Phase space volume preserving dynamics for non-Hamiltonian systems
Swetamber Das, Jason R. Green · 2025
Read Phase space volume preserving dynamics for non-Hamiltonian systems by Swetamber Das, Jason R. Green (2025) 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-2511-04960.txt.
Introduction
Swetamber Das and Jason Green propose new rules for how small volumes move with a flow in phase space, the space of positions and momenta. In chaotic flow, tracked volumes tend to crash flat as vectors line up with the fastest growth path.
That crash is an artifact, not a fact of the flow, so the authors build an update rule from the skew part of the stability matrix. Their rule keeps phase space volume fixed in time, even for systems that lose energy. It also gives a density law much like the quantum Liouville-von Neumann form.
Catalog record
| CREATOR | Swetamber Das, Jason R. Green |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 59 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-04960
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Phase space volume preserving dynamics for non-Hamiltonian systems Swetamber Das, Jason R. Green Abstract Infinitesimal volumes stretch and contract as they coevolve with classical phase space trajectories according to a linearized dynamics. Unless these tangent space dynamics are modified, the underlying chaotic dynamics will cause the volume to vanish as tangent vectors collapse on the most expanding direction. Here, we propose an alternative linearized dynamics and rectify the generalized Liouville equation to preserve phase space volume, even for non-Hamiltonian systems. Within a classical density matrix theory, we define the time evolution operator from the anti-symmetric part of the stability matrix so that phase space volume is time-invariant. The operator generates orthogonal transformations without distorting volume elements, providing an invariant measure for dissipative dynamics and a evolution equation for the density matrix akin to the quantum mechanical Liouville-von Neumann equation. The compressibility of volume elements is determined by a non-orthogonal operator made from the symmetric part of the stability matrix. We analyze complete sets of basis vectors for the tangent space dynamics of chaotic systems, which may be dissipative, transient or driven, without re-orthogonalization of tangent vectors. The linear harmonic oscillator, the Lorenz-Fetter model, and the Hénon-Heiles system demonstrate the computation of the instantaneous Lyapunov exponent spectrum and the local Gibbs entropy flow rate using these bases and show that it is numerically convenient. I Introduction Phase space volume conservation is a defining feature of Hamiltonian dynamics Gibbs 2010 . This time invariance of phase space volume is central to the Liouville’s theorem for divergence-free phase space flows. A more rigorous statement considers the symplectic geometry of the Hamiltonian flow which leads to the incompressiblity of phase space volumes Arnol’d 1989 . It is usually understood as the preservation of an infinitesimal volume element evolving in time despite its distortions (expansion, contraction, and rotation) due to the governing flow Tolman 1979 . However, many dynamical systems of interest are characterized by phase space variability . Systems that are dissipative, driven, or constrained do not conserve phase space volume. The statistical mechanics of these non-Hamiltonian systems must then be based on generalizations of Liouville’s theorem and Liouville’s equation. The most straightforward generalization Steeb 1979 ; Andrey 1985 ; Andrey 1986 of Liouville’s equation for an arbitrary deterministic system invokes the continuity equation in the phase space \partial_{t}\rho+\bm{\nabla}\cdot(\rho\bm{f})=0 . This equation governs the time evolution of the phase space density \rho({\bm{x}},t) for an ensemble of systems which themselves evolve according to a dynamical flow of the form \dot{{\bm{x}}}=\bm{f}({\bm{x}}) . Because Liouville’s equation also holds for out of equilibrium systems, its solution i.e., the phase space probability density \rho({\bm{x}},t) is used to compute mean phase space observables such as Gibbs entropy rate. For dissipative systems, the rate at which the phase space probability density \rho(t) grows with time is determined by the divergence of the flow \bm{\nabla}\cdot\dot{\bm{x}} . Equivalently, the flow divergence also measures the compressibility of the phase space which is the contraction rate of phase space volume element \delta V(t) . For non-equilibrium steady states, \rho diverges as \delta V collapses onto a structure of fractal dimension Hoover and Posch 1995 ; Klages 2007 . As a consequence, irreversible-thermodynamics concepts such as nonequilibrium Gibbs entropy rate becomes undefined. The non-smoothness of the \rho in non-Hamiltonian systems due to phase space variability remains a long-standing concern Ramshaw 1986 ; Ramshaw 2002 ; Tuckerman, Mundy, and Martyna 1999 ; Tuckerman et al. 2001 ; Ezra 2004 . This ambiguity may be resolved either by course-graining of the phase space or by determining a suitable invariant measure Ramshaw 1986 . The presence of an appropriate invariant measure implies that the phase space average of a specific property can be correlated with the time-average of that same property along a trajectory (assuming ergodicity). To derive an appropriate measure, a typical way is to modify the phase space probability density by including either a geometric factor or a “measure function” to normalize the probability density \rho({\bm{x}},t) Tuckerman, Mundy, and Klein 1997 ; Tuckerman, Mundy, and Martyna 1999 ; Tuckerman et al. 2001 ; Ezra 2004 ; Ramshaw 2019 ; Ramshaw 2020 ; Ramshaw 2022 . Obtaining such smooth invariant measures that satisfy the Liouville equation is a mathematically nontrivial and often subtle for non-Hamiltonian systems Holian et al. 1989 ; Ruelle 1999 ; Dorfman 1999 . Despite significant theoretical progress, computing the phase space density for large systems remains inherently challenging due to dynamical chaos. When phase space trajectories diverge at an exponential rate, the phase space volume develops intricate fine structures, causing \rho to become non-smooth and rendering Liouville’s equation inapplicable. This challenge persists even in chaotic Hamiltonian systems, where the conservation of phase space volume does not guarantee a smooth probability density due to continuous stretching and folding of phase space volume. As a result, direct numerical computation to obtain well-behaved phase space density is computationally intensive because chaos causes nearby trajectories to separate exponentially. This leads to a rapid alignment of tangent vectors with the most expanding direction, making phase-space volume computation difficult Skokos and Manos 2016 . The issue is conventionally addressed via Gram-Schmidt orthogonalization Ott 2002 , but its iterative nature is computationally expensive and suffers from loss of orthogonality due to round-off errors Leon, Björck, and Gander 2013 ; Ford 2015 . This challenge is even more pronounced in non-ergodic chaotic systems, where complex phase-space structures and high-dimensional dynamics further complicate computations Guralnik 2008 ; Hahn et al. 2021 . Here, we address two questions in the context of classical chaotic dynamics — (1) how to generalize Liouville’s equation to obtain a well-defined invariant measure and (2) how local Lyapunov exponents, in various bases, contribute to the local Gibbs entropy flow rate. To do that, we use the emerging framework of classical density matrix theory Das and Green 2022 which is akin to quantum density matrix formulation in its mathematical treatment. In Sec. II , we summarize the theory and show how quantities like the instantaneous Lyapunov exponents and local phase space volume contraction rates appear. In Sec. III , we obtain a set of evolution operators for a modified tangent space dynamics. Based on the mathematical properties of these operators we derive, in Sec. IV , a set of generalized Liouville’s theorem and equations. To describe local rates of stretching and contraction of phase space volume elements, we introduce several “basis states” in Sec. V and show how to derive those local rates for Hamiltonian and non-Hamiltonian systems. We compute these rates for a several model systems including the damped harmonic oscillator, Lorenz-Fetter model, and Hénon-Heiles system in Sec. VI . \,\ket{\delta\yu_1(t_0)} \,\ket{\delta\yu_2(t_0)} \theta \,{\bm{M}}_{-}\ket{\delta\yu_2(t_0)} \,{\bm{M}}_{-}\,\ket{\delta\yu_1(t_0)} \theta \,{\bm{M}}\ket{\delta\yu_1(t_0)} \,{\bm{M}}\ket{\delta\yu_2(t_0)} Figure 1: Time evolution of the initial pure states \ket{\delta\yu_1(t_0)} and \ket{\delta\yu_2(t_0)} at time t_{0} to the states \ket{\delta\yu_1(t)} and \ket{\delta\yu_2(t)} at time t , under two norm-preserving dynamics governed by the evolution operators \widetilde{{\bm{M}}} and {\bm{M}}_{-} along a given classical trajectory. The operator {\bm{M}}_{-} is orthogonal and therefore preserves both the norm and the angle \theta between the two states over time. While \widetilde{{\bm{M}}} preserves the norm, it does not preserve the angle. II Density matrix theory We have recently introduced a classical density matrix theory for deterministic systems to