A Fourier spectral method for the cutoff Boltzmann equation: Convergence analysis and numerical simulation
Yanzhi Gui, Ling-Bing He, Liu Liu · 2026
Read A Fourier spectral method for the cutoff Boltzmann equation: Convergence analysis and numerical simulation by Yanzhi Gui, Ling-Bing He, Liu Liu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-26220.txt.
Introduction
Gui, He, and Liu present a Fourier spectral method for the cutoff Boltzmann equation in a uniform gas. The equation tracks how gas particles collide and spread out in speed. Their method handles both Maxwellian and hard interaction potentials. They prove finite-time error bounds in a weighted L2 norm for the whole space. The bounds show in clear terms how the error depends on the velocity cutoff and the number of Fourier modes used.
Catalog record
| CREATOR | Yanzhi Gui, Ling-Bing He, Liu Liu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 164200 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.26220
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Fourier spectral method for the cutoff Boltzmann equation: Convergence analysis and numerical simulation Yanzhi Gui, Ling-Bing He, Liu Liu Abstract . This work addresses a central challenge in the numerical analysis of the cutoff spatially homogeneous Boltzmann equation: the development of rigorously justified, accurate numerical schemes. We present (i) a novel Fourier spectral method for the equation with Maxwellian and hard potentials, (ii) the derivation of the first rigorous error estimates for the proposed schemes. Comprehensive numerical experiments validate the theory, confirming the predicted accuracy and illustrating the method’s capability to capture solution dynamics, including the approach to equilibrium. The study thus provides a complete framework–from theoretical analysis to practical implementation–for the reliable computation of solutions to this foundational kinetic model. 2020 Mathematics Subject Classification 35Q20, 65M15, 65M70, 65M12, 35R09, 82C40 1. Introduction The main purpose of this work is to conduct a thorough numerical investigation of the spatially homogeneous Boltzmann equation for two important classes of interactions: Maxwellian potentials and hard potentials with angular cutoff. This involves: 1. The development of a novel numerical approximation method tailored to these potential types; 2. Rigorous error estimates for the proposed numerical scheme, establishing its accuracy and reliability; 3. Implementation of simulations to illustrate the method’s performance and to explore quantitative properties of solutions. This study is significant because it advances the computational toolkit for the Boltzmann equation—a fundamental model in kinetic theory that describes the evolution of particle systems through binary collisions. Obtaining accurate numerical solutions is essential for analyzing complex phenomena in rarefied gas dynamics, plasma physics, and other areas of statistical physics where analytical solutions are scarce or unavailable. By developing a novel numerical method with rigorous error analysis, this work provides a reliable framework for simulating the behavior and fundamental properties of solutions, thereby bridging theoretical kinetic theory with practical computational science. 1.1. A brief review on the Boltzmann equation The Boltzmann equation, formulated by Ludwig Boltzmann in the 1870s, governs the time evolution of the particle distribution function f(t,x,v) in phase space, representing the probability density of finding a particle at position x with velocity v at time t . In the absence of external forces and spatial inhomogeneity, the spatially homogeneous Boltzmann equation reads: (1.1) \frac{\partial f}{\partial t}(t,v)=Q(f,f)(t,v),\quad v\in\mathbb{R}^{d}, where the collision operator Q is a bilinear integral operator that accounts for binary elastic collisions between particles. It can be expressed in its gain-loss form as: Q(g,f)(v)=\int_{\mathbb{R}^{d}}\int_{\mathbb{S}^{d-1}}B(|v-v_{*}|,\cos\theta)\big[g(v^{\prime}_{*})f(v^{\prime})-g(v_{*})f(v)\big]\,d\sigma\,dv_{*}=:Q^{+}(g,f)-Q^{-}(g,f). The term Q^{+}(g,f) represents the gain term, modeling particles that enter the velocity state v after a collision, while Q^{-}(g,f) represents the loss term, modeling particles that leave velocity v due to a collision. Several remarks are in order to clarify the components of this operator: (1) Here v,v_{*} are the pre-collision velocities of two particles, and v^{\prime},v^{\prime}_{*} are their corresponding post-collision velocities. These are determined by the conservation laws of momentum and kinetic energy for elastic collisions. More precisely, they satisfy the following parametric relations: v^{\prime}=\frac{v+v_{*}}{2}+\frac{\left|v-v_{*}\right|}{2}\sigma,\quad v_{*}^{\prime}=\frac{v+v_{*}}{2}-\frac{\left|v-v_{*}\right|}{2}\sigma,\quad\sigma\in\mathbb{S}^{d-1}, where the unit vector \sigma parameterizes all possible post-collision directions for a given relative velocity. (2) The angle \theta is the deflection angle, defined by the relation \cos\theta:=\frac{v-v_{*}}{|v-v_{*}|}\cdot\sigma , where \cdot denotes the Euclidean dot product. This angle characterizes the scattering direction. (3) The function B is the collision kernel or cross-section, which encodes the microscopic details of the interaction potential between particles. It is a non-negative function of the relative speed |v-v_{*}| and the scattering angle \theta . In this work, we consider a specific class of kernels often used in theoretical and numerical studies: B\left(\left|v-v_{*}\right|,\cos\theta\right)=\Phi(|v-v_{*}|)\,b(\cos\theta),\quad\text{with}\quad\Phi(|v-v_{*}|)=|v-v_{*}|^{\gamma}\,\text{and}\,b\equiv 1. where \gamma is a parameter in [0,1] . The function b(\cos\theta) , which governs the angular scattering, is typically assumed to be integrable over the sphere \mathbb{S}^{d-1} (this is known as Grad’s angular cut-off assumption). For simplicity in our numerical analysis and simulations, we often take the isotropic case b\equiv 1 . Remark 1.1 . Two key families of collision kernels considered in this work, which fall under the general form \Phi(|v-v_{*}|)b(\cos\theta) , are: • Maxwellian molecules: In this idealized case, the kernel’s dependence on relative speed vanishes. We have \gamma=0 , so that \Phi(|v-v_{*}|)=1 . Thus, B=b(\cos\theta) , independent of the relative speed. • Hard-sphere model: This corresponds to the case \gamma=1 , where the collision frequency is proportional to the relative speed. Thus, the kernel is B=|v-v_{*}| . We note that the standard hard-sphere model also has a specific angular dependence b(\cos\theta)\propto|\cos\theta| , but under the cut-off assumption, b is taken to be an integrable function, often a constant for simplicity. Models with \gamma\in(0,1) are typically referred to as hard potentials (with cut-off). The Maxwellian case ( \gamma=0 ) often admits analytical simplifications, while the hard-sphere/hard potential cases ( \gamma>0 ) present greater numerical challenges due to the velocity dependence of the collision frequency, making them crucial benchmarks for any proposed numerical method. The fundamental properties of the spatially homogeneous Boltzmann equation underpin both its physical relevance and the mathematical challenges it presents. These properties are summarized as follows: 1.1.1. Conservation Laws Due to the microscopic conservation of mass, momentum, and energy during binary elastic collisions, solutions to the Boltzmann equation formally conserve these macroscopic quantities. For any solution f(t,v) with initial data f_{0}(v) , the following moments remain constant in time: (1.2) \int_{\mathop{\mathbb{R}\kern 0.0pt}\nolimits^{3}}f(t,v)\,\varphi(v)\,dv=\int_{\mathop{\mathbb{R}\kern 0.0pt}\nolimits^{3}}f_{0}(v)\,\varphi(v)\,dv\quad\text{for the collision invariants}\quad\varphi(v)=1,v_{1},v_{2},v_{3},\frac{|v|^{2}}{2}, where \varphi(v)=1,v_{i},\frac{|v|^{2}}{2} correspond to mass, the three components of momentum, and kinetic energy, respectively. This conservation law can be derived directly from the weak formulation of the collision operator. For any suitable test function \varphi(v) , we have: (1.3) \int_{\mathop{\mathbb{R}\kern 0.0pt}\nolimits^{3}}Q(f,f)(v)\,\varphi(v)\,dv=\frac{1}{2}\iint_{\mathop{\mathbb{R}\kern 0.0pt}\nolimits^{3}\times\mathop{\mathbb{R}\kern 0.0pt}\nolimits^{3}}\int_{\mathbb{S}^{2}}B\left(\left|v-v_{*}\right|,\sigma\right)ff_{*}\big(\varphi(v^{\prime})+\varphi(v^{\prime}_{*})-\varphi(v)-\varphi(v_{*})\big)\,d\sigma\,dv_{*}\,dv. The right-hand side vanishes identically precisely when \varphi(v) is a linear combination of the five collision invariants 1,v,|v|^{2} , due to the microscopic conservation laws \varphi(v^{\prime})+\varphi(v^{\prime}_{*})=\varphi(v)+\varphi(v_{*}) . 1.1.2. Entropy Structure and H-Theorem The Boltzmann equation exhibits a profound irreversible thermodynamic structure. Defining the macroscopic hydrodynamic fields associated with the distribution f(t,v) —namely, the density \rho_{f} , bulk velocity u_{f} , and temperature T_{f} —we have: (1.4) \rho_{f}(t):=\int_{\mathop{\mathbb{R}\kern 0.0pt}\nolimits^{3}}f(t,v)\,dv,\quad u_{f}(t):=\frac{1}{\rho_{f}(t)}\int_{\mathop{\mathbb{R}\kern 0.0pt}\nolimits^{3}}v\,f(t,v)\,dv,\quad T_{f}(t):=\frac{1}{3\rho_{f}(t)}\int_{\mathop{\mathbb{R}\kern 0.0pt}\nolimits^{3}}|v-u_{f}(t)|^{2}\,f(t,v)\,dv. These quantities are conserved according to ( 1.2 ). Given these fields, we can define the local Maxwellian (or Gibbs equilibrium state): (1.5) \mu_{\rho_{f},u_{f},T_{f}}(v)=\frac{\rho_{f}}{(2\pi T_{f})^{3/2}}\,\exp\left({-\frac{|v-u_{f}|^{2}}{2T_{f}}}\right), which shares the same mass, momentum, and energy as f . The system’s irreversibility is captured by the relative entropy (or Boltzmann H -functional relative to the local Maxwellian): (1.6) H(f|\mu_{\rho_{f},u_{f},T_{f}})(t):=\int_{\mathop{\mathbb{R}\kern 0.0pt}\nolimits^{3}}\left[f(t,v)\log\left(\frac{f(t,v)}{\mu_{\rho_{f},u_{f},T_{f}}(v)}\right)-f(t,v)+\mu_{\rho_{f},u_{f},T_{f}}(v)\right]\,dv. This functional is non-negative and serves as a measure of how far the distribution f is from its local equilibrium \mu_{\rho_{f},u_{f},T_{f}} . Boltzmann’s celebrated H-Theorem states that for solutions of the homogeneous equation, this relative entropy is non-increasing: (1.7) \frac{d}{dt}H(f|\mu_{\rho_{f},u_{f},T_{f}})(t)=-D(f(t,\cdot))\leq 0, where D(f) is the entropy dissipation rate , defined by: (1.8) D(f):=-\int_{\mathop{\mathbb{R}\kern 0.0pt}\nolimits^{3}}Q(f,f)(v)\,\log f(v)\,dv\geq 0. Using the symmetries of the collision operator, D(f) can be rewritten in the following symmetric and revealing form: D(f)=\frac{1}{4}\iint_{\mathop{\mathbb{R}\kern 0.0pt}\nolimits^{3}\times\mathop{\mathbb{R}\kern 0.0pt}\nolimits^{3}}\int_{\mathbb{S}^{2}}B\left(\left|v-v_{*}\right|,\sigma\right)\left(f^{\prime}f^{\prime}_{*}-ff_{*}\right)\log\frac{f^{\prime}f^{\prime}_{*}}{ff_{*}}\,d\sigma\,dv_{*}\,dv. This expression makes clear that D(f)\geq 0 because the sign of the factor (f^{\prime}f^{\prime}_{*}-ff_{*}) is always opposite to the sign of \log({f^{\prime}f^{\prime}_{*}}/{ff_{*}}) . The dissipation vanishes, D(f)=0 , if and only if f^{\prime}f^{\prime}_{*}=ff_{*} for almost all collisions, which is precisely the condition that f is a local Maxwellian (Boltzmann’s Lemma). Consequently, solutions are expected to converge (in entropy sense) to the unique global Maxwellian equilibrium \mu_{\rho_{0},u_{0},T_{0}} determined by the initial mass, momentum,