REVIEW 3 major objections 5 minor 1 cited by
Classical Fractons: Local chaos, global broken ergodicity and an arrow of time
T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash
Pith's one-line read Late-time four-fracton clusters are integrable billiards in a triangle or square, so local ergodicity coexists with global non-ergodicity.
desk verdict A novel billiard reduction for four-fracton late-time states is the real contribution, but the central integrability claim rests on an unproven confinement assumption that a referee should push on. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The machinery is the reduced-coordinate formulation together with the billiard reduction. Coordinates $q_1,q_2,q_3$ and conjugate momenta $\pi_1,\pi_2,\pi_3$ remove the conserved total momentum and dipole moment; the Hamiltonian is a sum of pair-inertia terms, each a momentum-difference squared weighted by a locality function $K(x)$ that is nonzero only when two particles are nearby. In the late-time cluster states only two or three pair terms remain active, so the trajectory is confined to a plane, and the boundaries of the triangular or square allowed region act as mirrors; the paper derives explicit reflection rules under which one reduced-momentum component reverses while another is conserved, making the dynamics exactly a polygonal billiard. Three integrals then follow: the plane containing the trajectory, the conserved kinetic energy, and the finite set of possible velocity directions for a rational polygon, which together imply Liouville integrability. For the Janus-point result, the central object is the complexity $P=\tfrac12\sum_{i\neq j}(p_i-p_j)^2$, whose growth on both sides of the central time is converted into a Boltzmann entropy through a hypersphere-shell phase-space count on a fixed cluster configuration.
What would settle it
Run long-time simulations of the four-fracton box-$K$ Hamiltonian from many big-bang-like initial conditions and monitor the reduced coordinates; finding any 3-1 or 2-2 trajectory that enters the large-triangle or outside region (a second particle at rest) or any positive Lyapunov exponent for compact $K$ would refute the integrable-billiard claim.
Extended reading notes
Core claim
The central claim is that the late-time dynamics of four classical fractons with box (compact) pair inertia $K$ reduces to integrable billiards in reduced coordinates. A 3-1 cluster, with three fractons moving together and one frozen, is confined to an equilateral triangle in the $(q_1,q_2)$ plane; a boundary collision flips one reduced-momentum component while leaving the component parallel to the wall unchanged, which is exactly specular billiard reflection. A 2-2 cluster is likewise confined to a square, with the reflected velocity components swapped in the new coordinates. In both cases the trajectory carries three independent integrals: the conserved plane, the conserved kinetic energy, and the finite set of allowed velocity directions inherited from the rationality of the polygon. Three integrals in a six-dimensional phase space imply Liouville integrability, so the cluster is locally ergodic (it fills the triangle or square) but globally non-ergodic (it cannot explore the full phase space). With non-compact $K$, some 3-1 trajectories reach the 'large triangle' where a second fracton momentarily stops, and these states show a positive Lyapunov exponent. Separately, the paper claims every generic non-stationary trajectory has a Janus point: a time of maximal homogeneity from which complexity $P=\frac{1}{2}\sum_{i\neq j}(p_i-p_j)^2$ and a Boltzmann entropy $S\simeq k_B\frac{N-4}{2}\ln P+\text{const}$ increase in both temporal directions.
Load-bearing premise
The billiard reduction of the 3-1 state rests on the empirical assertion that with box (compact) $K$, no trajectory ever enters the 'large triangle' region of reduced space where a second fracton would be momentarily at rest; if such trajectories exist for some initial conditions, the small-triangle mapping and the no-chaos conclusion would not hold for those states.
Editorial extensions
If this is right
- For box or compact pair inertia, the four-fracton 3-1 and 2-2 late-time states are exactly integrable billiards, so their long-time behavior is determined by the geometry of a triangle or square rather than by generic many-body chaos.
- Global ergodicity is broken in a controlled way: emergent conserved quantities confine the dynamics to a low-dimensional subspace even though the cluster motion fills that subspace densely.
- Changing the interaction range from compact to non-compact turns the same cluster state from regular to chaotic, so chaos is a property of the pair inertia's support rather than of clustering itself.
- Every generic non-stationary trajectory has a central time of maximal homogeneity, with complexity and the non-equilibrium Boltzmann entropy monotone in both time directions, yielding two arrows of time without invoking a past hypothesis.
- Clustered steady states are stable to small phase-space perturbations for compact $K$ because nearby trajectories do not separate exponentially, suggesting the clustering found for small $N$ persists for larger $N$.
Reading between the lines
- Editorial extension: the reflection calculations in the appendix give an algebraic route to proving that large-triangle 3-1 trajectories form a measure-zero set for compact $K$, which would turn the billiard reduction into a theorem rather than an empirical observation.
- Editorial extension: for $N>4$, the same reduction should produce billiards in rational polytopes of reduced coordinates; numerically checking whether each cluster type has zero Lyapunov exponents for compact $K$ would test whether integrability is generic for all cluster sizes.
- Editorial extension: smoothing the compact box $K$ into a function with a small tail should interpolate between the integrable and chaotic regimes; measuring the first Lyapunov exponent as a function of the tail length would make the compact/non-compact transition quantitative.
- Editorial extension: the Janus-point mechanism does not depend on fracton-specific details beyond cluster formation and linear momentum growth, so similar bidirectional arrows of time should appear in other non-ergodic Hamiltonian systems, including dipole-conserving lattice models in their fragmented phases.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. This paper studies one-dimensional classical fractons with pair-dependent inertia, continuing earlier work by the same authors. For the four-fracton problem with compact ('box') pair inertia K, it claims that late-time 3–1 and 2–2 cluster states reduce to billiard motion in an equilateral triangle and a square, respectively; because these polygonal billiards are integrable, the cluster dynamics is locally ergodic but not chaotic, while emergent conserved quantities break global ergodicity. The paper also introduces a complexity measure and a Boltzmann entropy, arguing that both increase away from a central 'Janus point', yielding a bidirectional arrow of time. The main technical content is the reduced-coordinate Hamiltonian, reflection rules at cluster boundaries, and numerical simulations of trajectories and Lyapunov exponents.
Significance. If the billiard reduction were rigorously established, this paper would provide a clean Hamiltonian many-body example in which global ergodicity is broken while cluster-level dynamics remains regular (or becomes chaotic for non-compact K). The explicit mapping of interacting fractons to integrable billiards is novel and would be of interest to the statistical-mechanics and fracton communities. The Janus-point/entropy construction also connects fracton dynamics to cosmological arrow-of-time ideas. Strengths include the transparent reduced-coordinate formulation, the concrete reflection rules, the numerical verification of trajectories, and the clear separation of compact and non-compact cases. However, the central reduction relies on an unproven confinement assumption, and the arrow-of-time argument depends on an assumed linear growth law rather than a derivation.
major comments (3)
- [III.B and III.C] The billiard reduction for the 3–1 state is supported only by the statement 'This is never observed in simulations of 3–1 states' (Section III.B), and the reflection rule is derived after the explicit clause 'Assuming the trajectory does not escape the triangle.' The manuscript never checks whether the boundary jump conditions admit a real transmission branch into the region where the previously isolated fourth fracton becomes coupled. This matters because the analogous three-fracton boundary jumps in Appendix A are resolved by real-solution discriminants (e.g., D1 and D2 in Appendix A), and both reflecting and transmitting branches occur there. The same gap appears in the 2–2 case: the square-wall reflection rule in Section III.C is stated without a discriminant analysis. Since the claimed three integrals of motion and the resulting Liouville integrability in Section IV.A apply only if every trajectory is reflected at the small-triangle or square walls, the confinement and the reflection branch must be proven, not assumed.
- [IV.A] The statement 'We now prove this' is not backed by a rigorous proof. The 'three integrals of motion' are not of the same type: the first two (the plane containing the trajectory and the Hamiltonian) are smooth conserved functions, while the third (the finite number of billiard directions) is a discrete invariant coming from the polygon being rational. The cited Liouville integrability theorem for a 2N-dimensional phase space with N integrals does not apply directly to a discrete, non-smooth invariant. The conclusion that equilateral-triangle and square billiards are regular is correct by standard results, but the proof as written is incomplete. Additionally, the phrase 'locally ergodic' is never defined; for these integrable billiards a trajectory is confined to a lower-dimensional invariant set (a fixed direction class and, on unfolding, a torus), so if 'local ergodicity' means exploration of the full cluster phase space, that is false. The authors should state which notion of local ergodicity they intend and prove it for the energy surface restricted to each invariant sector.
- [V, Eqs. (22)-(31)] The arrow-of-time result rests on the a priori assumption pi = alpha_i + beta_i t (Eq. 22), which is stated as a toy-model input rather than derived from the equations of motion. The complexity P is then shown to grow for t > t_+ (Eq. 24), and the Boltzmann entropy S is defined as a monotone function of P, so the 'second law' in Eq. (31) is a consequence of the choice of macroscopic variable. To support the abstract's claim of a 'generic' bidirectional arrow of time, the paper should either derive Eq. (22) from the dynamics or clearly state that this is a phenomenological model, and it should quantify how generic the observed growth of P is. As it stands, the entropic arrow is tautological once P is observed to increase.
minor comments (5)
- [Abstract, II, IV] The abstract and Section IV refer to 'various stadiums', but the paper treats equilateral-triangle and square billiards, not stadium billiards. Please use 'polygonal billiards' or specify the domains.
- [IV.B] The sentence 'this guarantees only 4 of the 6 exponents to be zero' is imprecise: since Lyapunov exponents in a Hamiltonian system come in pairs (lambda, -lambda), three zero exponents force at least four zeros out of six, leaving room for one nonzero pair. The wording should be corrected.
- [Appendix A] The arguments based on 'arbitrarily large region' Mathematica RegionPlots are informal; the analytic inequalities in Eqs. (A24)–(A31) are more convincing. Please replace the plot-based claims with the analytic discriminant conditions throughout Appendix A.
- [General] There are recurring typos and notation issues, e.g., 'π′2 = −π2' is written with an unusual prime placement, and the reflection rules in Sections III.B and III.C would be easier to follow if the conserved linear combinations were displayed before the jump rules.
- [References] Reference [15] appears incomplete: 'E. Ledberg, Introduction to rational billiards (2018)' lacks a journal or arXiv identifier. Please verify and complete the citation.
Circularity Check
The entropic arrow of time is a definitional relabeling of the observed complexity increase; the billiard reduction rests on an unproven confinement assumption, but the central integrability claim is otherwise self-contained.
-
self definitional
[Section V.B, Eq. (25) and Eq. (31), and the following paragraph.]
"S = kB ln(d/dP Vol(Γ[P,P+dP])). ... S ≃ kB ( N − 4 2 ) ln(P ) + const, where the constant does not depend on P. Importantly, S is an increasing function of P. Complexity P, and hence the entropy S, is observed to be non-decreasing for generic trajectories (e.g. Fig. 7), in accordance with the second law."
The Boltzmann entropy is defined as the log volume of the macro-set of states with a fixed value of the complexity P, and the subsequent calculation shows S(P) is monotonically increasing in P by construction. The paper then states that P, and hence S, is observed to be non-decreasing, and uses this to define a thermodynamic arrow of time. Thus the entropic arrow contains no dynamical content beyond the already-observed monotonicity of P: the entropy increase is a relabeling of the complexity increase, not an independent derivation. This is an acknowledged construction and a consistency check rather than a fitted prediction, and it does not affect the billiard/integrability results.
full rationale
The main derivation—the reduction of late-time 3–1 and 2–2 four-fracton states to billiards in an equilateral triangle and a square—is self-contained in the sense that the reflection rules are derived from the full four-fracton Hamiltonian (Eq. 9) together with conservation laws, and the integrability conclusion follows from standard polygonal-billiard results cited from the external literature. The paper's reliance on the authors' prior work [1,2] for the reduced-coordinate formalism and the three-fracton classification is a normal continuation of a research program, not a load-bearing self-citation chain. The one genuine circularity is in the entropy section: the entropy is defined in terms of the complexity P, so the statement that entropy increases with time reduces by construction to the observed increase of P. The paper is explicit that this is a construction, so the severity is limited. Separately, the billiard reduction in Section III.B assumes, rather than proves, that trajectories do not escape the small triangle; the text says 'Assuming the trajectory does not escape the triangle' and supports confinement with 'This is never observed in simulations.' This is an unverified empirical assumption that is load-bearing for the integrability claim, but it is a gap in proof or a correctness risk, not circularity, because the claim does not reduce to its own input by definition. Overall, the central local-ergodicity/global-ergodicity-breaking result is independent, and the circularity is confined to the arrow-of-time entropy argument.
Assumptions & free parameters
assumptions (5)
- domain assumption Hamiltonian Eq. (3) with pair inertia function K(x) describes dipole-conserving fractons with locality imposed by K.
- domain assumption Box pair inertia function K(x)=1 for |x|<1 and 0 otherwise yields instantaneous momentum jumps at region boundaries.
- ad hoc to paper Late-time momenta grow linearly in time, p_i = alpha_i + beta_i t, with fixed clustering configuration (Eq. 22).
- standard math Polygonal billiards in rational polygons have zero Lyapunov exponents and are Liouville integrable with the listed integrals of motion.
- standard math Hamiltonian systems have Lyapunov exponents in pairs (lambda, -lambda).
Cite this review
Pith. "Pith review of Classical Fractons: Local chaos, global broken ergodicity and an arrow of time." pith.science (2026). https://pith.science/paper/H4I4E2SU
@misc{pith2026250112445,
author = {Pith},
title = {Pith review of: Classical Fractons: Local chaos, global broken ergodicity and an arrow of time},
year = {2026},
howpublished = {\url{https://pith.science/paper/H4I4E2SU}},
note = {Machine review of arXiv:2501.12445}
}
read the original abstract
We report new results on classical nonrelativistic dipole conserving particles - fractons. These have been previously shown to exhibit "Machian" dynamics where the motion of one particle requires the presence of others in its proximity, such that dynamics produces ergodicity breaking steady states characterized by clusters. In this work, we show that although the global state breaks ergodicity, a limited version of ergodic behavior is retained within the clusters which may or may not be chaotic, depending on the nature of the microscopic Hamiltonian. In certain cases, we show that the dynamics can be mapped to that of a billiards particle in various stadiums. We also show that the many-fracton trajectories characteristically exhibit a central time or "Janus point" and thus a generic nonequilibrium bidirectional arrow of time.
Figures
Figures from the paper (4 more)
Forward citations
Cited by 1 Pith paper
-
Classical fractons with cosmological fixed points
In a special scale-invariant fracton Hamiltonian, late-time attractors reproduce the scale expansion, homogeneity, critical Newtonian dynamics, and arrow of time of a flat matter-dominated cosmology.
Reference graph
Works this paper leans on
-
[1]
First, thesefractonsappear dissipative [1], inanap- parent contradiction of Liouville’s theorem, which forbids attractors in Hamiltonian systems. This is clearly seen for systems of two fractons, which generically separate out to a fixed distance, and motion comes to a halt, reminiscent of a system with friction. Of course, as a rigorous theorem, Li- ouvi...
-
[2]
Conventional Hamiltonian systems have a linear re- lationship between position and velocity for each particle. However, velocities of individual fractons involve the pair inertia function, and momenta of all other fractons. As the apparent attractor is ap- proached, shrinking in position space, momenta of particles diverge, conserving true phase space vol...
-
[3]
Ergodicityisfoundtobebrokeninanunusualman- ner [2]. Late time states always converge to attrac- tors, with the emergence of new conserved quanti- ties. Most interestingly, late time states generically lead to the breaking of translation symmetry into crystalline states, even in low dimensions, where the naive invocation of the Hohenberg-Mermin- Wagner-Col...
-
[4]
1a) A trajectory starting in Region3, moving towards Region 2a, will always pass into2a
(Fig. 1a) A trajectory starting in Region3, moving towards Region 2a, will always pass into2a. Then it will jump from2a to the first sub-region of1 it encounters, and remain trapped there
-
[5]
(Fig.1b, c)AtrajectorystartinginRegion 2a, mov- ing towards Region1b, either remains trapped in1b (depending on the initial momenta), or it escapes to 2f, from where it necessarily goes into1a, and is trapped there
-
[6]
A trajectory starting in Region 1a either gets trapped in 1a or will reflect off the1a-2e and 1a-2f boundaries a finite number of times, before escap- ing to Region 2e or 2f. The number of such re- flections depends on the ratio of reduced momenta 4 π2/π1 at the start of the trajectory. Once the tra- jectory has escaped to2e or 2f, it ends up trapped in o...
-
[7]
Box K: K(x) = 1 for |x| < 1, and 0 for |x| ≥1
-
[8]
Compact K: K(x) = 0 for |x| ≥1, but generally may take any value for|x| < 1
Show all 38 references
-
[9]
big-bang
Non-compact K:K(x) → 0 only at x → ±∞. We first focus on the boxK, where we rigorously un- derstand dynamics by analyzing trajectories in reduced coordinates. As we shall show, most results from the box K case carry over to the general compactK case. We then analyze non-compac...
-
[10]
Generalizing the approach in [1], we derive from the full Hamiltonian Eq
= const . Generalizing the approach in [1], we derive from the full Hamiltonian Eq. (9) at the boundary: π1 = const and√ 2π2 + π3 = const. Writing the newπ′ i in terms of the old πi, we have:π′ 1 = π1, π′ 2 = −π2 and π′ 3 = 2 √ 2π2+π3. The component of the velocity parallel to...
-
[11]
The full Hamiltonian Eq
Similarly to the 3–1 case, the Hamiltonian H2−2 = π2 1 + √ 2π3 − π2 2 /3 = const. The full Hamiltonian Eq. (9) yields: π3 = const and √ 3π1 − π2 = const. Solving for the new coordinates π′ i in terms of the old onesπi: π′ 1 = √ 2π3 − π2√ 3 , π′ 2 = √ 2π3 − √ 3π1, π′ 3 = π3. (1...
-
[12]
The plane containing the trajectory:q3 = const for 3–1 states and √ 2q2 + q3 = const for 2–2 states
-
[13]
The magnitude of velocity:H3−1 = 3/2(π2 1 + π2
-
[14]
H2−2 = 1 4 ( ˙x2 + ˙y2) = const for 2–2 states
= const for 3–1 states. H2−2 = 1 4 ( ˙x2 + ˙y2) = const for 2–2 states
-
[15]
galaxies
The finite number of directions of the billiard. In 8 (a) (b) FIG. 5: (a): Time evolution of particle positions in a 2–2 trajectory discussed in Section IIIC (b): the trajectory in reduced position coordinates. an extended zone scheme, with the squares or tri- angles being til...
2020
-
[16]
Prakash, A
A. Prakash, A. Goriely, and S. L. Sondhi, Classi- cal nonrelativistic fractons, Physical Review B 109, 10.1103/physrevb.109.054313 (2024)
2024 doi
-
[17]
Prakash, Y
A. Prakash, Y. Sadki, and S. L. Sondhi, Machian fractons, hamiltonian attractors, and nonequilibrium steady states, Physical Review B 110, 10.1103/phys- revb.110.024305 (2024)
2024 doi
-
[18]
Gromov and L
A. Gromov and L. Radzihovsky, Colloquium: Fracton matter, Rev. Mod. Phys.96, 011001 (2024)
2024
-
[19]
R. M. Nandkishore and M. Hermele, Fractons, Annual Review of Condensed Matter Physics 10, 295 (2019), https://doi.org/10.1146/annurev-conmatphys-031218- 013604
2019 doi
-
[20]
Pretko, X
M. Pretko, X. Chen, and Y. You, Fracton phases of matter, International Journal of Modern Physics A35, 2030003 (2020)
2020
-
[21]
P. C. Hohenberg, Existence of long-range order in one and two dimensions, Phys. Rev.158, 383 (1967)
1967
-
[22]
N. D. Mermin and H. Wagner, Absence of ferromag- netism or antiferromagnetism in one- or two-dimensional isotropic heisenberg models, Phys. Rev. Lett.17, 1133 (1966)
1966
-
[23]
Coleman, There are no Goldstone bosons in two di- mensions, Communications in Mathematical Physics31, 259 (1973)
S. Coleman, There are no Goldstone bosons in two di- mensions, Communications in Mathematical Physics31, 259 (1973)
1973
-
[24]
S. M. Carroll and J. Chen, Spontaneous inflation and the origin of the arrow of time, arXiv preprint hep- th/0410270 (2004)
2004
-
[25]
Barbour, T
J. Barbour, T. Koslowski, and F. Mercati, Identification of a gravitational arrow of time, Physical Review Letters 113, 10.1103/physrevlett.113.181101 (2014)
2014 doi
-
[26]
Goldstein, R
S. Goldstein, R. Tumulka, and N. Zanghì, Is the hypoth- esis about a low entropy initial state of the universe nec- essary for explaining the arrow of time?, Physical Review D 94, 10.1103/physrevd.94.023520 (2016)
2016 doi
-
[27]
Lazarovici and P
D. Lazarovici and P. Reichert, Arrow (s) of time with- out a past hypothesis, inStatistical mechanics and scien- tific explanation: Determinism, indeterminism and laws of nature(World Scientific, 2020) pp. 343–386
2020
-
[28]
Pretko, Emergent gravity of fractons: Mach’s princi- ple revisited, Phys
M. Pretko, Emergent gravity of fractons: Mach’s princi- ple revisited, Phys. Rev. D96, 024051 (2017)
2017
-
[29]
S. V. Naydenov, D. M. Naplekov, and V. V. Yanovsky, New mechanism of chaos in triangular billiards, JETP Letters 98, 496 (2013)
2013
-
[30]
Ledberg, Introduction to rational billiards (2018)
E. Ledberg, Introduction to rational billiards (2018)
2018
-
[31]
Arutyunov, Liouville integrability, in Elements of Classical and Quantum Integrable Systems(Springer In- ternational Publishing, Cham, 2019) pp
G. Arutyunov, Liouville integrability, in Elements of Classical and Quantum Integrable Systems(Springer In- ternational Publishing, Cham, 2019) pp. 1–68
2019
-
[32]
M. D. Hartl, Lyapunov exponents in constrained and unconstrained ordinary differential equations (2003), arXiv:arXiv:physics/0303077 [physics.comp-ph]
2003 arXiv
-
[33]
Skokos, The lyapunov characteristic exponents and their computation, in Dynamics of Small Solar System Bodies and Exoplanets(Springer, 2010) pp
C. Skokos, The lyapunov characteristic exponents and their computation, in Dynamics of Small Solar System Bodies and Exoplanets(Springer, 2010) pp. 63–135
2010
-
[34]
Goldstein, J
S. Goldstein, J. L. Lebowitz, R. Tumulka, and N. Zanghì, Gibbs and boltzmann entropy in classical and quantum mechanics, inStatistical Mechanics and Scientific Expla- nation (World Scientific, 2020) Chap. 14, pp. 519–581
2020
-
[35]
G. N. Remmen and S. M. Carroll, Attractor solutions in scalar-field cosmology, Phys. Rev. D88, 083518 (2013)
2013
-
[36]
Classen-Howes, R
J. Classen-Howes, R. Senese, and A. Prakash, Universal freezing transitions of dipole-conserving chains (2025), arXiv:2408.10321 [cond-mat.str-el]
2025
-
[37]
A.Morningstar, V.Khemani,andD.A.Huse,Kinetically constrained freezing transition in a dipole-conserving sys- tem, Phys. Rev. B101, 214205 (2020)
2020
-
[38]
Pozderac, S
C. Pozderac, S. Speck, X. Feng, D. A. Huse, and B. Skin- ner, Exact solution for the filling-induced thermalization transition in a one-dimensional fracton system, Phys. Rev. B107, 045137 (2023)
2023
Reviewed August 10, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.