Benjamin–Ono Equation Overview

• The Benjamin–Ono equation is a nonlocal dispersive PDE modeling one-dimensional internal gravity waves with rich integrability and soliton dynamics.
• It is rigorously analyzed in Sobolev spaces, employing gauge transforms and dyadic frequency analysis to establish optimal well-posedness.
• Its integrable structure enables explicit soliton formulas, inverse scattering reconstructions, and conservation laws that underpin modern PDE research.

The Benjamin–Ono equation is a completely integrable, nonlocal, dispersive partial differential equation modeling one-dimensional internal gravity waves at the interface of two fluids with different densities. Its mathematical structure, rich integrability, and sharp analysis have established it as a canonical model in the theory of nonlocal dispersive equations, with deep links to nonlinear Fourier analysis, inverse scattering, and soliton dynamics.

1. Equation, Physical Context, and Fundamental Properties

The real-valued Benjamin–Ono (BO) equation on the real line,

$$u_t + H u_{xx} + u\,u_x = 0, \qquad u(0,x) = u_0(x),$$

features the Hilbert transform

$$Hf(x) = \mathrm{p.v.}~\frac{1}{\pi} \int \frac{f(y)}{x-y} \, dy,$$

which acts as a Fourier multiplier $$\widehat{Hf}(\xi) = -i\,\mathrm{sgn}(\xi)\,\hat f(\xi)$$. The scaling symmetry $u_\lambda(t,x) = \lambda u(\lambda^2 t, \lambda x)$ leaves the equation invariant, identifying $s=-2$ as the scale-critical Sobolev exponent.

The BO equation was derived to describe the propagation of long internal waves in deep, stratified fluids and is physically interpreted as a model for internal waves with weak nonlinearity and weak, nonlocal (non-KdV type) dispersion (Ifrim et al., 2017).

2. Well-Posedness and Functional Analysis

2.1. Sobolev Spaces and Sharp Thresholds

The initial value problem for BO has been rigorously analyzed in Sobolev spaces. For any $s > -\frac{1}{2}$ and either on $\mathbb{R}$ or $\mathbb{T}$, the problem is globally well-posed in $H^s$: $$q_0 \in H^s \ \Rightarrow\ \exists!~q \in C(\mathbb{R}; H^s), \text{ with } q(0) = q_0.$$ The flow map $$Hf(x) = \mathrm{p.v.}~\frac{1}{\pi} \int \frac{f(y)}{x-y} \, dy,$$0 is continuous $$Hf(x) = \mathrm{p.v.}~\frac{1}{\pi} \int \frac{f(y)}{x-y} \, dy,$$1 (Killip et al., 2023). The sharp threshold $$Hf(x) = \mathrm{p.v.}~\frac{1}{\pi} \int \frac{f(y)}{x-y} \, dy,$$2 is critical and optimal. The proof utilizes a family of commuting $$Hf(x) = \mathrm{p.v.}~\frac{1}{\pi} \int \frac{f(y)}{x-y} \, dy,$$3–flows corresponding to auxiliary Hamiltonians, a nonlinear gauge transformation $$Hf(x) = \mathrm{p.v.}~\frac{1}{\pi} \int \frac{f(y)}{x-y} \, dy,$$4, and contraction in negative regularity.

2.2. Analytic Tools

Key technical elements include:

• Normal-form and gauge transforms to remove derivative loss and high-low frequency resonance.
• Dyadic frequency analysis, Strichartz and bilinear $$Hf(x) = \mathrm{p.v.}~\frac{1}{\pi} \int \frac{f(y)}{x-y} \, dy,$$5-based norm bootstraps.
• Paradifferential reductions and commutator estimates (e.g., Coifman–Meyer lemma: $$Hf(x) = \mathrm{p.v.}~\frac{1}{\pi} \int \frac{f(y)}{x-y} \, dy,$$6) (Ifrim et al., 2017).

2.3. Unconditional Uniqueness

Unconditional uniqueness holds in $$Hf(x) = \mathrm{p.v.}~\frac{1}{\pi} \int \frac{f(y)}{x-y} \, dy,$$7 for $$Hf(x) = \mathrm{p.v.}~\frac{1}{\pi} \int \frac{f(y)}{x-y} \, dy,$$8, achieved by gauge transforms (Tao), refined Strichartz, and double normal-form analysis yielding uniqueness below $$Hf(x) = \mathrm{p.v.}~\frac{1}{\pi} \int \frac{f(y)}{x-y} \, dy,$$9 (Mosincat et al., 2021).

3. Integrability and Exact Structures

3.1. Lax Pair, Hierarchy, and Conservation Laws

The BO equation admits a nonlocal Lax pair on appropriate Hardy spaces: $$\widehat{Hf}(\xi) = -i\,\mathrm{sgn}(\xi)\,\hat f(\xi)$$0 (with $$\widehat{Hf}(\xi) = -i\,\mathrm{sgn}(\xi)\,\hat f(\xi)$$1 the Szegő projector), and the flow preserves the spectrum of $$\widehat{Hf}(\xi) = -i\,\mathrm{sgn}(\xi)\,\hat f(\xi)$$2. The generating function

$$\widehat{Hf}(\xi) = -i\,\mathrm{sgn}(\xi)\,\hat f(\xi)$$3

expands to encode all polynomial conservation laws: mass, momentum, energy, and higher Hamiltonians as coefficients in powers of $$\widehat{Hf}(\xi) = -i\,\mathrm{sgn}(\xi)\,\hat f(\xi)$$4 (Killip et al., 2023). This unification connects Fokas–Fuchssteiner vector-field recursion to operator invariants.

Negative-regularity conservation laws, derived from a renormalized perturbation determinant

$$\widehat{Hf}(\xi) = -i\,\mathrm{sgn}(\xi)\,\hat f(\xi)$$5

control and propagate $$\widehat{Hf}(\xi) = -i\,\mathrm{sgn}(\xi)\,\hat f(\xi)$$6-norms for $$\widehat{Hf}(\xi) = -i\,\mathrm{sgn}(\xi)\,\hat f(\xi)$$7 (Talbut, 2018).

3.2. Explicit Solution Formulas

Recent explicit solutions for the Cauchy problem on $$\widehat{Hf}(\xi) = -i\,\mathrm{sgn}(\xi)\,\hat f(\xi)$$8 and $$\widehat{Hf}(\xi) = -i\,\mathrm{sgn}(\xi)\,\hat f(\xi)$$9 express the evolution in terms of the spectral data of the Lax operator (Gerard, 2022): $u_\lambda(t,x) = \lambda u(\lambda^2 t, \lambda x)$0 where $u_\lambda(t,x) = \lambda u(\lambda^2 t, \lambda x)$1 is the adjoint of multiplication by $u_\lambda(t,x) = \lambda u(\lambda^2 t, \lambda x)$2 on Hardy space. This framework allows for a direct "integration by quadrature", providing a concrete nonlinear Fourier (Birkhoff) coordinate system (Gerard et al., 2019), with global action–angle variables for all $u_\lambda(t,x) = \lambda u(\lambda^2 t, \lambda x)$3 solutions on the torus.

3.3. Inverse Scattering and Jost Solutions

The direct and inverse Fokas–Ablowitz scattering theory has been fully developed for BO, rigorously constructing Jost solutions for the Lax operator, defining scattering data, establishing their analytic properties, and detailing the associated nonlocal Riemann–Hilbert problem for reconstruction of the potential (Wu, 2017).

4. Solitons, Multisolitons, and Resolution

4.1. One- and Multi-soliton Solutions

BO admits explicit soliton solutions: $u_\lambda(t,x) = \lambda u(\lambda^2 t, \lambda x)$4 Multi-soliton solutions are given in terms of determinants involving the soliton parameters and centers, and propagate via explicit modulation, with exact formulas for the $u_\lambda(t,x) = \lambda u(\lambda^2 t, \lambda x)$5-soliton manifold (Badreddine et al., 17 Sep 2025).

4.2. Orbital Stability

Recent work proves uniform orbital stability of $u_\lambda(t,x) = \lambda u(\lambda^2 t, \lambda x)$6-solitons in $u_\lambda(t,x) = \lambda u(\lambda^2 t, \lambda x)$7:

• Any data $u_\lambda(t,x) = \lambda u(\lambda^2 t, \lambda x)$8-close to the $u_\lambda(t,x) = \lambda u(\lambda^2 t, \lambda x)$9-soliton manifold remains $s=-2$0-close for all time, up to translations.
• The proof exploits a new variational characterization: the generating functional $s=-2$1 is minimized precisely by multisolitons, with the Wu identity controlling the discrete spectrum of the Lax operator (Badreddine et al., 17 Sep 2025).
• Concentration–compactness and molecular approximation ensure stability at low regularity.

4.3. Soliton Resolution

The soliton resolution conjecture for BO is resolved: for initial data $s=-2$2 with sufficient decay, the solution decomposes as

$s=-2$3

with soliton parameters $s=-2$4 determined by negative eigenvalues of the Lax operator and $s=-2$5 the radiation profiles from the continuous spectrum. Orthogonality, stationary phase, and detailed control of the distorted Fourier transform underpin this asymptotic completeness (Gassot et al., 15 Jan 2026).

5. Dispersive Decay, Long-Time Dynamics, and Measures

5.1. Dispersive Dynamics

For small, localized data, solutions exhibit nearly completely dispersive behavior up to exponentially long times $s=-2$6: $s=-2$7 with stronger decay off maximal velocity $s=-2$8 (Ifrim et al., 2017).

5.2. Invariant Measures, Recurrence, and Almost-Periodicity

Higher-order Gibbs-type Gaussian measures $s=-2$9 are rigorously constructed and shown to be invariant under the BO flow on the torus for $s > -\frac{1}{2}$0 (even), providing almost-sure global bounds and establishing Poincaré recurrence for almost every initial data. These measures concentrate on Sobolev regularities corresponding to the order of the invariant (Tzvetkov et al., 2012).

On $s > -\frac{1}{2}$1, the existence of global Birkhoff coordinates and the complete integrability structure ensures that all $s > -\frac{1}{2}$2-data generate almost-periodic orbits under the flow (Gerard et al., 2019).

6. Extensions, Generalizations, and Related Models

6.1. Dispersion Generalizations and Multi-Dimensional BO

Generalizations to dispersion exponents $s > -\frac{1}{2}$3 yield the so-called dispersion-generalized BO (dgBO) equation: $s > -\frac{1}{2}$4 Local well-posedness in $s > -\frac{1}{2}$5 is achieved down to $s > -\frac{1}{2}$6, with a unified approach using pseudodifferential gauge transforms and paradifferential normal-form reductions (Ai et al., 2024). In higher dimensions, the equation

$s > -\frac{1}{2}$7

with $s > -\frac{1}{2}$8 the Riesz transform, is locally well-posed for $s > -\frac{1}{2}$9 in 2D; the critical exponent is $\mathbb{R}$0 (Linares et al., 2019).

6.2. Bidirectional and Rotational Variants

The bidirectional Benjamin–Ono (2BO) equation extends the integrable structure to two interacting fields, embedding into a real reduction of the modified KP hierarchy, and possessing multi-phase solutions and solitons (Abanov et al., 2008).

The rotation-generalized Benjamin–Ono equation,

$\mathbb{R}$1

models the effect of planetary rotation and admits solitary waves with variational characterization, algebraic decay, and a stability criterion governed by sign of the second derivative of the action (Esfahani et al., 2011).

7. Zero-Dispersion and Long-Time Limits

The zero-dispersion limit (as $\mathbb{R}$2) of BO,

$\mathbb{R}$3

exhibits a transition to a multivalued solution structure reminiscent of Burgers' equation. For suitable initial data, the weak limit is the signed sum of branches of the multivalued inviscid Burgers solution. This rigorous limiting behavior is established both on $\mathbb{R}$4 and the torus, with explicit connection to the spectral data of the Lax operator (Gérard, 2023, Gassot, 2021).

Table: Major Theoretical Results for the Benjamin–Ono Equation

Result	Statement/Threshold	Reference
Global well-posedness in $\mathbb{R}$5	$\mathbb{R}$6	(Killip et al., 2023)
Orbital stability of multisolitons	$\mathbb{R}$7	(Badreddine et al., 17 Sep 2025)
Soliton resolution in $\mathbb{R}$8	Asymptotic $\mathbb{R}$9-soliton + radiation	(Gassot et al., 15 Jan 2026)
Negative-regularity conservation	$\mathbb{T}$0	(Talbut, 2018)
Invariant Gibbs measures on $\mathbb{T}$1	Measures supported in $\mathbb{T}$2, $\mathbb{T}$3	(Tzvetkov et al., 2012)
Dispersive decay for small/localized data	$\mathbb{T}$4	(Ifrim et al., 2017)
Unconditional uniqueness	$\mathbb{T}$5	(Mosincat et al., 2021)

• (Ifrim et al., 2017): Well-posedness and dispersive decay of small data solutions for the Benjamin-Ono equation
• (Badreddine et al., 17 Sep 2025): Orbital stability of Benjamin–Ono multisolitons
• (Gassot et al., 15 Jan 2026): A proof of the soliton resolution conjecture for the Benjamin–Ono equation
• (Killip et al., 2023): Sharp well-posedness for the Benjamin–Ono equation
• (Talbut, 2018): Low regularity conservation laws for the Benjamin-Ono equation
• (Tzvetkov et al., 2012): Invariant measures and long time behaviour for the Benjamin-Ono equation
• (Gerard, 2022): An explicit formula for the Benjamin–Ono equation
• (Gerard et al., 2019): On the integrability of the Benjamin-Ono equation on the torus
• (Abanov et al., 2008): Integrable hydrodynamics of Calogero-Sutherland model: Bidirectional Benjamin-Ono equation
• (Esfahani et al., 2011): Solitary waves of the rotation-generalized Benjamin-Ono equation
• (Ai et al., 2024): The dispersion generalized Benjamin-Ono equation
• (Gérard, 2023, Gassot, 2021): Zero-dispersion limits

---

The Benjamin–Ono equation occupies a central position in the modern analysis of integrable, nonlocal dispersive systems: its long-time soliton dynamics, hierarchy of conservation laws, explicit nonlinear Fourier transform, and connection to statistical mechanics and random data dynamics collectively exemplify the sharpest known synthesis of PDE, harmonic analysis, and spectral theory in the nonlocal regime.