Pith. sign in

REVIEW 4 major objections 6 minor 17 references

Adaptive Traffic-Following Scheme for Orderly Distributed Control of Multi-Vehicle Systems

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper claims that letting each aircraft adjust its traffic-following strength from local traffic density yields lower travel times than any fixed setting, with only minimal extra airspace disorder.

desk verdict Interesting incremental heuristic with a solid time-discounting result, but the central local-adaptation claim is untested and Eq. 6 has what looks like a sign error. read the letter →

arxiv 2506.00703 v1 pith:ER6PHYIZ submitted 2025-05-31 cs.MA cs.ETcs.ROcs.SYeess.SY

Anahita Jain , Husni Idris , John-Paul Clarke , Daniel Delahaye This is my paper
classification cs.MAcs.ETcs.ROcs.SYeess.SY
keywords self-organizingcontroltraffic-followingmulti-agentsystemsairspaceoperationsentropydistributedadaptiveroutingtrafficpatternmap
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

Previous work on this system found that at high densities aircraft save time by following established traffic patterns, while at low densities direct paths win. This paper turns that dichotomy into a distributed control rule: each aircraft continuously recomputes a traffic-following factor $k_t$ from a sigmoid function of the local aircraft density in its sensing radius, and re-plans its path accordingly. In simulations with a time-varying traffic profile, this adaptive rule produced travel times 11% below the best fixed setting ($k_t=6$) and 21% below no traffic-following ($k_t=0$), while adding only a small amount of airspace entropy. The paper presents the result as evidence that self-organizing behavior, rather than fixed rules or central coordination, is the scalable way to manage dense autonomous airspace operations.

What carries the argument

The load-bearing object is the adaptive traffic-following factor $k_t$, a sigmoid of local aircraft density, $k_t = 6.024/(1+e^{(-\rho/0.0005 - 15.193)})$, with $\rho$ measured as aircraft per square mile inside the sensing radius $R_s$. This gain modulates the cell traversal cost $(1 - k_t \hat{t}_{i,j})$, where $\hat{t}_{i,j}$ is the normalized count of prior aircraft that used that entry-exit pair; higher $k_t$ makes well-used paths cheaper and therefore more attractive. Each ownship recomputes $k_t$ every 100 seconds and replans its route with Dijkstra's algorithm on a graph whose nodes are cell edges, so the same mechanism produces direct routing at low density and pattern-following at high density. A time-discounted traffic pattern map (the preceding 500 seconds) supplies the $\hat{t}_{i,j}$ values, and Shannon entropy over the traversal counts measures the resulting airspace order.

What would settle it

Run the identical traffic profile, path planning, and $R_s = 25$ miles, but shift the sigmoid's crossover density in Eq. (6) by a factor of two; if the resulting travel times are no better than the fixed $k_t = 6$ or $k_t = 0$ cases, then the benefit comes from the specific fitted curve, not from the adaptive mechanism itself.

Watch

Extended reading notes

Core claim

The central claim is that a purely local, density-driven adjustment of traffic-following behavior gives the collective the best of both congestion regimes: aircraft follow traffic only when it pays, and return to direct routing when the airspace empties. To show this, the authors model a two-dimensional hexagonal-cell airspace in which each ownship records entry-exit traversals in a traffic pattern map, discounts patterns older than 500 seconds, plans a least-cost path with Dijkstra's algorithm, and sets $k_t$ using the fitted sigmoid $k_t = 6.024/(1+e^{(-\rho/0.0005 - 15.193)})$ based on the density $\rho$ in its range $R_s$. The dynamic rule achieved the lowest average travel times of every setting tested, and entropy stayed close to the high-following fixed cases. The authors conclude that adaptive traffic-following should replace fixed-rule following, and that an intermediate sensing range ($R_s = 25$ miles) is best because very short ranges lack information while very long ranges trigger unhelpful following in distant congestion.

Load-bearing premise

The sigmoid in Eq. (6) was fitted to earlier simulations in which every aircraft had complete grid-wide information and kept one fixed $k_t$ for the entire flight; the method assumes this same mapping is optimal when aircraft sense only a local region and change $k_t$ every 100 seconds. If the fitted mapping does not transfer to local, dynamic conditions, the reported travel-time benefits collapse.

Editorial extensions

If this is right

  • Adaptive $k_t$ reduces average travel time by 11% compared with the best fixed setting ($k_t=6$, p = 0.0045) and by 21% compared with no traffic-following ($k_t=0$), while adding only minimal airspace entropy.
  • Discounting the traffic pattern map to the preceding 500 seconds cuts average travel time by 18% (p = 2.19E-13), because old 'selfish' paths from low-density periods mislead aircraft once density rises.
  • A sensing radius of $R_s = 25$ miles outperforms both smaller and larger radii by 12% in travel time, suggesting the existence of an optimal spatial horizon for local traffic-following decisions.
  • The authors position this scheme as the self-organization layer between self-separation and self-limitation, arguing that such adaptive order creation is necessary for scaling autonomous airspace operations to high density.

Reading between the lines

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

  • A natural extension the paper does not develop is to learn the density-to-$k_t$ mapping online instead of fitting it once to global-information simulations; a system that adapts the mapping per sector could be more robust to traffic distributions the sigmoid was never fitted on.
  • Because the entropy metric counts only direction diversity and traversal counts, the 'minimal additional disorder' finding leaves open whether the extra path variety trades against separation buffers, controller workload, or fuel burn in a more realistic setting.
  • The optimal $R_s = 25$ miles is tied to the 2.5-mile cell edge, 250-knot speed, and 500-second discount window, so the concrete radius should be rescaled rather than transferred directly to other airspace designs or to ground-robot swarms.
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

4 major / 6 minor

Summary. The paper proposes an adaptive extension of the authors' earlier traffic-following model for distributed autonomous aircraft. In the model, each aircraft computes shortest paths over a hexagonal grid using a cost function that discounts edge pairs used by previous traffic, with the strength of the discount controlled by a traffic-following factor k_t. The new contribution is to let each aircraft update k_t every 100 seconds from the local aircraft density rho within a sensing radius R_s, using a sigmoid curve (Eq. 6) fitted to the authors' previous fixed-k_t simulations. Experiments compare travel time and a cumulative entropy metric under temporal discounting of the traffic map, under fixed versus dynamic k_t, and under R_s values of 15, 25, 35, and 50 miles. The headline results are that dynamic k_t reduces average travel time by 11% relative to the best fixed value tested (k_t=6) and by 21% relative to k_t=0, with reported p-values of 0.0045 and smaller, and that R_s=25 yields a 12% reduction relative to the other ranges; Section V further states that the scheme 'always result[s] in the best travel times.'

Significance. If the dynamic local rule were validated, this would be a useful step toward scalable self-organizing airspace operations: it would show that agents can obtain much of the benefit of global traffic-following from local density information, and the sensitivity analysis of temporal discounting and range would give practical design guidance. The paper deserves credit for comparing against fixed-k_t baselines, for reporting p-values over multiple simulation runs, for specifying the simulation protocol in detail (traffic profile, cell size, hold time, update period), and for including a sensitivity analysis over R_s. However, the manuscript does not ship code or data, and the key transfer from a globally fitted sigmoid to a dynamic, locally sensed control law is not validated; moreover, the printed form of Eq. (6) appears inconsistent with the described adaptive behavior, and the main optimality statement is stronger than the evidence. These gaps, together with the cumulative entropy metric, make the contribution conditional rather than established.

major comments (4)
  1. [II-E, Eq. (6)] Equation (6) as printed is k_t = 6.024 / (1 + exp(-rho/0.0005 - 15.193)). At rho=0 this evaluates to approximately 6.024, and because the exponential term is at most exp(-15.193) for all non-negative rho, k_t remains essentially 6.024 at every density. The text and Fig. 6 describe k_t as low at low density and increasing with density, so the printed formula cannot be the function used in the simulations. If the intended argument was exp(-(rho/0.0005 - 15.193)) or exp(-rho/0.0005 + 15.193), the typesetting must be corrected; as written, the adaptive mechanism is effectively constant and the reported travel-time comparison with fixed k_t=6 cannot be reproduced from the stated equation.
  2. [II-E; III-B; III-C] The sigmoid in Eq. (6) is fitted to data points from the authors' previous fixed-k_t simulations (Fig. 4 and Refs. [12][13]), where every aircraft used the same global k_t and had a range equal to the whole grid; the manuscript itself notes this in Section II-E. In the headline comparison of Section III-B, R_s is set to the entire grid, so all aircraft select the same k_t at each update; that experiment therefore validates a time-varying global gain, not independent local adaptation. Section III-C varies R_s but does not compare dynamic local k_t against fixed-k_t baselines under the same traffic-map window and cell-capacity rules, so the benefit observed at R_s=25 could be an effect of the sensing radius rather than of adaptation. The transferability of the fitted mapping to locally sensed, dynamically updated conditions is the load-bearing assumption and is untested; without it, the travel-time claims do not establish the paper's central thesis.
  3. [III-B and V] The statement in Section V that the scheme results in 'the best travel times', and the analogous 'should always be lower than or equal' assertion in Section III-B, is not supported by the evidence. The simulations compare the dynamic rule with four fixed values of k_t under one traffic profile; no optimality bound is derived, and Eq. (6) is an approximation without reported residuals, confidence intervals, or validation on held-out data. The supported claim is that the dynamic rule improves on the tested fixed settings, not that it is always optimal.
  4. [II-D, Figs. 9, 12, 15] The entropy metric is the sum over cells and over simulation time of entropies computed from the accumulated traffic matrices, so it is monotonically non-decreasing by construction, as the paper itself states. Consequently, the observation that entropy increased over time is a property of the estimator, and comparisons of cumulative entropy curves do not directly support the abstract's claim of 'minimal levels of additional disorder.' The authors should report incremental or instantaneous entropy (for example, entropy added per update window or normalized by the number of traversals), or explicitly justify why the cumulative quantity is the right measure of the disorder cost.
minor comments (6)
  1. [I, last paragraph] The text says 'In Section V we discuss potential applications' and then 'in Section V we present conclusions'; the first reference should be to Section IV and the second to Section V.
  2. [II-E, Eq. (6)] Even after correcting the sign, please report how the sigmoid was fitted to the selected 'least travel time' data points from Fig. 4, including residuals, confidence intervals, and whether any data were held out for validation.
  3. [III, Experimental setup] The statement that 15 simulations were 'found sufficient by statistical analysis' should be accompanied by the actual variability or power calculation, and the p-values should be complemented with confidence intervals or effect sizes.
  4. [III-A] The choice of the 500-second discount threshold is described as found by some trial runs; either include a sensitivity analysis over this threshold or state clearly that the threshold is an experimental parameter rather than a tuned result.
  5. [II-B, Eq. (3)] The notation '16x6' should be written as a bold one-matrix, for example \mathbf{1}_{6\times6}, and defined explicitly at first use.
  6. [Figs. 5 and 6] Figures 5 and 6 should include axis labels and units, and the caption should state whether the curves are averages over the 15 runs; the shape of Fig. 6 should also be reconciled with the printed Eq. (6).

Circularity Check

1 steps flagged · score 4.0 of 10

The dynamic traffic-following benefit is partly inherited from Eq. (6), a sigmoid fitted to the authors' own prior fixed-k_t simulations; however, the fixed-k_t comparisons and R_s sensitivity runs give the claim independent empirical content.

  1. fitted input called prediction [Section II-E (Eq. 6); Section III-B; Section V]
    "By referring to past results, as shown in Fig. 4, we model a function that will output the ideal value of the traffic-following factor k_t, given the amount of vehicular congestion detected by an ownship. This ideal value of k_t will enable an aircraft to reach its destination in the shortest possible time, given the current state of the airspace. ... It should be noted here that when R_s spans the entire grid, each aircraft considers all aircraft being in its range, resulting in all aircraft selecting the same value for k_t."

    Eq. (6) is a sigmoid fitted to the least-travel-time points from the authors' prior fixed-k_t simulations (Fig. 4, Refs. [12][13]). In the headline comparison of Section III-B, R_s is the entire grid, so all aircraft follow the same time-varying k_t schedule, determined by Eq. (6) from global density. Thus the dynamic case is essentially the fitted envelope of the best fixed-k_t choices applied over time; reporting that it beats any single fixed k_t partly restates that the envelope of the fixed-k_t travel-time curves lies below each individual curve. The label 'ideal' and the conclusion that adaptive behavior 'always result[s] in the best travel times' (Section V) import optimality from the fit rather than establish it.

full rationale

The paper's core experiment compares a dynamically adjusted k_t against fixed k_t values in the same simulation framework. This comparison has independent content: the dynamic rule could in principle fail to preserve the benefit due to switching effects, and the R_s sensitivity study is a new contribution. However, the dynamic rule itself is not derived from first principles; Eq. (6) is fit to the authors' own prior fixed-k_t simulation data, and the paper then calls this fitted mapping 'ideal' and claims it 'will enable an aircraft to reach its destination in the shortest possible time.' When R_s covers the whole grid, all aircraft share the same k_t, so the dynamic case is, by construction, a time-varying version of the very fixed-k_t curves used to fit the sigmoid. This makes the reported travel-time improvement partially a restatement of the fit rather than an independent prediction. The self-citations to [12][13] are load-bearing for the 'ideal' mapping, but the paper also re-runs fixed-k_t baselines, so the comparison is not wholly circular. The strongest overclaim, 'always resulting in the best travel times,' is an assertion of the fit's optimality, not a proven bound. Overall, the central claim retains independent empirical content, but the fitted-input issue warrants a moderate circularity score.

Assumptions & free parameters 8 free parameters · 7 assumptions · 0 invented entities

The central claim rests on a sigmoid controller fitted to data from the authors' own prior simulations, several hand-picked simulation parameters (discount threshold, traffic profile, hold time), and domain assumptions about cell capacity, entropy, and information availability. No new physical entities are introduced.

free parameters (8)
  • sigmoid maximum k_t = 6.024
    Upper asymptote fitted to selected minimum-travel-time points from Fig. 4 of prior fixed-k_t studies.
  • sigmoid density scale = 0.0005 aircraft per square mile
    Steepness parameter fitted to the observed crossover from k_t=0 to k_t=6 in Fig. 4.
  • sigmoid offset = 15.193
    Location parameter fitted to the crossover density in Fig. 4.
  • discount threshold = 500 seconds
    Chosen ad hoc: 'was found to be adequate for this experiment based on some trial runs' (Section III-A).
  • hold time t_hold = 150 seconds
    Simulation parameter for how long an ownship waits for an occupied cell before replanning.
  • traffic profile = N = {4,5,4,5,40,20,10,4,5,6,4,3,10,40,20,10,5,10,30,20,10,6,5,3} over 24 time points
    Hand-picked schedule to exercise varying congestion levels; no justification or comparison to other profiles.
  • range values R_s = {15, 25, 35, 50} miles
    Experimental factor; 25 miles identified as optimal for this setup, but this may vary with traffic profile, cell size, and speed.
  • cell edge length and aircraft speed = 2.5 miles and 250 mph
    Hand-chosen simulation constants; conclusions might change with different values.
assumptions (7)
  • domain assumption The airspace is two-dimensional and partitioned into regular hexagonal cells.
    Section II-A: the paper explicitly assumes a 2D airspace partition.
  • domain assumption Cell capacity limit of one aircraft ensures minimum separation without a conflict resolution model.
    Section II-C: 'we limit cell capacity to one... We assume this capacity limit, along with the cell size, ensures aircraft do not violate minimum horizontal separation requirements.'
  • domain assumption Traffic counts in each entry-exit pair are a sufficient basis for path cost.
    Section II-B: the traffic cost is defined as (1 - k_t * normalized count), capturing direction similarity and volume only.
  • domain assumption Shannon entropy of cumulative traffic counts measures airspace order.
    Section II-D: entropy is computed from the cumulative traffic matrix, so it rises by construction as new directions accumulate.
  • ad hoc to paper A sigmoid is a sufficient functional form for the density-to-k_t mapping.
    Section II-E: 'the crossover point... suggests that a sigmoid curve might be sufficient'; no comparison to other functional forms.
  • standard math Dijkstra's algorithm on the edge-graph returns the least-cost path.
    Section II-C, standard shortest-path result.
  • domain assumption Aircraft have access to the traffic pattern map via sensors or a broadcast service.
    Section II-A: information availability is assumed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Adaptive Traffic-Following Scheme for Orderly Distributed Control of Multi-Vehicle Systems." pith.science (2026). https://pith.science/paper/ER6PHYIZ

@misc{pith2026250600703,
  author       = {Pith},
  title        = {Pith review of: Adaptive Traffic-Following Scheme for Orderly Distributed Control of Multi-Vehicle Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ER6PHYIZ}},
  note         = {Machine review of arXiv:2506.00703}
}
read the original abstract

We present an adaptive control scheme to enable the emergence of order within distributed, autonomous multi-agent systems. Past studies showed that under high-density conditions, order generated from traffic-following behavior reduces travel times, while under low densities, choosing direct paths is more beneficial. In this paper, we leveraged those findings to allow aircraft to independently and dynamically adjust their degree of traffic-following behavior based on the current state of the airspace. This enables aircraft to follow other traffic only when beneficial. Quantitative analyses revealed that dynamic traffic-following behavior results in lower aircraft travel times at the cost of minimal levels of additional disorder to the airspace. The sensitivity of these benefits to temporal and spatial horizons was also investigated. Overall, this work highlights the benefits, and potential necessity, of incorporating self-organizing behavior in making distributed, autonomous multi-agent systems scalable.

Figures

Figures reproduced from arXiv: 2506.00703 by the authors.

Figure 1
Figure 1. Layered control loops of self-separation, self [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Partitioning the airspace into a hexagonal grid, with [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Simulation snapshots demonstrating the effect of [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: Results showing the effect of fixed values of the [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 6
Figure 6. Figure 6: Plot depicting an ownship’s adjustment of the [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Effect of discounting the traffic pattern map with [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Effect of discounting the traffic pattern map with [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: Effect of discounting the traffic pattern map with [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 10
Figure 10. Figure 10: Traffic profiles for cases when the traffic-following [PITH_FULL_IMAGE:figures/full_fig_p008_10.png]
Figure 12
Figure 12. Figure 12: Average airspace entropy for cases when the [PITH_FULL_IMAGE:figures/full_fig_p008_12.png]
Figure 11
Figure 11. Figure 11: Average aircraft travel times for cases when the [PITH_FULL_IMAGE:figures/full_fig_p008_11.png]
Figure 14
Figure 14. Figure 14: Effect of varying the range (Rs) on the average aircraft travel times, with all Rs values in miles. of this might not ultimately be beneficial. For example, if one corner of the airspace is congested, an aircraft in a less congested corner might not benefit from follo…
Figure 13
Figure 13. Figure 13: shows how the average number of aircraft in the airspace varied over the course of the simulation across different cases, even though they were introduced into the airspace at the same time in each case. Although they all showed similar traffic volumes at the beginnin…
Figure 15
Figure 15. Figure 15: shows the average amount of airspace entropy over the course of the simulation for the different cases. Similar to III-B, airspace entropy remained at similar levels for all cases. This indicates that, while there might be travel time benefits that certain ranges offe…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

17 extracted references · 14 canonical work pages

  1. [12]

    Benefits of Traffic-Following in High-Density Au- tonomous Airspace Operations

    Anahita Jain, Husni R Idris, and John-Paul Clarke. “Benefits of Traffic-Following in High-Density Au- tonomous Airspace Operations”. In:AIAA AVIATION FORUM AND ASCEND 2024. 2024, p. 4629.DOI: https://doi.org/10.2514/6.2024-4629

    work page doi:10.2514/6.2024-4629 2024

  2. [13]

    Impact of Traffic-Following on Order of Autonomous Airspace Operations

    Anahita Jain, Husni R Idris, and John-Paul Clarke. “Impact of Traffic-Following on Order of Autonomous Airspace Operations”. In:11th International Confer- ence on Research in Air Transportation. 2024

    work page 2024

  3. [1]

    David Wing et al.Digital Flight: A New Cooperative Operating Mode to Complement VFR and IFR. Tech. rep. 2022

    work page 2022

  4. [2]

    Tom G Reynolds et al.Structure, Intent and Confor- mance Monitoring in ATC. Tech. rep. 2002

    work page 2002

  5. [3]

    Estimating airspace ca- pacity based on risk mitigation metrics

    Husni Idris and Ni Shen. “Estimating airspace ca- pacity based on risk mitigation metrics”. In:Tenth USA/Europe Air Traffic Management R&D Seminar, Chicago. 2013

    work page 2013

  6. [4]

    Airspace complexity measurement: An air traffic control simulation anal- ysis

    Parimal H Kopardekar et al. “Airspace complexity measurement: An air traffic control simulation anal- ysis”. In:International Journal of Industrial Engineer- ing: Theory, Applications and Practice16.1 (2009), pp. 61–70.DOI: https://doi.org/10.23055/ijietap.2009. 16.1.18

    work page doi:10.23055/ijietap.2009 2009

  7. [5]

    Air Traffic Complexity Map Based on Linear Dynamical Systems

    Daniel Delahaye et al. “Air Traffic Complexity Map Based on Linear Dynamical Systems”. In:Aerospace 9.5 (2022), p. 230.DOI: https : / / doi . org / 10 . 3390 / aerospace9050230

    work page 2022

  8. [6]

    A Framework for Sense and Follow Convoys for Collective Autonomous Mobility

    Abraham K Ishihara, Husni R Idris, and Min Xue. “A Framework for Sense and Follow Convoys for Collective Autonomous Mobility”. In:AIAA AVIATION FORUM. 2021, p. 2329.DOI: https://doi.org/10.2514/ 6.2021-2329

    work page 2021

Show all 17 references
  1. [7]

    Aircraft Route Optimization for For- mation Flight

    Jia Xu et al. “Aircraft Route Optimization for For- mation Flight”. In:Journal of Aircraft51.2 (2014), pp. 490–501.DOI: https://doi.org/10.2514/1.C032154

    2014 doi

  2. [8]

    Self-organized free-flight arrival for urban air mobil- ity

    Martin Waltz, Ostap Okhrin, and Michael Schultz. “Self-organized free-flight arrival for urban air mobil- ity”. In:Transportation Research Part C: Emerging Technologies167 (2024), p. 104806.DOI: 10 . 1016 / j.trc.2024.104806

    2024

  3. [9]

    Non-Communicative Negotiation-Free Collision Avoidance

    Anahita Jain, John-Paul Clarke, and Fanruiqi Zeng. “Non-Communicative Negotiation-Free Collision Avoidance”. In:AIAA AVIATION Forum. 2023, p. 3627.DOI: https://doi.org/10.2514/6.2023-3627

    2023 doi

  4. [10]

    Distributed Trajectory Flexibility Preservation for Traffic Complexity Mitigation

    Husni Idris, Daniel Delahaye, and David Wing. “Distributed Trajectory Flexibility Preservation for Traffic Complexity Mitigation”. In:ATM 2009, 8th USA/Europe Air Traffic Management R&D Seminar

    2009

  5. [11]

    Benefits of Traffic-Following in High-Density Au- tonomous Airspace Operations

    Anahita Jain, Husni R Idris, and John-Paul Clarke. “Benefits of Traffic-Following in High-Density Au- tonomous Airspace Operations”. In:AIAA Journal of Air Transportation 2025. 2025.DOI: https://doi.org/10. 2514/1.D0486

    2025

  6. [14]

    Patel.Hexagonal grids

    Amit J. Patel.Hexagonal grids. Mar. 2013.URL: https: //www.redblobgames.com/grids/hexagons/

    2013

  7. [15]

    Courier Corporation, 1998.DOI: 10.1109/TASSP.1984

    Christos H Papadimitriou and Kenneth Steiglitz.Com- binatorial Optimization: Algorithms and Complexity. Courier Corporation, 1998.DOI: 10.1109/TASSP.1984. 1164450

    1998 doi

  8. [16]

    A note on two problems in con- nexion with graphs

    Edsger W Dijkstra. “A note on two problems in con- nexion with graphs”. In:Edsger Wybe Dijkstra: His Life, Work, and Legacy. 2022, pp. 287–290.DOI: https: //doi.org/10.1145/3544585.3544600

    2022

  9. [17]

    A generalization of Shannon’s infor- mation theory

    Chen-Guang Lu. “A generalization of Shannon’s infor- mation theory”. In:International Journal Of General System28.6 (1999), pp. 453–490.DOI: https://doi.org/ 10.1080/03081079908935247. BIOGRAPHIES Anahita Jainis a Ph.D. candidate in the Aerospace Engineering and Engineering Me...

    1999 doi

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.