mass by the sharp Gagliardo–Nirenberg inequality

$x$3

where $x$4 denotes the positive radial ground state solving

$x$5

equivalently $x$6 (Bhattacharya et al., 2019, Bhattacharya, 2020).

This mass threshold is structurally central. It governs the focusing global theory below $x$7, appears in concentration phenomena at blow-up, and identifies the borderline between coercive and noncoercive energy behavior.

The complex torus equation has a different structure. There the $x$8 norm is formally conserved, because

$x$9

but no Hamiltonian conservation law is used or asserted in that paper. This reflects a basic distinction between the real derivative form $\mathbb{R}^d$0 and the complex nonlinearity $\mathbb{R}^d$1 (Kenig et al., 28 May 2026).

3. Solitary waves and traveling-wave structure

For the standard two-dimensional real equation, solitary waves propagate only along the distinguished $\mathbb{R}^d$2-axis. The traveling-wave ansatz is

$\mathbb{R}^d$3

with speed $\mathbb{R}^d$4 and shift $\mathbb{R}^d$5. The profile $\mathbb{R}^d$6 solves

$\mathbb{R}^d$7

The ground state is unique up to translations, and up to sign in the cubic case, is positive and radially symmetric, smooth, and exponentially decaying together with all derivatives (Valet, 2020).

Linearization around a soliton introduces the self-adjoint operator

$\mathbb{R}^d$8

At $\mathbb{R}^d$9, the kernel is

$\partial_t u+\partial_{x_1}(\Delta u+u^p)=0,$0

and the spectrum contains a single negative eigenvalue $\partial_t u+\partial_{x_1}(\Delta u+u^p)=0,$1 with positive radial eigenfunction $\partial_t u+\partial_{x_1}(\Delta u+u^p)=0,$2. In the mass-critical cubic case, coercivity requires orthogonality not only to the translation modes but also to $\partial_t u+\partial_{x_1}(\Delta u+u^p)=0,$3 and $\partial_t u+\partial_{x_1}(\Delta u+u^p)=0,$4; this additional negative-mode projection is a defining analytical feature of the critical mZK theory (Valet, 2020).

A wider traveling-wave classification has been obtained for the coefficient-retaining equation

$\partial_t u+\partial_{x_1}(\Delta u+u^p)=0,$5

via the oblique ansatz $\partial_t u+\partial_{x_1}(\Delta u+u^p)=0,$6. Writing

$\partial_t u+\partial_{x_1}(\Delta u+u^p)=0,$7

the traveling-wave reduction yields the first integral

$\partial_t u+\partial_{x_1}(\Delta u+u^p)=0,$8

The complete classification is then determined by the degree, sign, and root multiplicity pattern of the quartic $\partial_t u+\partial_{x_1}(\Delta u+u^p)=0,$9, producing rational, kink-type, trigonometric, hyperbolic, and Jacobi-elliptic families (Pan-Collantes et al., 2024).

This reduction makes explicit how oblique propagation enters only through the effective dispersion coefficient $(d,p)=(2,3)$0. A plausible implication is that many qualitative distinctions between bright, kink, and periodic traveling waves are encoded more fundamentally in the quartic phase portrait than in the particular coordinate normalization.

4. Well-posedness theory

Well-posedness results for mZK are highly sensitive to domain, regularity scale, and whether one works in classical Sobolev spaces, Bourgain spaces, atomic spaces, or anisotropically weighted critical spaces. The thresholds therefore organize into parallel rather than strictly nested theories.

Setting	Representative result	Source
$(d,p)=(2,3)$1, classical $(d,p)=(2,3)$2	Local well-posedness for $(d,p)=(2,3)$3; failure of $(d,p)=(2,3)$4 flow below $(d,p)=(2,3)$5	(Kinoshita, 2019)
$(d,p)=(2,3)$6, low-regularity global theory	Global well-posedness for $(d,p)=(2,3)$7 in the defocusing case, and in the focusing case under $(d,p)=(2,3)$8	(Bhattacharya et al., 2019)
$(d,p)=(2,3)$9, anisotropic critical scale	Local well-posedness for $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0.$0, $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0.$1; small-data global well-posedness and scattering in $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0.$2	(Correia et al., 31 Jul 2025)
$\partial_t u+\partial_{x_1}(\Delta u+u^3)=0.$3	Local well-posedness for $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0.$4, improved later to $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0.$5	(Mezher, 2024, Nowicki-Koth, 18 Jun 2025)
$\partial_t u+\partial_{x_1}(\Delta u+u^3)=0.$6, complex-valued equation	Local well-posedness for $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0.$7 and failure of uniform continuity of the flow map	(Kenig et al., 28 May 2026)
$\partial_t u+\partial_{x_1}(\Delta u+u^3)=0.$8	Local well-posedness for $\partial_t u+\partial_{x_1}(\Delta u+u^3)=0.$9 and small-data global well-posedness in $\mathbb{R}^2$0; more generally small-data global well-posedness at $\mathbb{R}^2$1 for $\mathbb{R}^2$2	(Grünrock, 2013, Kinoshita, 2019)

On $\mathbb{R}^2$3, the sharp classical Sobolev threshold is $\mathbb{R}^2$4. Kinoshita’s result yields local well-posedness at $\mathbb{R}^2$5, and the same paper proves that for $\mathbb{R}^2$6 the data-to-solution map fails to be $\mathbb{R}^2$7 at the origin, so the Picard-iteration threshold is sharp in that sense (Kinoshita, 2019).

Below the energy space, Bhattacharya, Farah, and Roudenko used the $\mathbb{R}^2$8-method to globalize the two-dimensional real equation for $\mathbb{R}^2$9. In the defocusing case this holds for arbitrary data, while in the focusing case it requires

$$v_t+\partial_x(\Delta v)+\sigma\,\partial_x(v^3)=0, \qquad \sigma\in\{+1,-1\},$$0

They also obtained the polynomial-in-time bound

$$v_t+\partial_x(\Delta v)+\sigma\,\partial_x(v^3)=0, \qquad \sigma\in\{+1,-1\},$$1

for the corresponding global solution (Bhattacharya et al., 2019).

A later critical-scale refinement introduced the anisotropic spaces

$$v_t+\partial_x(\Delta v)+\sigma\,\partial_x(v^3)=0, \qquad \sigma\in\{+1,-1\},$$2

with norm

$$v_t+\partial_x(\Delta v)+\sigma\,\partial_x(v^3)=0, \qquad \sigma\in\{+1,-1\},$$3

These spaces were designed to penalize concentration near the weakly dispersive axes in the symmetrized phase $$v_t+\partial_x(\Delta v)+\sigma\,\partial_x(v^3)=0, \qquad \sigma\in\{+1,-1\},$$4. In this framework, local well-posedness holds for $$v_t+\partial_x(\Delta v)+\sigma\,\partial_x(v^3)=0, \qquad \sigma\in\{+1,-1\},$$5 with $$v_t+\partial_x(\Delta v)+\sigma\,\partial_x(v^3)=0, \qquad \sigma\in\{+1,-1\},$$6, while small-data global well-posedness and scattering hold at the critical endpoint $$v_t+\partial_x(\Delta v)+\sigma\,\partial_x(v^3)=0, \qquad \sigma\in\{+1,-1\},$$7; the results are sharp in the sense of $$v_t+\partial_x(\Delta v)+\sigma\,\partial_x(v^3)=0, \qquad \sigma\in\{+1,-1\},$$8 flows (Correia et al., 31 Jul 2025).

On semiperiodic domains, the thresholds are different. For the cylinder $$v_t+\partial_x(\Delta v)+\sigma\,\partial_x(v^3)=0, \qquad \sigma\in\{+1,-1\},$$9, a Bourgain-space trilinear estimate first yielded local well-posedness for $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$00, and a later almost optimal linear $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$01 estimate plus bilinear refinements lowered the threshold to $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$02 for mZK (Mezher, 2024, Nowicki-Koth, 18 Jun 2025). On the torus $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$03, the complex-valued equation is locally well-posed for $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$04, but the flow map is not uniformly continuous on bounded subsets of $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$05 (Kenig et al., 28 May 2026).

In three dimensions the critical index becomes $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$06, so the equation is no longer mass-critical. Grünrock proved local well-posedness on $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$07 for $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$08 and, using mass and energy conservation, small-data global well-posedness in $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$09 (Grünrock, 2013). More generally, small-data global well-posedness at the scaling-critical regularity $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$10 for all $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$11 was established using $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$12 spaces (Kinoshita, 2019).

These statements should not be read as a single monotone regularity hierarchy. They mix different geometries, distinct nonlinear structures, and different solution spaces.

5. Nonlinear dynamics: multi-solitons, blow-up, and scattering

The asymptotic soliton theory of the two-dimensional real mZK equation is unusually complete for a mass-critical model. Given $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$13 distinct velocities

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$14

shifts $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$15, and signs $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$16, define

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$17

Valet proved that there exist $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$18 and a unique multi-soliton solution $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$19 such that

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$20

and more precisely, for every $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$21,

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$22

for $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$23, with $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$24 depending only on the soliton speeds. The proof combines backward construction, modulation of positions and speeds, localized mass and modified energy monotonicity, and coercivity after projection onto translation modes and the negative mode (Valet, 2020).

The focusing equation exhibits mass concentration at blow-up. If

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$25

blows up in finite time $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$26 in $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$27 with $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$28, and if $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$29 satisfies

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$30

then there exists a center $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$31 such that

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$32

For $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$33, the limsup strengthens to liminf, and the stronger liminf conclusion also holds for $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$34 under an additional upper bound on the blow-up rate (Bhattacharya, 2020).

At the opposite end of the dynamical spectrum lies scattering. For small and localized initial data satisfying

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$35

Anjolras proved that the global solution of the two-dimensional real mZK equation scatters: the profile $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$36 converges in $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$37 as $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$38, where $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$39 is the linearized Airy-type operator after diagonalization by a linear change of variables. The proof uses space-time resonances and the factorization of the linear kernel into one-dimensional Airy kernels (Anjolras, 20 Jun 2025). The later critical-space theory in $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$40 contains a small-data scattering result at the scale-invariant level, but in a different function-space architecture and without the weighted $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$41 localization assumption (Correia et al., 31 Jul 2025).

These results show that the same mass-critical equation supports three sharply distinct regimes: asymptotic resolution into prescribed soliton packets, concentration of at least the ground-state mass at finite-time blow-up, and linear scattering for sufficiently small data.

6. Analyticity, weighted decay, and current directions

Beyond Sobolev well-posedness, recent work has focused on how regularity is distributed in frequency and space. In Gevrey spaces

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$42

the two-dimensional symmetrized mZK equation preserves the initial radius of spatial analyticity $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$43 on a short interval for both focusing and defocusing signs. In the defocusing case, if $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$44 with $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$45, then for any $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$46 the global analyticity radius satisfies

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$47

The focusing case is excluded from the global Gevrey analysis because the corresponding Gevrey-weighted energy is not coercive (Baldasso et al., 2024).

Weighted decay and regularity are also coupled in anisotropic Sobolev spaces. For the symmetrized flow analyzed in

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$48

persistence holds when

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$49

Moreover, if at two distinct times $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$50 the solution obeys

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$51

then the solution gains regularity

$$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$52

subject to the parameter ranges established in that paper. The gain is controlled by the weaker directional decay, not the stronger one (Bustamante et al., 2024).

Several open problems recur across these developments. Three-dimensional mZK multi-solitons remain untreated in the asymptotic theory, and the available discussion suggests that the supercritical character would require different, likely topological, arguments and may not lead to uniqueness in the same form as in two dimensions (Valet, 2020). Large-data scattering at critical regularity is open in the real two-dimensional equation (Correia et al., 31 Jul 2025). For the complex torus model, lowering the threshold $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$53, extending local theory to global dynamics, and understanding invariant measures remain open directions (Kenig et al., 28 May 2026).

A recurring misconception is that “the” modified Zakharov–Kuznetsov equation has a single settled critical theory. The current literature instead points to a more differentiated picture: the standard real two-dimensional whole-space equation is now understood at the level of sharp local $$\partial_t u+\partial_x\Delta u+\partial_x(u^3)=0, \qquad (t,x,y)\in\mathbb{R}\times\mathbb{R}^2,$$54 theory, low-regularity global existence, critical small-data scattering in anisotropic spaces, soliton resolution within prescribed multi-soliton classes, and blow-up concentration; yet each of these advances depends on a specific functional setting, and no single framework presently subsumes all of them.