Ladyzhenskaya–Prodi–Serrin Condition

• Ladyzhenskaya–Prodi–Serrin condition is a mixed space–time integrability criterion that defines the threshold for regularity and uniqueness in weak solutions of Navier–Stokes and related PDEs.
• It employs scale-invariant exponents to balance diffusion and rough transport, marking the borderline between subcritical, critical, and supercritical regimes in various fluid and stochastic models.
• Generalizations of this condition extend to coupled hydrodynamic systems, stochastic dynamics, and transport problems, offering practical insights into strong well-posedness and singular behavior.

The Ladyzhenskaya–Prodi–Serrin condition is a mixed space–time integrability criterion that is classical in the regularity theory of the incompressible Navier–Stokes equations and now functions as a structural threshold across parabolic PDE, stochastic analysis, kinetic equations, and coupled fluid models. In its standard three-dimensional Navier–Stokes form it requires

$$u \in L^q(0,T;L^p(\mathbb{R}^3)), \qquad \frac{2}{q}+\frac{3}{p}\le 1,$$

while in diffusion-with-drift problems it appears as

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}\le 1$$

or, in many subcritical arguments, with strict inequality. Its role is to mark the borderline at which diffusion or viscosity can still dominate rough transport, yielding regularity, uniqueness, stability, or strong well-posedness (Veiga et al., 2019, Kinzebulatov et al., 2021, Li et al., 2017).

1. Classical formulation and canonical variants

In the three-dimensional incompressible Navier–Stokes setting, the classical Ladyzhenskaya–Prodi–Serrin criterion is formulated for a Leray–Hopf weak solution $u$ by requiring

$$u \in L^p(0,T;L^r(\Omega)), \qquad \frac{2}{p}+\frac{3}{r}=1,\qquad 3\le r\le \infty,$$

and under this condition the solution is smooth; uniqueness and energy equality then follow from smoothness (Veiga et al., 2019). The limiting endpoint $u\in L^\infty(0,T;L^3)$ lies on the same critical line and is explicitly identified as the marginal case of the condition (Choe et al., 2016).

A distinct but parallel formulation arises for stochastic differential equations and parabolic transport problems with singular drift. There the drift plays the role of the advecting field, and the standard condition becomes

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}<1,$$

in the strictly subcritical regime, or

$\frac{d}{p}+\frac{2}{q}\le 1$

in critical versions (Kinzebulatov et al., 2021, Neves et al., 2013). In this setting the condition is imposed on the coefficient of a diffusion or transport equation rather than on the unknown velocity itself.

The literature represented here uses several distinct variants of the condition. Some works insist on the strict inequality $<1$, especially when a Zvonkin transform or maximal $L^q_tL^p_x$ regularity estimate is needed (Li et al., 2017, Lê et al., 2021). Other works treat the borderline critical case $\le 1$ and regard it as the natural threshold for regularity, uniqueness, or strong solvability (Kinzebulatov et al., 2021). There are also componentwise and anisotropic analogues, where only selected components of a vector field are assumed to satisfy an LPS-type integrability bound rather than the full field (Wang, 2014, Larios et al., 2016).

2. Scaling, criticality, and threshold structure

The organizing principle behind the Ladyzhenskaya–Prodi–Serrin condition is parabolic scaling. For the heat operator, the natural transformation is

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}\le 1$$0

In the Fokker–Planck setting with drift $$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}\le 1$$1, the quantity

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}\le 1$$2

measures the scaling of the $$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}\le 1$$3 norm of the drift relative to diffusion: $$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}\le 1$$4 is subcritical, $$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}\le 1$$5 is critical, and $$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}\le 1$$6 is supercritical. In the subcritical regime, small scales diminish the effective size of the drift, which is precisely why diffusion can regularize rough transport (Li et al., 2017).

For the Navier–Stokes equations, the critical line

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}\le 1$$7

is the scale-invariant one under

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}\le 1$$8

This is the geometric meaning of the classical LPS condition: it is neither merely an integrability assumption nor merely a technical hypothesis, but a criticality statement tied to the scaling of the equation (Choe et al., 2016, Li et al., 2022).

Later developments preserve this scaling logic while changing the underlying noise or blowup rescaling. For additive fractional Brownian noise with Hurst parameter $$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}\le 1$$9, the Brownian threshold is replaced by

$u$0

and for symmetric $u$1-stable Lévy noise by

$u$2

These are presented as generalized or extended Krylov–LPS conditions adapted to the scaling of the noise (Butkovsky et al., 2023). In a different direction, Type II blowup analysis for Navier–Stokes under an Euler-type rescaling uses a generalized relation

$u$3

which is explicitly described as looking like a generalization of the Ladyzhenskaya–Prodi–Serrin condition (Seregin, 11 Jul 2025).

This scaling perspective also clarifies why endpoint and supercritical questions are delicate. Several works represented here treat only the strictly subcritical side, and some state explicitly that the borderline case is not handled (Li et al., 2017). Others show that the critical class remains meaningful but becomes analytically much more rigid (Kinzebulatov et al., 2021). Still others prove non-existence or non-uniqueness once the generalized threshold is crossed (Butkovsky et al., 2023, Miao et al., 24 May 2026).

3. Navier–Stokes regularity, endpoint phenomena, and neighboring criteria

In Navier–Stokes theory, the classical function of the LPS condition is to upgrade a weak solution to a smooth one. That role remains central in recent formulations, but several neighboring criteria sharpen the picture. The Lions–Prodi condition

$u$4

implies energy equality, and Shinbrot’s criterion

$u$5

is shown to follow from Lions–Prodi by interpolation. An extension to $u$6 with

$u$7

still yields energy equality, though it is more restrictive than the classical Shinbrot relation and does not coincide with the LPS regularity regime except at the endpoints $u$8 and $u$9 (Veiga et al., 2019).

At the endpoint $$u \in L^p(0,T;L^r(\Omega)), \qquad \frac{2}{p}+\frac{3}{r}=1,\qquad 3\le r\le \infty,$$0, the Escauriaza–Seregin–Šverák theorem is recalled as the marginal extension of the LPS regularity criterion (Choe et al., 2016). A weaker assumption,

$$u \in L^p(0,T;L^r(\Omega)), \qquad \frac{2}{p}+\frac{3}{r}=1,\qquad 3\le r\le \infty,$$1

does not imply full regularity in the quoted work, but it yields a new local regularity criterion and the conclusion that at any singular time there are at most finitely many blowup points (Choe et al., 2016). This places weak-$$u \in L^p(0,T;L^r(\Omega)), \qquad \frac{2}{p}+\frac{3}{r}=1,\qquad 3\le r\le \infty,$$2 strictly beyond the endpoint LPS space while preserving nontrivial geometric control on singularities.

A further localization replaces global $$u \in L^p(0,T;L^r(\Omega)), \qquad \frac{2}{p}+\frac{3}{r}=1,\qquad 3\le r\le \infty,$$3 norms by uniformly local norms

$$u \in L^p(0,T;L^r(\Omega)), \qquad \frac{2}{p}+\frac{3}{r}=1,\qquad 3\le r\le \infty,$$4

and assumes

$$u \in L^p(0,T;L^r(\Omega)), \qquad \frac{2}{p}+\frac{3}{r}=1,\qquad 3\le r\le \infty,$$5

together with

$$u \in L^p(0,T;L^r(\Omega)), \qquad \frac{2}{p}+\frac{3}{r}=1,\qquad 3\le r\le \infty,$$6

Under these hypotheses one obtains an a priori $$u \in L^p(0,T;L^r(\Omega)), \qquad \frac{2}{p}+\frac{3}{r}=1,\qquad 3\le r\le \infty,$$7 estimate and hence regularity. This is a localized Serrin–Prodi–Ladyzhenskaya criterion with the same critical exponent relation but different spatial geometry (Li et al., 2022).

The LPS framework has also been used as a target in computational studies of extreme Navier–Stokes behavior. One recent work formulates PDE-constrained optimization problems that maximize

$$u \in L^p(0,T;L^r(\Omega)), \qquad \frac{2}{p}+\frac{3}{r}=1,\qquad 3\le r\le \infty,$$8

or $$u \in L^p(0,T;L^r(\Omega)), \qquad \frac{2}{p}+\frac{3}{r}=1,\qquad 3\le r\le \infty,$$9, over periodic initial data. No unbounded growth of the LPS quantities is detected, but the maximizing flows enter transient regimes in which the growth rates of $u\in L^\infty(0,T;L^3)$0 and of enstrophy are consistent with finite-time singularity formation, though not sustained long enough for singularities to occur (Ramírez et al., 14 Apr 2026).

4. Coupled hydrodynamic systems and componentwise criteria

The LPS principle extends to several coupled systems, but often in modified or componentwise form. For the three-dimensional MHD equations, one continuation criterion assumes

$u\in L^\infty(0,T;L^3)$1

and imposes an LPS-type condition only on the horizontal magnetic components $u\in L^\infty(0,T;L^3)$2, either directly,

$u\in L^\infty(0,T;L^3)$3

or at the derivative level,

$u\in L^\infty(0,T;L^3)$4

Under these assumptions no blowup occurs up to the terminal time (Wang, 2014). The structural point is that the velocity is controlled in the critical endpoint $u\in L^\infty(0,T;L^3)$5, while only selected magnetic components satisfy an LPS-type bound.

For the three-dimensional MHD–Boussinesq system without thermal diffusion, the paper cited proves a Prodi–Serrin-type regularity criterion in terms of only two velocity and two magnetic components: $u\in L^\infty(0,T;L^3)$6 with

$u\in L^\infty(0,T;L^3)$7

This is explicitly presented as the first Prodi–Serrin-type criterion for a hydrodynamic system that is not fully dissipative (Larios et al., 2016). The exponent relation differs from the Navier–Stokes line because the system is coupled and the temperature equation lacks diffusion.

In a fluid–structure–polymer setting, the classical three-dimensional LPS condition is transplanted to a moving domain. For the solute–solvent–structure Oldroyd-B system, if

$u\in L^\infty(0,T;L^3)$8

and simultaneously

$u\in L^\infty(0,T;L^3)$9

then a global weak solution becomes a unique global strong solution. The paper stresses that no additional LPS-type assumption is imposed on the polymer density or extra stress tensor (Mensah, 2024).

Sharpness on the supercritical side also appears in MHD. For the two-dimensional viscous and resistive MHD equations, weak solutions are proved non-unique in

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}<1,$$0

which is interpreted as showing the sharpness of the LPS condition at the endpoint $$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}<1,$$1. The same work states analogous non-uniqueness consequences for Navier–Stokes and for large $$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}<1,$$2 data (Miao et al., 24 May 2026).

5. Parabolic equations, stochastic dynamics, and transport by rough drifts

Outside fluid equations proper, the LPS condition governs rough drifts in linear and stochastic parabolic problems. In the non-degenerate Fokker–Planck equation

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}<1,$$3

one cited work assumes

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}<1,$$4

and derives quantitative stability estimates for solutions in a logarithmic Kantorovich–Rubinstein distance and in Wasserstein distance. The argument combines Trevisan’s superposition principle with a Zvonkin transform obtained from a backward parabolic equation whose solvability and gradient bound rely crucially on the subcritical LPS assumption (Li et al., 2017).

For singular SDEs, the classical subcritical condition

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}<1,$$5

appears in the Krylov–Röckner theory, while the critical class

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}<1,$$6

is singled out in later work as the critical Ladyzhenskaya–Prodi–Serrin class. One paper cited embeds this class into a larger form-bounded framework and proves unique weak solvability, a Feller evolution family, and weak uniqueness for the associated SDE (Kinzebulatov et al., 2021).

A different extension concerns weak existence below the LPS threshold. For uniformly elliptic SDEs with singular drift, one work reproves Krylov’s weak existence regime

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}<1,$$7

by means of a partial Zvonkin transform and allows uniformly local Lebesgue spaces and linear growth at infinity (Galeati, 2023). This separates the integrability threshold for weak existence from the classical LPS threshold associated with strong well-posedness and uniqueness.

The same condition also regularizes transport-type SPDEs. For the stochastic divergence-free continuity equation driven by Stratonovich Brownian noise, if

$$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}<1,$$8

and $$b\in L^q(0,T;L^p(\mathbb{R}^d)), \qquad \frac{d}{p}+\frac{2}{q}<1,$$9, then weak $\frac{d}{p}+\frac{2}{q}\le 1$0-solutions are unique and admit the representation

$\frac{d}{p}+\frac{2}{q}\le 1$1

where $\frac{d}{p}+\frac{2}{q}\le 1$2 is the stochastic flow associated with the SDE $\frac{d}{p}+\frac{2}{q}\le 1$3 (Neves et al., 2013). In numerical analysis, the same strict LPS regime underlies strong convergence results for explicit tamed Euler–Maruyama schemes with multiplicative noise, and the paper cited states that suitable approximations of the drift can deliver rates arbitrarily close to the benchmark $\frac{d}{p}+\frac{2}{q}\le 1$4 rate (Lê et al., 2021).

6. Generalizations, singular interactions, and sharpness beyond the classical setting

The LPS condition has been generalized in several orthogonal directions. For the Dean–Kawasaki equation with singular interaction kernel $\frac{d}{p}+\frac{2}{q}\le 1$5 on $\frac{d}{p}+\frac{2}{q}\le 1$6, the kernel is assumed to satisfy

$\frac{d}{p}+\frac{2}{q}\le 1$7

This is explicitly called the Ladyzhenskaya–Prodi–Serrin condition for $\frac{d}{p}+\frac{2}{q}\le 1$8. Under this assumption the paper proves existence of probabilistic weak renormalized kinetic solutions; an additional condition on $\frac{d}{p}+\frac{2}{q}\le 1$9 yields pathwise uniqueness and strong well-posedness (Wang et al., 2022).

Noise-adapted LPS analogues also arise in rough-noise SDEs. For additive fractional Brownian noise, weak existence is proved under

$<1$0

and for symmetric $<1$1-stable Lévy noise under

$<1$2

The same source constructs a counterexample when the fractional-Brownian inequality is reversed, which is stated as demonstrating optimality on the existence side (Butkovsky et al., 2023). This suggests that the Ladyzhenskaya–Prodi–Serrin philosophy persists after the Brownian scaling is replaced by a different noise scaling.

In blowup theory, an additional assumption

$<1$3

is described as looking like a generalization of the LPS condition and is used, together with scale-weighted bounds, to exclude a class of Type II blowup scenarios for suitable weak solutions of Navier–Stokes (Seregin, 11 Jul 2025). By contrast, sharp non-uniqueness results for 2D MHD show that once one moves to the supercritical side of the endpoint $<1$4, uniqueness can fail (Miao et al., 24 May 2026). Taken together, these works depict LPS not merely as a sufficient condition but as a threshold whose sharpness can be probed from both sides.

A final nearby development is the heat equation with singular drift under Morrey assumptions. There maximum modulus estimates are proved for all $<1$5 satisfying

$<1$6

and one section gives an application to drifts satisfying the critical LPS condition

$<1$7

The LPS case appears there as a critical special case within a broader Morrey-drift framework (Krylov, 13 Apr 2026).

Across these disparate settings, the common theme is stable: the Ladyzhenskaya–Prodi–Serrin condition identifies the borderline at which transport remains subordinate to diffusion or viscosity. In Navier–Stokes it is a regularity and uniqueness criterion; in stochastic analysis it governs strong solvability, weak existence, or stochastic regularization by noise; in coupled and nonlocal systems it persists in modified, componentwise, or kernel-based forms; and in sharpness results it emerges as the boundary separating provable regularity from supercritical phenomena such as non-uniqueness or loss of control (Li et al., 2017, Mensah, 2024, Miao et al., 24 May 2026).