Pith. sign in

REVIEW 3 major objections 5 minor 56 references

Universal Scaling Laws in Freeway Traffic

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Real freeway traffic is a self-organized critical system whose jam clusters scale like percolation clusters and whose vehicle-count interface roughens with the Kardar-Parisi-Zhang exponents of growing interfaces.

desk verdict First trajectory-level test of SOC and KPZ conjectures in freeway traffic, but the SOC claim rests on a threshold plateau that the authors' own finite-size scaling contradicts. read the letter →

arxiv 2507.09530 v1 pith:LKW3YVGR submitted 2025-07-13 nlin.CD physics.data-an

Garyoung Lee , Aryaman Jha , Kurt Wiesenfeld , Jorge Laval This is my paper · ORCID
classification nlin.CDphysics.data-an
keywords trafficflowself-organizedcriticalitypercolationtransitionKardar-Parisi-Zhanguniversalityjamclusterscriticalexponentsvehicletrajectorydatainterfaceroughening
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Traffic congestion, which resists traditional physical modeling, may obey sharp universal statistical laws. Using vehicle-by-vehicle trajectories from an instrumented 4.2-mile stretch of Interstate 24 sampled at 25 Hz, this paper argues that traffic jams form a percolation phase transition that is self-organized critical, and that the fluctuations of the vehicle-count interface belong to the Kardar-Parisi-Zhang (KPZ) universality class, the family of growing random interfaces with identical fluctuation statistics. The evidence is a set of measured critical exponents — cluster-size exponent $\tau \approx 1.5$, fractal dimension $D_f \approx 1$, dynamic exponent $z_P \approx 1$, delay dimension $D \approx 1.5$ — that stay roughly constant across a wide speed-threshold plateau ($10 \lesssim v_c \lesssim 35$ mph), together with interface-width growth and data collapses matching $\alpha = 1/2$, $\beta = 1/3$, $z = 3/2$. If the claim holds, seemingly chaotic stop-and-go traffic has predictable statistical structure, which would ground congestion forecasting and control in measurable scaling exponents rather than static thresholds.

What carries the argument

The argument is carried by two objects. The first is the jam cluster: a connected component of congested cells in a time-space grid (0.02 mile by 6 second cells) obtained by thresholding measured speeds at $v_c$; its size $S$, spatial extent $R$, and temporal duration $T$ supply every percolation exponent — $\tau$, $\alpha_R$, $\alpha_T$, $D_f$, $D_R$, $D_T$, $z_P$ (computed as the ratio $D_R/D_T$ rather than fitted directly), and the delay dimension $D$ — via power-law fits and the hyper-scaling relations $\alpha_R = D_R(\tau-1)+1$, $\alpha_T = D_T(\tau-1)+1$, and $D = d + D_f(2-\tau)$. The second is the detrended cumulative vehicle-count interface $h'(x,t) = h(x,t) - \rho x$, built by assigning +1 to spatial segments of about one vehicle length that contain a car and summing along the road at fixed time; its width $W(t)$, local width $W(\ell,t)$, height-height correlation $C(r,t)$, and Hurst exponent provide the KPZ exponents. The mechanism connecting them is the Fortuin–Kasteleyn-style identification of criticality with cluster geometry: if jam clusters are scale-free and their exponents obey the hyper-scaling identities, traffic sits at a percolation critical point, and the plateau of stable exponents across thresholds is the signature of self-organized criticality.

What would settle it

Recompute the cluster statistics on the same trajectories but with a different time-space discretization (for example 3-second or 12-second time bins, with the box aspect ratio adjusted to the shockwave speed): if the exponent plateau at $\tau \approx 1.5$, $D_f \approx 1$, $z_P \approx 1$ does not survive the resolution change, the “critical plateau” is a thresholding artifact of the chosen grid. A direct confirmation would come from the KPZ side: collect the full distribution of the detrended interface height at fixed time and test whether it matches the universal KPZ fixed-point distribution, since matching exponents with non-matching distributions would show the system is rough but not in the KPZ class.

Watch

Extended reading notes

Core claim

The paper's central claim is that real freeway traffic, examined at the microscopic trajectory level, is a genuinely critical system. Binarizing a time-space speed field at a speed threshold $v_c$ and grouping adjacent slow cells into jam clusters — the traffic shockwaves that propagate upstream — the authors find power-law cluster-size, spatial-extent, and temporal-duration distributions whose exponents are stable across a broad plateau, $10 \lesssim v_c \lesssim 35$ mph: $\tau \approx 1.5$, $D_f \approx 1$, $z_P = D_R/D_T \approx 1$, with the delay fractal dimension $D \approx 1.5$–$1.75$. Because critical behavior persists over a range of thresholds rather than at a single tuned point, the authors conclude the system is self-organized critical, with two crossover transitions bounding the regime: a shockwave critical point near 10 mph and a lateral-merging critical point near 30 mph. On the fluctuations side, the detrended cumulative vehicle-count curve $h'(x,t)$ roughens in time with growth exponent $\beta \approx 1/3$, its spatial correlations collapse under dynamic scaling with $z \approx 3/2$, and its Hurst exponent sits at $\alpha \approx 1/2$ — the exact KPZ values in 1+1 dimensions. The measured percolation exponents coincide with exact results for the deterministic limit of TASEP, which the authors read as evidence that real traffic behaves nearly deterministically yet keeps its KPZ membership, and they argue the observed exponents distinguish traffic from the directed-percolation universality class. The width-growth analysis is restricted to the two days with complete early-morning trajectory coverage, since gaps in the other days produce artificial bursts in the interface width.

Load-bearing premise

The whole argument stands on the assumption that slicing vehicle trajectories into slow and fast cells with a speed cut-off gives a true picture of jams as percolation clusters, and that the power laws seen across the claimed 10–35 mph range are real and not artifacts of the cut-off and curve-fitting — an assumption the paper's own finite-size scaling analysis in the Supplementary Text only clearly supports near 10–15 mph, where it concedes the results “do not exhibit widespread robustness” across the broader interval.

Editorial extensions

If this is right

  • Because the exponents are stable across a wide threshold plateau, the precise operational definition of a congested cluster matters little, making the scaling signatures stable enough to serve as dynamic indicators of impending congestion.
  • A real-time system that monitors fluctuation statistics such as speed or headway variance could detect early signs of the phase transition and steer traffic away from the critical state, rather than applying static speed or density thresholds.
  • The match between the measured exponents and exact deterministic-TASEP results ($\tau = 1.5$, $z_P = 1$) identifies the near-deterministic limit $p \to 1$ of stochastic traffic models as the theoretically relevant regime, and it sharpens the open question of whether TASEP itself is critical.
  • The observed exponents are distinct from the directed-percolation universality class ($z_P \approx 1.58$, $\tau \approx 1.277$), so despite the anisotropic, shockwave-like growth of jam clusters, traffic belongs to a different universality class.
  • Excluding clusters smaller than ten cells markedly improves the hyper-scaling consistency, indicating that small fragmented clusters — frequent in outer lanes near on- and off-ramps — are the main source of lane-to-lane variation in the exponents.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A sharper test the paper does not run: at fixed time, the full one-point distribution of the detrended interface height should approach the universal KPZ fixed-point distribution (Tracy–Widom-like), not merely reproduce the exponents; matching exponents with a non-matching distribution would show the system is rough but not in the KPZ class.
  • If the plateau is genuine self-organized criticality, then on longer instrumented corridors the cutoff cluster size should grow with system length as $L^{D_f}$ while $\tau$ stays constant; a multi-site replication across segments of different lengths would either confirm or break the claim.
  • The two crossover thresholds suggest lane-changing intensity is an effective control parameter: varying ramp density or lane-change rates should shift the lateral-merging critical point and widen or narrow the plateau, a prediction that could be tested in simulation before instrumenting new sites.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper claims the first empirical demonstration, using the I-24 MOTION microscopic trajectory dataset, that freeway traffic exhibits both self-organized critical percolation and fluctuations consistent with the Kardar-Parisi-Zhang (KPZ) universality class in 1+1 dimensions. The percolation half of the claim is based on binarizing a time-space speed field with a threshold v_c, extracting jam clusters, and estimating the exponents tau, alpha_R, alpha_T, D_f, D, and z_P; the authors report a broad plateau of critical behavior for 10 ≲ v_c ≲ 35 mph and interpret this plateau as evidence for self-organized criticality. The KPZ half is based on detrended cumulative vehicle-count interfaces, from which the authors measure the growth exponent beta from the time evolution of the interface width W(t), the roughness exponent alpha from Hurst-type analyses, and test dynamic scaling collapses using the KPZ values beta = 1/3 and z = 3/2. The paper concludes that real traffic belongs to a distinct universality class from directed percolation, with percolation exponents matching a deterministic TASEP limit and interface fluctuations matching KPZ.

Significance. If the claims are substantiated, this is a potentially important empirical bridge between real-world traffic dynamics and statistical-physics universality classes, with practical implications for congestion forecasting and control. The paper has notable strengths: the analysis uses a rich, publicly available trajectory dataset; the authors provide code and processed data; the supplementary material contains extensive sensitivity analyses for box-counting, Hurst estimation, and finite-size scaling; and the discussion is candid about data artifacts such as missing coverage and tracking fragmentation. However, the current manuscript contains internal inconsistencies that directly affect the central claims, most importantly the contradiction between the broad 10-35 mph plateau asserted in the main text and the authors' own finite-size scaling analysis, which supports only a 10-15 mph critical range. These issues must be resolved before the paper can be accepted.

major comments (3)
  1. [Criticality of traffic flow; Supplementary Text, Finite-Size Scaling] The main text (Figure 3 and the surrounding text) claims a robust critical plateau for tau, alpha_R, and alpha_T over 10 ≲ v_c ≲ 35 mph and uses this broad plateau as the primary evidence for self-organized criticality. The supplementary FSS analysis, however, explicitly states that 'Unlike the broader 10-30 mph range suggested in the main text, the FSS results do not exhibit widespread robustness across this interval' and that the pronounced local minimum of the collapse quality C defined in Eq. S5 lies near 10-15 mph (Figure S3). This is a direct internal contradiction. Because the SOC interpretation depends on a wide threshold range rather than a single tuned point, the averages reported over 10-35 mph mix potentially critical and non-critical thresholds, so the reported exponents do not currently support the SOC claim as stated. Please reconcile the main-text plateau with the FSS result, restrict the plateau to the range actually supported, or provide independent evidence that the flat regions in Figure 3 reflect genuine scale invariance rather than a slow variation of piecewise-fit exponents with threshold.
  2. [KPZ universality signatures in traffic flow; Figure 4; Supplementary Text, Dataset] The growth exponent beta = 1/3 is supported by W(t) data for only two days, Nov. 22 and Nov. 23, because all other days have incomplete trajectory coverage during the early free-flow period. The supplementary materials, however, identify Nov. 23 as an anomalous day with significantly lower demand and a minor incident, and state that the scaling analysis excludes two anomalous days (the narrative identifies Nov. 21 and Nov. 23). Using Nov. 23 as one of only two days for the central KPZ growth-exponent estimate while also excluding that day as non-representative is internally inconsistent. The KPZ growth claim therefore currently rests on a single fully representative day. Please either justify the inclusion of Nov. 23 for this particular observable, supplement the analysis with additional days, or explicitly weaken the beta claim.
  3. [KPZ universality signatures; Figure 5; Eq. 9] The dynamic scaling collapse in Figure 5 is constructed using the KPZ values beta = 1/3 and z = 3/2 as inputs, and the slopes of the collapsed curves are then used to infer alpha. This is a legitimate consistency check, but it is not an independent measurement of the dynamic exponent z, and the paper does not provide a direct estimate of z from the traffic data. Since the central claim includes the full KPZ exponent set (alpha, beta, z), please either provide an independent estimate of z from the data (for example, from the time dependence of the crossover position in the correlation function or local width) or explicitly state that z is assumed rather than measured.
minor comments (5)
  1. [Supplementary Text, Finite-Size Scaling] The supplementary text refers to 'the broader 10-30 mph range suggested in the main text,' but the main text states 10 ≲ v_c ≲ 35 mph; this range should be quoted consistently.
  2. [Supplementary Text, Dataset] The supplement says the scaling analysis excludes two anomalous days but does not explicitly name both days in the exclusion sentence; the surrounding narrative implies Nov. 21 and Nov. 23, but this should be stated explicitly to avoid ambiguity.
  3. [Materials and Methods, Box-counting Method] Equation S3 gives the wave-speed range as 10-14 mph in one place and 10-15 mph in the following sentence; please harmonize these values.
  4. [KPZ universality signatures; Figure 5 caption] The caption for Figure 5 states that averaging is performed over 18 groups of 10 time slices, but the text in the main body does not explain why this grouping was chosen; a brief justification in the Methods would improve reproducibility.
  5. [Supplementary Text, Growth Exponent of Eastbound Data] The discussion of the eastbound data correctly notes that missing coverage obscures the early-time growth exponent, but the main text should state clearly that the westbound beta estimate is also subject to the same missing-data limitation for all but two days, so readers are not misled about the robustness of beta.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the empirical exponents are measured from I-24 MOTION data, the KPZ collapse is a consistency check, and the FSS/plateau tension is a robustness caveat rather than a circular derivation.

full rationale

We find no circular step. The percolation exponents (tau, alpha_R, alpha_T, D_f, D) are estimated directly from thresholded cluster statistics in the I-24 MOTION data, and the KPZ exponents are measured from interface widths, dynamic correlations, and Hurst/R-S/MLE estimates. The theoretical values alpha=1/2, beta=1/3, z=3/2 are used as comparison targets and as inputs to the Figure 5 data collapse, but the collapse is a consistency check, not a fitted prediction: alpha is independently obtained from the Hurst exponent and from the small-u slope, and the collapse quality is not optimized over beta and z. The paper explicitly states that all critical exponents except z_P were 'independently extracted from the empirical data by fitting power-laws,' and z_P is then computed as the ratio D_R/D_T, which is a definition rather than a circular fit. The self-citations (refs 5, 12, 24, all by Laval/Jha/Wiesenfeld/Lee) supply interpretive conjectures and a model comparison, but the empirical derivation does not reduce to them; no uniqueness theorem or ansatz is imported by citation to force the conclusion. The supplement's FSS statement, 'Unlike the broader 10–30 mph range suggested in the main text, the FSS results do not exhibit widespread robustness across this interval,' is an internal consistency limitation that weakens the broad-plateau/SOC claim, but it is not circularity: it is a failed or partial cross-check over part of the claimed critical range, not a parameter renamed as a prediction. Therefore the paper's central claims retain independent empirical content and the circularity score is 0.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The paper introduces no new physical entities. The free parameters are methodological choices (threshold, cluster cutoff, box geometry, R/S window, detrending density) that affect the estimated exponents and are explored in sensitivity analyses. The axioms are predominantly standard percolation and KPZ theory plus domain assumptions about the traffic system; the KPZ-height interpretation and the deterministic-TASEP comparison rely on self-cited foundations from the same group.

free parameters (5)
  • Critical speed threshold v_c = scanned 5-40 mph; plateau claimed 10-35 mph
    Defines congested cells in the binarized speed field; exponents are estimated as functions of v_c, and the plateau is the empirical basis for the SOC claim.
  • Minimum cluster size = >=2 or >=10 cells
    Used in exponent estimation; results differ depending on this cut, with outer lanes sensitive to small clusters, affecting the reported tau values.
  • Box-counting aspect ratio W/H = 1 (with Delta t = 6 s, Delta x = 0.02 mi)
    Chosen so square boxes align with the assumed shockwave speed of 10-14 mph; sensitivity analysis shows D_f depends strongly on this choice.
  • R/S window parameter r = L/3 (Gaussian filter sigma = r/2)
    Used to detrend cumulative profiles before Hurst estimation; varying r changes the R/S estimates, though the H = 1/2 conclusion is robust.
  • Density rho for detrending = empirical per lane and day
    Subtracted from h(x,t) to form the detrended height h' = h - rho x; this is an empirical fit to the spatial trend and shapes the fluctuation signal.
assumptions (6)
  • domain assumption The cumulative vehicle count h(x,t) is the KPZ height, and detrending by rho x yields KPZ fluctuations.
    Lifted from ref 5 (Laval 2024, self-cited); without this identification the KPZ interpretation of the cumulative count curves has no theoretical basis.
  • domain assumption Speed-threshold binarization substitutes for percolation occupation probability, and connected low-speed cells are the relevant clusters.
    Used throughout the percolation analysis; the paper states this departs from classical site percolation where each site has an occupation probability.
  • domain assumption The 4.2 mi segment is effectively a 1D system of size L approximately 1000 vehicle lengths with self-averaging dynamics.
    Used for the KPZ finite-size scaling and the comparison to TASEP simulations with L=1000; assumes the segment behaves as a single representative system.
  • standard math The finite-size scaling form P(S>s) ~ s^-(tau-1) f(s/L^D_f) and the hyperscaling relations of percolation theory apply to this empirical system.
    Standard percolation theory results invoked to interpret the empirical cluster statistics and to test consistency via Eqs. 3, 4, and 7.
  • domain assumption Morning peak traffic on I-24 is quasi-stationary and repeatable enough that six days and four lanes are representative.
    The analysis excludes two anomalous days and aggregates over the remaining days; this assumes the scaling behavior is stable across the sample period and lanes.
  • domain assumption The deterministic TASEP exponents from ref 12 are the correct benchmark for the observed percolation exponents.
    Comparison benchmark from the authors' companion paper; the 'exact match' claim depends on the correctness of that self-cited result, though the empirical fits are independent.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Universal Scaling Laws in Freeway Traffic." pith.science (2026). https://pith.science/paper/LKW3YVGR

@misc{pith2026250709530,
  author       = {Pith},
  title        = {Pith review of: Universal Scaling Laws in Freeway Traffic},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LKW3YVGR}},
  note         = {Machine review of arXiv:2507.09530}
}
read the original abstract

Traffic congestion, a daily frustration for millions and a multi-billion dollar drain on economies, has long resisted deep physical understanding. While simple theoretical models of traffic flow have suggested connections to critical phenomena and non-equilibrium universality, direct empirical validation is lacking. Using extensive, high-resolution vehicle trajectory data from the I-24 MOTION testbed, we show that traffic flow exhibits both a percolation phase transition that is self-organized critical and fluctuations consistent with the Kardar-Parisi-Zhang universality in 1+1 dimensions. This suggests that the complex and seemingly chaotic formation of traffic jams has predictable statistical properties, which opens new avenues in traffic science for developing advanced forecasting and management strategies grounded in universal scaling laws.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

56 extracted references · 52 canonical work pages

  1. [1]

    Herman, I

    R. Herman, I. Prigogine, A two-fluid approach to town traffic. Science 204 (4389), 148–151 (1979)

    work page 1979

  2. [2]

    Nagel, M

    K. Nagel, M. Paczuski, Emergent traffic jams. Physical Review E 51 (4), 2909 (1995)

    work page 1995

  3. [3]

    Paczuski, K

    M. Paczuski, K. Nagel, Self-organized criticality and 1/f noise in traffic, inTraffic and granular flow (World Scientific) (1996), p. 73

    work page 1996

  4. [4]

    Nagel, P

    K. Nagel, P. Wagner, R. Woesler, Still flowing: Approaches to traffic flow and traffic jam modeling. Operations research 51 (5), 681–710 (2003)

    work page 2003

  5. [5]

    J. A. Laval, Traffic Flow as a Simple Fluid: Toward a Scaling Theory of Urban Congestion. Transportation Research Record2678 (3), 376–386 (2024)

    work page 2024

  6. [6]

    Zhang, G

    L. Zhang, G. Zeng, D. Li, H.-J. Huang, H. E. Stanley, S. Havlin, Scale-free resilience of real traffic jams. Proceedings of the National Academy of Sciences 116 (18), 8673–8678 (2019)

    work page 2019

  7. [7]

    Nagel, M

    K. Nagel, M. Schreckenberg, A cellular automaton model for freeway traffic. Journal de physique I 2 (12), 2221–2229 (1992)

    work page 1992

  8. [8]

    Roters, S

    L. Roters, S. L¨ ubeck, K. Usadel, Critical behavior of a traffic flow model. Physical Review E 59 (3), 2672 (1999)

    work page 1999

Show all 56 references
  1. [9]

    Critical behavior of a traffic flow model

    D. Chowdhury, J. Kert ´esz, K. Nagel, L. Santen, A. Schadschneider, Comment on “Critical behavior of a traffic flow model”. Physical Review E 61 (3), 3270 (2000)

    2000

  2. [10]

    Comment on ‘Critical behavior of a traffic flow model’

    L. Roters, S. L¨ ubeck, K. Usadel, Reply to “Comment on ‘Critical behavior of a traffic flow model’”. Physical Review E 61 (3), 3272 (2000)

    2000

  3. [11]

    Schadschneider, D

    A. Schadschneider, D. Chowdhury, K. Nishinari,Stochastic transport in complex systems: from molecules to vehicles (Elsevier) (2010)

    2010

  4. [12]

    A. Jha, K. Wiesenfeld, G. Lee, J. Laval, Evidence of Non-Equilibrium Critical Phenomena in a Simple Model of Traffic. arXiv preprint arXiv:2501.08417 (2025). 16

    2025

  5. [13]

    P. Bak, C. Tang, K. Wiesenfeld, Self-organized criticality: An explanation of the 1/f noise. Physical review letters 59 (4), 381 (1987)

    1987

  6. [14]

    Kardar, G

    M. Kardar, G. Parisi, Y.-C. Zhang, Dynamic scaling of growing interfaces. Physical Review Letters 56 (9), 889 (1986)

    1986

  7. [15]

    Halpin-Healy, K

    T. Halpin-Healy, K. A. Takeuchi, A KPZ cocktail-shaken, not stirred... Journal of Statistical Physics 160 (4), 794–814 (2015)

    2015

  8. [16]

    ´Odor, Universality classes in nonequilibrium lattice systems

    G. ´Odor, Universality classes in nonequilibrium lattice systems. Reviews of modern physics 76 (3), 663–724 (2004)

    2004

  9. [17]

    Johansson, Discrete polynuclear growth and determinantal processes

    K. Johansson, Discrete polynuclear growth and determinantal processes. Communications in Mathematical Physics 242 (1), 277–329 (2003)

    2003

  10. [18]

    Corwin, J

    I. Corwin, J. Quastel, D. Remenik, Renormalization fixed point of the KPZ universality class. Journal of Statistical Physics 160 (4), 815–834 (2015)

    2015

  11. [19]

    Spitzer, Interaction of Markov processes, in Random Walks, Brownian Motion, and In- teracting Particle Systems: A Festschrift in Honor of Frank Spitzer (Springer), pp

    F. Spitzer, Interaction of Markov processes, in Random Walks, Brownian Motion, and In- teracting Particle Systems: A Festschrift in Honor of Frank Spitzer (Springer), pp. 66–110 (1991)

    1991

  12. [20]

    Johansson, Shape fluctuations and random matrices

    K. Johansson, Shape fluctuations and random matrices. Communications in mathematical physics 209 (2), 437–476 (2000)

    2000

  13. [21]

    de Gier, A

    J. de Gier, A. Schadschneider, J. Schmidt, G. M. Sch¨ utz, Kardar-parisi-zhang universality of the nagel-schreckenberg model. Physical Review E 100 (5), 052111 (2019)

    2019

  14. [22]

    Gloudemans, Y

    D. Gloudemans, Y. Wang, J. Ji, G. Zachar, W. Barbour, E. Hall, M. Cebelak, L. Smith, D. B. Work, I-24 MOTION: An instrument for freeway traffic science.Transportation Research Part C: Emerging Technologies155, 104311 (2023)

    2023

  15. [23]

    Fortuin, P

    C. Fortuin, P. Kasteleyn, On the random-cluster model: I. Introduction and relation to other models. Physica 57 (4), 536–564 (1972), doi:10.1016/0031-8914(72)90045-6

    1972 doi

  16. [24]

    J. A. Laval, Self-organized criticality of traffic flow: Implications for congestion management technologies. Transportation Research Part C: Emerging Technologies149, 104056 (2023). 17

    2023

  17. [25]

    Materials and methods are available as supplementary material

  18. [26]

    Matetski, J

    K. Matetski, J. Quastel, D. Remenik, The KPZ fixed point.Acta Mathematica227 (1), 115–203 (2021)

    2021

  19. [27]

    Barab ´asi, H

    A.-L. Barab ´asi, H. E. Stanley,Fractal concepts in surface growth(Cambridge university press) (1995)

    1995

  20. [28]

    H. E. Hurst, Long-term storage capacity of reservoirs. Transactions of the American society of civil engineers 116 (1), 770–799 (1951)

    1951

  21. [29]

    Hinrichsen, Non-equilibrium critical phenomena and phase transitions into absorbing states

    H. Hinrichsen, Non-equilibrium critical phenomena and phase transitions into absorbing states. Advances in physics 49 (7), 815–958 (2000)

    2000

  22. [30]

    K. Jang, N. Lichtl ´e, E. Vinitsky, A. Shah, M. Bunting, M. Nice, B. Piccoli, B. Seibold, D. B. Work, M. L. D. Monache, J. Sprinkle, J. W. Lee, A. M. Bayen, Reinforcement Learning Based Oscillation Dampening: Scaling up Single-Agent RL algorithms to a 100 A V highway field ope...

    2024 arXiv

  23. [31]

    J. W. Lee, H. Wang, K. Jang, A. Hayat, M. Bunting, A. Alanqary, W. Barbour, Z. Fu, X. Gong, G. Gunter, et al., Traffic Control via Connected and Automated Vehicles: An Open-Road Field Experiment with 100 CA Vs.arXiv preprint arXiv:2402.17043 (2024)

    2024 arXiv

  24. [32]

    Zhang, M

    Y. Zhang, M. Quinones-Grueiro, Z. Zhang, Y. Wang, W. Barbour, G. Biswas, D. Work, Mar- vel: Multi-agent reinforcement-learning for large-scale variable speed limits. arXiv preprint arXiv:2310.12359 (2023)

    2023 arXiv

  25. [33]

    Gloudemans, D

    D. Gloudemans, D. B. Work, Vehicle tracking with crop-based detection, in 2021 20th IEEE International Conference on Machine Learning and Applications (ICMLA)(IEEE) (2021), pp. 312–319

    2021

  26. [34]

    Gloudemans, G

    D. Gloudemans, G. Zach ´ar, Y. Wang, J. Ji, M. Nice, M. Bunting, W. W. Barbour, J. Sprinkle, B. Piccoli, M. L. D. Monache, et al. , So you think you can track?, in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision(2024), pp. 4528–4538. 18

    2024

  27. [35]

    Y. Wang, D. Gloudemans, J. Ji, Z. N. Teoh, L. Liu, G. Zach´ar, W. Barbour, D. Work, Automatic vehicle trajectory data reconstruction at scale. Transportation Research Part C: Emerging Technologies 160, 104520 (2024)

    2024

  28. [36]

    Y. Wang, J. Ji, W. Barbour, D. B. Work, Online Min Cost Circulation for Multi-Object Tracking on Fragments, in 2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC) (IEEE) (2023), pp. 2803–2810

    2023

  29. [37]

    J. Ji, Y. Wang, W. Barbour, D. B. Work, Platoon trajectory reconstruction with conflict resolu- tion using semidefinite relaxation, in 2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC)(IEEE) (2023), pp. 4332–4338

    2023

  30. [38]

    J. Ji, Y. Wang, D. Gloudemans, G. Zach ´ar, W. Barbour, D. B. Work, Virtual Trajectories for I–24 MOTION: Data and Tools, in 2024 Forum for Innovative Sustainable Transportation Systems (FISTS) (IEEE) (2024), pp. 1–6

    2024

  31. [39]

    K. A. Takeuchi, M. Sano, Universal fluctuations of growing interfaces: evidence in turbulent liquid crystals. Physical review letters 104 (23), 230601 (2010)

    2010

  32. [40]

    R. J. Allen, B. Waclaw, Bacterial growth: a statistical physicist’s guide.Reports on Progress in Physics 82 (1), 016601 (2018)

    2018

  33. [41]

    Myllys, J

    M. Myllys, J. Maunuksela, M. Alava, T. Ala-Nissila, J. Merikoski, J. Timonen, Kinetic rough- ening in slow combustion of paper. Physical Review E 64 (3), 036101 (2001)

    2001

  34. [42]

    Christensen, Percolation theory

    K. Christensen, Percolation theory. Imperial College London 1, 87 (2002)

    2002

  35. [43]

    B. B. Mandelbrot, The Fractal Geometry of Nature (W. H. Freeman and Company, San Fran- cisco) (1982)

    1982

  36. [44]

    Cheng, Multifractal modeling and lacunarity analysis

    Q. Cheng, Multifractal modeling and lacunarity analysis. Mathematical Geology 29, 919–932 (1997)

    1997

  37. [45]

    M. A. Reiss, N. Sabathiel, H. Ahammer, Noise dependency of algorithms for calculating fractal dimensions in digital images. Chaos, Solitons & Fractals 78, 39–46 (2015). 19

    2015

  38. [46]

    Treiber, A

    M. Treiber, A. Kesting, D. Helbing, Three-phase traffic theory and two-phase models with a fundamental diagram in the light of empirical stylized facts. Transportation Research Part B: Methodological 44 (8-9), 983–1000 (2010)

    2010

  39. [47]

    D. Chen, J. A. Laval, S. Ahn, Z. Zheng, Microscopic traffic hysteresis in traffic oscillations: A behavioral perspective. Transportation Research Part B: Methodological46 (10), 1440–1453 (2012)

    2012

  40. [48]

    Serafino, G

    M. Serafino, G. Cimini, A. Maritan, A. Rinaldo, S. Suweis, J. R. Banavar, G. Caldarelli, True scale-free networks hidden by finite size effects. Proceedings of the National Academy of Sciences 118 (2), e2013825118 (2021). Acknowledgments We sincerely thank Dr. Daniel Remenik f...

    2021

  41. [49]

    Make sure the data (𝑥1,...𝑥 𝑚) exhibits no deterministic trends

  42. [50]

    Compute the first differences (increments) of the data: Δ𝑥𝑖 =𝑥𝑖+1−𝑥𝑖

  43. [51]

    Subtract the mean of the segment to obtain centered deviations: 𝑑𝑖 =Δ𝑥𝑖− ¯Δ𝑥

  44. [52]

    Construct the cumulative deviation series: 𝑍𝑘 = Í𝑘 𝑖=1𝑑𝑖 for𝑘 = 1,...,𝑛 − 1

  45. [53]

    Compute the range of cumulative deviations: 𝑅= max(𝑍𝑘)− min(𝑍𝑘)

  46. [54]

    Compute the standard deviation 𝑆 of the segment𝑥𝑖

  47. [55]

    This process is repeated for varying subinterval lengths 𝑛

    The rescaled range is defined as𝑅/𝑆. This process is repeated for varying subinterval lengths 𝑛. The slope of the best-fit line on a log-log plot of 𝑅/𝑆 versus𝑛 corresponds to the Hurst exponent𝐻. Step 1 above introduces an additional parameter, the window size parameter 𝑟, ne...

  48. [56]

    3 and 4 show improved agreement in the plateau region when small clusters are excluded

    We observe that the hyper-scaling relationships of Eqs. 3 and 4 show improved agreement in the plateau region when small clusters are excluded. In contrast, the hyper-scaling relationship for the delay fractal dimension (Eq. 7) exhibits higher errors within the plateau range b...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.