Hypocoercive Lyapunov Distance for Kinetic Systems

• The hypocoercive Lyapunov-type distance is an energy functional defined via modal decomposition that quantifies exponential decay in non-self-adjoint, degenerate dissipative kinetic equations.
• It synthesizes block-diagonal matrix weights to couple non-dissipative and dissipative modes, ensuring explicit decay rates and norm equivalence between the functional and the physical norm.
• The framework also provides quantitative control of waiting-time phenomena and is extendable to nonlinear and infinite-dimensional models with higher Sobolev regularity.

A hypocoercive Lyapunov-type distance is a rigorously constructed energy-like (entropy) functional adapted to the analysis of non-self-adjoint, degenerate dissipative evolution equations such as linear kinetic models, in particular those exhibiting hypocoercivity. Unlike classical coercive dynamics, where coercivity of the dissipative generator induces immediate exponential decay in natural norms, hypocoercive systems feature non-coercive generators combined with a mode-mixing mechanism (typically via skew-adjoint parts) that induces exponential relaxation to equilibrium after a short transient regime. The hypocoercive Lyapunov-type distance quantitatively encodes this structure by equipping the phase space with an explicitly constructed time-monotone distance, often derived by synthesizing mode-wise or block-diagonal matrix weights, ensuring both control of the original norm and strict decay under the flow.

1. Construction of the Hypocoercive Lyapunov-Type Functional

The construction begins with the decomposition of the phase space, usually a weighted $L^2$ or Sobolev space (e.g., $$\mathcal{H}_\gamma = H^\gamma(\mathbb{T}^d)\otimes L^2(\mathbb{R}^d;M^{-1}dv)$$ with $M(v)$ a Maxwellian), into orthogonal modes indexed by spatial Fourier variables and velocity Hermite basis. The perturbation $h(x,v,t) = f(x,v,t)-M(v)$ is expanded as

$$h(x,v,t) = \sum_{k\in\mathbb{Z}^d} h_k(v,t) e^{i (2\pi/L) k\cdot x}.$$

Each mode can be further represented in a Hermite basis for the velocity variable. The evolution reduces to a system of linear ODEs for the mode coefficients: $$\partial_t \hat{h}_k + i(2\pi/L)(k\cdot L_1)\hat{h}_k = -L_2 \hat{h}_k,$$ where $L_1$ is a skew-Hermitian streaming matrix, and $L_2$ is a diagonal collision matrix with possible degeneracies (nontrivial kernel).

To obtain a Lyapunov functional, one introduces for each nonzero mode $k$ a positive-definite matrix $P_k$, typically constructed via a block-ansatz involving the non-dissipative and dissipative components coupled through the hypocoercivity mechanism. The entropy functional is then defined as

$$\mathcal{H}_\gamma = H^\gamma(\mathbb{T}^d)\otimes L^2(\mathbb{R}^d;M^{-1}dv)$$0

This functional is equivalent to the physical norm; that is,

$$\mathcal{H}_\gamma = H^\gamma(\mathbb{T}^d)\otimes L^2(\mathbb{R}^d;M^{-1}dv)$$1

with explicit $$\mathcal{H}_\gamma = H^\gamma(\mathbb{T}^d)\otimes L^2(\mathbb{R}^d;M^{-1}dv)$$2-dependent constants, ensuring the entropy determines the distance to equilibrium in the original norm (Achleitner et al., 2017, Achleitner et al., 2015).

2. Hypocoercivity Index and Kalman-Type Condition

The hypocoercivity index $$\mathcal{H}_\gamma = H^\gamma(\mathbb{T}^d)\otimes L^2(\mathbb{R}^d;M^{-1}dv)$$3 quantifies the minimal number of steps needed for the non-dissipative part to couple all modes to the dissipative block. Formally, for a system $$\mathcal{H}_\gamma = H^\gamma(\mathbb{T}^d)\otimes L^2(\mathbb{R}^d;M^{-1}dv)$$4 with $$\mathcal{H}_\gamma = H^\gamma(\mathbb{T}^d)\otimes L^2(\mathbb{R}^d;M^{-1}dv)$$5, $$\mathcal{H}_\gamma = H^\gamma(\mathbb{T}^d)\otimes L^2(\mathbb{R}^d;M^{-1}dv)$$6, $$\mathcal{H}_\gamma = H^\gamma(\mathbb{T}^d)\otimes L^2(\mathbb{R}^d;M^{-1}dv)$$7 is the smallest integer such that

$$\mathcal{H}_\gamma = H^\gamma(\mathbb{T}^d)\otimes L^2(\mathbb{R}^d;M^{-1}dv)$$8

for some $$\mathcal{H}_\gamma = H^\gamma(\mathbb{T}^d)\otimes L^2(\mathbb{R}^d;M^{-1}dv)$$9. This is equivalent to a Kalman-type rank condition, ensuring that no $M(v)$0-invariant subspace remains uncoupled to $M(v)$1 (Roschkowski et al., 8 Dec 2025, Achleitner et al., 2017).

For the BGK model on the torus, the value of $M(v)$2 depends on dimension:

• $M(v)$3: $M(v)$4,
• $M(v)$5: higher-dimensional Hermite block; typically $M(v)$6.

This index directly influences both the structure of $M(v)$7 and the leading-order rate in short-time dissipation estimates.

3. Mode-wise Lyapunov Estimates and Exponential Decay

Given the explicit construction of $M(v)$8, a key differential inequality is established for each Fourier-Hermite mode: $M(v)$9 where $h(x,v,t) = f(x,v,t)-M(v)$0 is an explicit spectral gap obtained through, e.g., Sylvester’s criterion on principal minors or via determinant estimates on mode-restricted blocks (Achleitner et al., 2015, Achleitner et al., 2017). Summing these bounds across all $h(x,v,t) = f(x,v,t)-M(v)$1 yields

$h(x,v,t) = f(x,v,t)-M(v)$2

with $h(x,v,t) = f(x,v,t)-M(v)$3, implying

$h(x,v,t) = f(x,v,t)-M(v)$4

Through norm equivalence, this immediately establishes exponential relaxation to equilibrium in the original norm.

4. Two-Scale Decay and Waiting-Time Phenomenon

An essential feature of hypocoercive dynamics, encoded and quantified by the Lyapunov-type distance, is a two-timescale relaxation structure:

1. An initial plateau (waiting time) during which the $h(x,v,t) = f(x,v,t)-M(v)$5-distance to equilibrium remains nearly constant if the initial data are phase-space concentrated.
2. An onset of uniform exponential decay at rate $h(x,v,t) = f(x,v,t)-M(v)$6 after a waiting period $h(x,v,t) = f(x,v,t)-M(v)$7 given explicitly as

$h(x,v,t) = f(x,v,t)-M(v)$8

The full $h(x,v,t) = f(x,v,t)-M(v)$9 bound reads

$$h(x,v,t) = \sum_{k\in\mathbb{Z}^d} h_k(v,t) e^{i (2\pi/L) k\cdot x}.$$0

demonstrating the direct control of classical distances by the hypocoercive entropy functional (Achleitner et al., 2017).

5. Extensions to Unbounded Generators and General Modal Decompositions

For unbounded generators admitting modal decomposition, the theory extends via block-diagonal Lyapunov operators $$h(x,v,t) = \sum_{k\in\mathbb{Z}^d} h_k(v,t) e^{i (2\pi/L) k\cdot x}.$$1 on the Hilbert sum $$h(x,v,t) = \sum_{k\in\mathbb{Z}^d} h_k(v,t) e^{i (2\pi/L) k\cdot x}.$$2, with blocks

$$h(x,v,t) = \sum_{k\in\mathbb{Z}^d} h_k(v,t) e^{i (2\pi/L) k\cdot x}.$$3

reflecting the hypocoercivity index $$h(x,v,t) = \sum_{k\in\mathbb{Z}^d} h_k(v,t) e^{i (2\pi/L) k\cdot x}.$$4. The resulting Lyapunov norm $$h(x,v,t) = \sum_{k\in\mathbb{Z}^d} h_k(v,t) e^{i (2\pi/L) k\cdot x}.$$5 is strictly monotone under propagation and norm-equivalent to the physical norm: $$h(x,v,t) = \sum_{k\in\mathbb{Z}^d} h_k(v,t) e^{i (2\pi/L) k\cdot x}.$$6 Short- and long-time behavior is governed by the index $$h(x,v,t) = \sum_{k\in\mathbb{Z}^d} h_k(v,t) e^{i (2\pi/L) k\cdot x}.$$7:

• For $$h(x,v,t) = \sum_{k\in\mathbb{Z}^d} h_k(v,t) e^{i (2\pi/L) k\cdot x}.$$8, $$h(x,v,t) = \sum_{k\in\mathbb{Z}^d} h_k(v,t) e^{i (2\pi/L) k\cdot x}.$$9, $$\partial_t \hat{h}_k + i(2\pi/L)(k\cdot L_1)\hat{h}_k = -L_2 \hat{h}_k,$$0,
• For $$\partial_t \hat{h}_k + i(2\pi/L)(k\cdot L_1)\hat{h}_k = -L_2 \hat{h}_k,$$1, $$\partial_t \hat{h}_k + i(2\pi/L)(k\cdot L_1)\hat{h}_k = -L_2 \hat{h}_k,$$2.

This formalism generalizes to port-Hamiltonian PDEs and other infinite-dimensional systems under modal-splitting (Roschkowski et al., 8 Dec 2025).

6. Role in Nonlinear Models and Higher Sobolev Regularity

While the explicit construction and exponential decay are proved for linear and linearized BGK-type models, the methodologies, including the Lyapunov-type functional, carry over to establishing local asymptotic stability of corresponding nonlinear models by incorporating higher regularity and modified entropy functionals in Sobolev settings. The analytic framework is robust to higher dimensions and allows for extension to models with additional structure or degeneracy in the dissipation (Achleitner et al., 2017, Achleitner et al., 2015).

7. Summary Table: Key Properties of Hypocoercive Lyapunov-Type Distance

Property	Mathematical Expression	Reference
Functional form	$$\partial_t \hat{h}_k + i(2\pi/L)(k\cdot L_1)\hat{h}_k = -L_2 \hat{h}_k,$$3	(Achleitner et al., 2017)
Hypocoercivity index	$$\partial_t \hat{h}_k + i(2\pi/L)(k\cdot L_1)\hat{h}_k = -L_2 \hat{h}_k,$$4	(Roschkowski et al., 8 Dec 2025)
Norm equivalence	$$\partial_t \hat{h}_k + i(2\pi/L)(k\cdot L_1)\hat{h}_k = -L_2 \hat{h}_k,$$5	(Achleitner et al., 2015)
Differential inequality	$$\partial_t \hat{h}_k + i(2\pi/L)(k\cdot L_1)\hat{h}_k = -L_2 \hat{h}_k,$$6	(Achleitner et al., 2017)
L1 control	$$\partial_t \hat{h}_k + i(2\pi/L)(k\cdot L_1)\hat{h}_k = -L_2 \hat{h}_k,$$7	(Achleitner et al., 2017)
Short-time decay	$$\partial_t \hat{h}_k + i(2\pi/L)(k\cdot L_1)\hat{h}_k = -L_2 \hat{h}_k,$$8, $$\partial_t \hat{h}_k + i(2\pi/L)(k\cdot L_1)\hat{h}_k = -L_2 \hat{h}_k,$$9	(Roschkowski et al., 8 Dec 2025)

The hypocoercive Lyapunov-type distance thus serves as a canonical quantitative tool for proving decay to equilibrium in kinetic equations and related dissipative systems lacking immediate coercivity, both in finite- and infinite-dimensional modal frameworks (Roschkowski et al., 8 Dec 2025, Achleitner et al., 2017, Achleitner et al., 2015).