pith. sign in

arxiv: 2604.19641 · v1 · submitted 2026-04-21 · 🧮 math.OC

Regulation Zero 2: A Flow-Centric Sequential Regulation Planning Framework to Counter Regulation Cascading in Pre-tactical Air Traffic Flow Management

Thinh Hoang , Zhengyi Wang , Leila Zerrouki , Daniel Delahaye This is my paper

Pith reviewed 2026-05-10 01:55 UTC · model grok-4.3

classification 🧮 math.OC
keywords regulation cascadingair traffic flow managementMonte Carlo Tree Searchsequential regulation planningflow-level regulationspre-tactical ATFMFPFS allocatornetwork optimization
0
0 comments X

The pith

A sequential flow-centric planner using Monte Carlo tree search mitigates regulation cascading in air traffic management.

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

The paper tries to establish that planning ordered sequences of flow-level regulations with a hierarchical Monte Carlo Tree Search can effectively counter the unpredictable compounding interactions known as regulation cascading in air traffic flow management. This matters because rising demand and workforce shortages increase the use of regulations, making their interactions a key barrier to reliable sector protection. The method samples congestion hotspots, proposes regulations via a local engine, and evaluates them with a fast First-Planned-First-Served allocator to guide the search. Experiments on pan-European peak days show higher objective gains with less network impact than flight-centric baselines, and ablations confirm cascading can cut effectiveness by half. It also keeps slot fairness and allows expert input for practical automation.

Core claim

Regulation Zero 2 is a flow-centric framework that optimizes compatible sequences of regulations using hierarchical MCTS, where hotspot sampling and local proposals are scored by fast FPFS to find effective plans. On multiple summer-peak traffic days, it outperforms flight-centric simulated annealing and NSGA-II in objective improvements while limiting the scope of network impact. Ablations reveal that regulation cascading can reduce up to 50% of potential effectiveness, underscoring the value of sequential compatibility-aware planning.

What carries the argument

Hierarchical Monte Carlo Tree Search that first identifies congestion hotspots and then selects regulation proposals from a local engine, with rewards estimated by a fast First-Planned-First-Served allocator.

If this is right

  • Ordered regulation sequences can be found that are compatible with existing slot-allocation systems such as CASA and RBS++.
  • The planner achieves consistent performance improvements across various pan-European summer-peak traffic scenarios.
  • A tighter scope of impact on the network is maintained compared to flight-centric optimization approaches.
  • FPFS fairness is preserved while supporting the injection of expert knowledge into the planning process.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Applying similar sequential planning to other domains with cascading constraints, like railway scheduling or supply chain flows, could yield comparable benefits.
  • Deploying this in operational ATFM systems might enable more automated, less disruptive regulation during high-demand periods.
  • Testing the framework with real-time updates could extend its use beyond pre-tactical phases.

Load-bearing premise

The fast FPFS allocator supplies sufficiently accurate reward estimates to steer the MCTS toward effective global sequences, and the local proposal engine produces candidates that include the high-impact regulations.

What would settle it

If experiments on new peak traffic days show that Regulation Zero 2 does not deliver markedly higher objective improvements than the flight-centric baselines, the claim of effective cascading mitigation would be weakened.

Figures

Figures reproduced from arXiv: 2604.19641 by Daniel Delahaye, Leila Zerrouki, Thinh Hoang, Zhengyi Wang.

Figure 1
Figure 1. Figure 1: The general Flow Regulation Workflow in the European Context. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Regulations (EU) and Flow Programs (US) that will serve as input for slot allo [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Comparison of the MPR and RZ delay-arbitration schemes when the same flight [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Illustration of Trajectory (left) and Regulation Spaces (right). In Trajectory Space, [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Comparison between the optimization pathways of slot-allocation algorithms and [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Illustration of the computation of the NomRel and InLoad heuristics. In essence, [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: NomRel’s flow selection power adjusted for the optimal entry rate. [PITH_FULL_IMAGE:figures/full_fig_p016_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: InLoad vs Actual Rate-Optimal Network-wide Excess Relief [PITH_FULL_IMAGE:figures/full_fig_p017_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Bin Trend Plots of NomRel and InLoad against [PITH_FULL_IMAGE:figures/full_fig_p017_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Regulation Zero Key Components variance on the positive half of JCAP increases much more rapidly than the general trend. For InLoad, the low-InLoad regime, where the InLoad value is less than 80, has the best chance of yielding a good JCAP score after rate tuning. These findings can be explained as follows: a good flow candidate is one that is not too large in size; otherwise, it can significantly increas… view at source ↗
Figure 11
Figure 11. Figure 11: Demand-impact similarity calculation for two flights [PITH_FULL_IMAGE:figures/full_fig_p019_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Sample flow extracted by the community detection algorithm for hotspot segment [PITH_FULL_IMAGE:figures/full_fig_p020_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: The Regulation Proposal Engine’s Main Components, with [PITH_FULL_IMAGE:figures/full_fig_p022_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Illustration of the three main phases of the MCTS algorithm employed by Regu [PITH_FULL_IMAGE:figures/full_fig_p023_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Objective Improvement by Case, by Algorithm. Higher is better. RZ shows [PITH_FULL_IMAGE:figures/full_fig_p028_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Traffic-Volume Gini indices measure the delay distribution fairness across Traffic [PITH_FULL_IMAGE:figures/full_fig_p031_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Comparison of objective improvement score against relative run time. Higher is [PITH_FULL_IMAGE:figures/full_fig_p032_17.png] view at source ↗
Figure 19
Figure 19. Figure 19: Comparison of JCAP score improvement against relative run time. Higher is better. 34 [PITH_FULL_IMAGE:figures/full_fig_p034_19.png] view at source ↗
Figure 20
Figure 20. Figure 20: Comparison of step-wise contribution of JCAP to the general objective function. Higher is better. 35 [PITH_FULL_IMAGE:figures/full_fig_p035_20.png] view at source ↗
Figure 21
Figure 21. Figure 21: Reward per step and cumulative reward with the number of regulations imposed. [PITH_FULL_IMAGE:figures/full_fig_p036_21.png] view at source ↗
Figure 22
Figure 22. Figure 22: Comparison of ratio metrics for 3 subcases: 6h, 12h and full-day for the 17/07 [PITH_FULL_IMAGE:figures/full_fig_p037_22.png] view at source ↗
Figure 23
Figure 23. Figure 23: Comparison of ratio metrics for 3 subcases: 6h, 12h and full-day for the 18/07 [PITH_FULL_IMAGE:figures/full_fig_p038_23.png] view at source ↗
Figure 24
Figure 24. Figure 24: Comparison of ratio metrics for 3 subcases: 6h, 12h and full-day for the 20/07 [PITH_FULL_IMAGE:figures/full_fig_p039_24.png] view at source ↗
Figure 25
Figure 25. Figure 25: Comparison of ratio metrics for 3 subcases: 6h, 12h and full-day for the 05/08 [PITH_FULL_IMAGE:figures/full_fig_p040_25.png] view at source ↗
Figure 26
Figure 26. Figure 26: Comparison of Flow Threshold / Flow Size effects for four full-day cases of 17/07, [PITH_FULL_IMAGE:figures/full_fig_p041_26.png] view at source ↗
Figure 27
Figure 27. Figure 27: Final Objective Improvement comparison between RZ v1 and v2 for reduced [PITH_FULL_IMAGE:figures/full_fig_p042_27.png] view at source ↗
Figure 28
Figure 28. Figure 28: RZ’s Pareto Frontier for six full-day cases with [PITH_FULL_IMAGE:figures/full_fig_p043_28.png] view at source ↗
Figure 29
Figure 29. Figure 29: Variation of Exceedance Reduced per Minute Delay when [PITH_FULL_IMAGE:figures/full_fig_p043_29.png] view at source ↗
Figure 30
Figure 30. Figure 30: Improvement of Objective between RZ v2 (MCTS-enabled) and BRPP (MCTS [PITH_FULL_IMAGE:figures/full_fig_p045_30.png] view at source ↗
Figure 31
Figure 31. Figure 31: Positioning of RZ against flight-centric methods in categories of DCB problems. [PITH_FULL_IMAGE:figures/full_fig_p046_31.png] view at source ↗
read the original abstract

Air Traffic Flow Management (ATFM) traffic regulations are being increasingly used as rising demand meets persistent workforce shortages. This operational strain has amplified a critical phenomenon that we call \emph{regulation cascading}: the compounding, non-linear interactions that occur when multiple regulations influence one another in unpredictable ways. As the number and complexity of regulations grow, cascading effects become more pronounced, undermining the network operator's ability to protect sectors reliably. To address this challenge, we introduce Regulation Zero 2, an updated sequential planning framework that natively operates in the regulation space, optimizing over ordered sequences of flow-level regulations that remain compatible as much as possible with existing slot-allocation systems such as CASA and RBS++. We equipped Regulation Zero 2 with new heuristics to render flow finding more efficient. At its core, the method employs a hierarchical Monte Carlo Tree Search (MCTS) that first samples congestion hotspots and then selects candidate regulations synthesized by a local proposal engine. Each proposal is evaluated by a fast First-Planned-First-Served (FPFS) allocator to estimate its reward, with these feedbacks guiding the subsequent MCTS exploration. Experiments on many pan-European summer-peak traffic days that Regulation Zero delivers promising and consistent performance. Compared to a flight-centric simulated-annealing and NSGA-II baselines, it achieves markedly higher objective improvements, while maintaining a tighter scope of impact on the network. Ablation studies also found that Regulation Cascading could reduce up to 50\% of potential effectiveness. RZ also preserves FPFS fairness and supports expert knowledge injection, offering a pragmatic and low-disruption pathway toward automation in operations.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript introduces Regulation Zero 2, a flow-centric sequential planning framework for pre-tactical air traffic flow management that employs hierarchical Monte Carlo Tree Search (MCTS) over ordered sequences of flow-level regulations. Proposals are generated by a local engine and evaluated via a fast FPFS allocator to estimate rewards, with the goal of mitigating regulation cascading effects. Experiments on pan-European summer-peak traffic days are reported to show markedly higher objective improvements than flight-centric simulated-annealing and NSGA-II baselines, a tighter network impact scope, and an ablation indicating that cascading can reduce up to 50% of potential effectiveness, while preserving FPFS fairness.

Significance. If the performance claims and proxy validity hold, the framework could provide a pragmatic, low-disruption pathway for automating regulation planning in ATFM operations that respects existing slot-allocation systems. The explicit modeling of cascading interactions and use of MCTS for sequence optimization address a real and growing operational strain from rising demand and workforce shortages.

major comments (2)
  1. [Abstract and Experiments section] Abstract and Experiments section: The central claim of superior performance rests on FPFS serving as a faithful reward proxy inside the MCTS tree search. No demonstration is given that FPFS rankings or relative magnitudes correlate with outcomes from the operational allocator (CASA/RBS++) on identical candidate sets, despite the abstract noting that cascading effects are non-linear and can reduce effectiveness by 50%. This is load-bearing for the claim that the hierarchical MCTS yields globally effective sequences.
  2. [Experiments section] Experiments section: The abstract asserts 'promising and consistent performance' and 'markedly higher objective improvements' across 'many' pan-European summer-peak days, yet supplies no details on the exact number of days, the precise objective function, statistical significance testing, variance across instances, or controls against post-hoc selection of favorable scenarios. Without these, the reported gains cannot be independently verified or compared to the baselines.
minor comments (1)
  1. [Abstract] Abstract: The sentence beginning 'Experiments on many pan-European summer-peak traffic days that Regulation Zero delivers...' is grammatically incomplete and should be revised for clarity (e.g., insert 'show that').

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback and the recommendation for major revision. The comments highlight important aspects of proxy validation and experimental transparency that we will address to strengthen the manuscript. Below we respond point by point to the major comments.

read point-by-point responses

  1. Referee: [Abstract and Experiments section] Abstract and Experiments section: The central claim of superior performance rests on FPFS serving as a faithful reward proxy inside the MCTS tree search. No demonstration is given that FPFS rankings or relative magnitudes correlate with outcomes from the operational allocator (CASA/RBS++) on identical candidate sets, despite the abstract noting that cascading effects are non-linear and can reduce effectiveness by 50%. This is load-bearing for the claim that the hierarchical MCTS yields globally effective sequences.

    Authors: We agree that demonstrating the correlation between FPFS-based rewards and those from an operational allocator such as CASA/RBS++ would provide stronger support for the proxy's validity, particularly given the non-linear cascading effects noted in the ablation. FPFS was selected as the reward estimator because it is computationally efficient, deterministic, and preserves the first-planned-first-served fairness principle central to existing slot allocation systems, enabling scalable MCTS search. The 50% effectiveness reduction from cascading was quantified via ablation under this same proxy. In the revised manuscript we will add a dedicated validation subsection that applies both FPFS and a re-implemented CASA-like allocator to the same set of candidate regulation sequences drawn from the test instances, reporting rank correlation (Spearman) and relative objective error to quantify alignment. revision: yes

  2. Referee: [Experiments section] Experiments section: The abstract asserts 'promising and consistent performance' and 'markedly higher objective improvements' across 'many' pan-European summer-peak days, yet supplies no details on the exact number of days, the precise objective function, statistical significance testing, variance across instances, or controls against post-hoc selection of favorable scenarios. Without these, the reported gains cannot be independently verified or compared to the baselines.

    Authors: We concur that greater specificity is required for independent verification and fair comparison with the baselines. The current abstract and Experiments section use the term 'many' without enumerating the instances or providing supporting statistics. In the revision we will (i) state the exact number of pan-European summer-peak days evaluated, (ii) give a formal mathematical definition of the objective function, (iii) report mean performance together with variance (standard deviation) across all instances, (iv) include statistical significance tests (e.g., paired t-tests or Wilcoxon signed-rank tests with p-values) against the simulated-annealing and NSGA-II baselines, and (v) describe the day-selection criteria and confirm that no post-hoc filtering of favorable scenarios occurred. These additions will appear in both the abstract and the expanded Experiments section. revision: yes

Circularity Check

0 steps flagged

No significant circularity; framework and claims are self-contained against external benchmarks

full rationale

The paper describes a hierarchical MCTS framework that samples hotspots, proposes regulations via a local engine, and evaluates each via an external fast FPFS allocator to obtain rewards that guide search. Performance is assessed via direct comparison to independent flight-centric baselines (simulated annealing and NSGA-II) on pan-European traffic data, plus ablation studies on cascading. No equations or steps reduce a claimed result to a fitted parameter or self-citation by construction; FPFS is treated as an independent proxy rather than derived from the method itself. The 50% cascading reduction is reported as an experimental observation, not a definitional tautology. The derivation chain therefore remains non-circular and externally falsifiable.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Abstract-only review limits visibility into exact parameters and assumptions; the approach relies on standard MCTS exploration heuristics and the accuracy of FPFS as a proxy evaluator, which are not detailed here.

free parameters (1)
  • MCTS exploration parameters
    Typical MCTS constants for balancing exploration and exploitation are likely tuned but not specified in abstract.
axioms (1)
  • domain assumption FPFS allocator provides reliable reward estimates for regulation sequences
    Central to guiding the search; invoked implicitly when using FPFS for evaluation.

pith-pipeline@v0.9.0 · 5612 in / 1351 out tokens · 35714 ms · 2026-05-10T01:55:07.126194+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages

  1. [1]

    2024 , month =

    Whitaker, Michael , title =. 2024 , month =

    work page 2024

  2. [2]

    CBS News , year =

    Kelly, John , title =. CBS News , year =

    work page

  3. [3]

    and Cech, M

    Niarchakou, S. and Cech, M. , title =. 2018 , note =

    work page 2018

  4. [4]

    12th SESAR Innovation Days (SID) , address =

    Optimal Air Traffic Flow Management Regulations Scheme with Adaptive Large Neighbourhood Search , author =. 12th SESAR Innovation Days (SID) , address =. 2022 , note =

    work page 2022

  5. [5]

    11th SESAR Innovation Days (SID) , year =

    Dalmau, Ramon and Zerrouki, Zinedine and Anoraud, Camille and Smith, Duncan and Cramet, Benjamin , title =. 11th SESAR Innovation Days (SID) , year =

    work page

  6. [6]

    Odoni , title =

    Octavio Richetta and Amedeo R. Odoni , title =. Transportation Science , year =

    work page

  7. [7]

    Vranas and Dimitris J

    Peter B. Vranas and Dimitris J. Bertsimas and Amedeo R. Odoni , title =. Operations Research , year =

    work page

  8. [8]

    Operations Research , year =

    Dimitris Bertsimas and Sarah Stock Patterson , title =. Operations Research , year =

    work page

  9. [9]

    Transportation Science , year =

    Dimitris Bertsimas and Sarah Stock Patterson , title =. Transportation Science , year =

    work page

  10. [10]

    2024 , note =

    ATFCM Operations Manual , organization =. 2024 , note =

    work page 2024

  11. [11]

    2002 , url =

    Innovative Slot Allocation: An Overview , institution =. 2002 , url =

    work page 2002

  12. [12]

    Proceedings of the 29th Congress of the International Council of the Aeronautical Sciences (ICAS) , year =

    Alexander Lau and Robert Budde and Jan Berling and Volker Gollnick , title =. Proceedings of the 29th Congress of the International Council of the Aeronautical Sciences (ICAS) , year =

    work page

  13. [13]

    Air Traffic Flow Management Slot Allocation to Minimize Propagated Delay and Improve Airport Slot Adherence , journal =

    Nikola Ivanov and Fedja Netjasov and Radosav Jovanovi. Air Traffic Flow Management Slot Allocation to Minimize Propagated Delay and Improve Airport Slot Adherence , journal =. 2017 , volume =

    work page 2017

  14. [14]

    Transportation Research Part B: Methodological , year =

    Jun Chen and Lijian Chen and Dengfeng Sun , title =. Transportation Research Part B: Methodological , year =

    work page

  15. [15]

    CEAS Aeronautical Journal , year =

    Jan Berling and Alexander Lau and Volker Gollnick , title =. CEAS Aeronautical Journal , year =. doi:10.1007/s13272-023-00708-4 , note =

    work page doi:10.1007/s13272-023-00708-4

  16. [16]

    Transportation Research Part C: Emerging Technologies , year =

    Wang, Yanjun and Liu, Chang and Wang, Hai and Duong, Vu , title =. Transportation Research Part C: Emerging Technologies , year =

    work page

  17. [17]

    Journal of Air Transport Management , year =

    Liu, Wenjing and Zhao, Qiuhong and Delahaye, Daniel , title =. Journal of Air Transport Management , year =

    work page

  18. [18]

    , title =

    Keskin, Merve and Zografos, Konstantinos G. , title =. Transportation Research Part B: Methodological , year =

    work page

  19. [19]

    An Auction-Based Mechanism for a Privacy-Preserving Marketplace for ATFM Slots , booktitle =

    Sch. An Auction-Based Mechanism for a Privacy-Preserving Marketplace for ATFM Slots , booktitle =. 2022 , address =

    work page 2022

  20. [20]

    Multi-Objective Air Traffic Flow Management Through Lexicographic Optimisation , booktitle =

    Dalmau, Ricard and Kyne, Sin. Multi-Objective Air Traffic Flow Management Through Lexicographic Optimisation , booktitle =. 2024 , address =

    work page 2024

  21. [21]

    Aerospace , year =

    Yan, Yang and Hu, Qiang , title =. Aerospace , year =

    work page

  22. [22]

    Aerospace , year =

    Liu, Fengfan and Hu, Minghua and Zhang, Qingxian and Yang, Lei , title =. Aerospace , year =

    work page

  23. [23]

    Transactions of the Japan Society for Aeronautical and Space Sciences , year =

    Liu, Chang and Liao, Chaohao and Hang, Xu and Wang, Yanjun and Delahaye, Daniel , title =. Transactions of the Japan Society for Aeronautical and Space Sciences , year =

    work page

  24. [24]

    PLOS ONE , year =

    He, Peinan and Pan, Weijun , title =. PLOS ONE , year =

    work page

  25. [25]

    Transportation Research Part D: Transport and Environment , year =

    Feng, Huilin and Hu, Rong and Wang, Deyun and Zhang, Junfeng and Wu, Chuntao , title =. Transportation Research Part D: Transport and Environment , year =

    work page

  26. [26]

    Journal of Air Transport Management , year =

    Lee, Heeyeon and Jung, Jihyeok and Lee, Deok-Joo , title =. Journal of Air Transport Management , year =

    work page

  27. [27]

    Computers & Industrial Engineering , year =

    Tian, Jing and Hao, Xinchang and Huang, Jibo and Huang, Jinglei and Gen, Mitsuo , title =. Computers & Industrial Engineering , year =

    work page

  28. [28]

    2011 , doi =

    Pellegrini, Paola and Castelli, Lorenzo and Pesenti, Raffaele , title =. 2011 , doi =

    work page 2011

  29. [29]

    and Abdelghany, Ahmed F

    Abdelghany, Khaled F. and Abdelghany, Ahmed F. and Niznik, Tim , title =. Journal of Air Transport Management , year =

    work page

  30. [30]

    IEEE Computational Intelligence Magazine , year =

    Chaimatanan, Supatcha and Delahaye, Daniel and Mongeau, Marie , title =. IEEE Computational Intelligence Magazine , year =

    work page

  31. [31]

    Proceedings of the 2007 IEEE Intelligent Transportation Systems Conference (ITSC) , year =

    Zhang, Xuejun and Zhou, Yan and Liu, Bo and Wang, Zheng , title =. Proceedings of the 2007 IEEE Intelligent Transportation Systems Conference (ITSC) , year =. doi:10.1109/ITSC.2007.4357707 , isbn =

    work page doi:10.1109/itsc.2007.4357707 2007

  32. [32]

    International Conference on Machine Learning , pages=

    Monte-Carlo tree search as regularized policy optimization , author=. International Conference on Machine Learning , pages=. 2020 , organization=

    work page 2020

  33. [33]

    2010 , publisher=

    Modeling purposeful adaptive behavior with the principle of maximum causal entropy , author=. 2010 , publisher=

    work page 2010

  34. [34]

    2024 , month =

    Flow CONOPS: Concept of Operations , institution =. 2024 , month =

    work page 2024

  35. [35]

    Science , volume=

    A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play , author=. Science , volume=. 2018 , publisher=

    work page 2018

  36. [36]

    Proceedings of the International Conference on Automated Planning and Scheduling , volume=

    PROST: Probabilistic planning based on UCT , author=. Proceedings of the International Conference on Automated Planning and Scheduling , volume=

    work page

  37. [37]

    Scientific reports , volume=

    From Louvain to Leiden: guaranteeing well-connected communities , author=. Scientific reports , volume=. 2019 , publisher=

    work page 2019

  38. [38]

    International conference on machine learning , pages=

    Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor , author=. International conference on machine learning , pages=. 2018 , organization=

    work page 2018

  39. [39]

    Aerospace , volume=

    Selective simulated annealing for large scale airspace congestion mitigation , author=. Aerospace , volume=. 2021 , publisher=

    work page 2021

  40. [40]

    MATEC Web of Conferences , volume=

    Multi-objective optimization approach for air traffic flow management , author=. MATEC Web of Conferences , volume=

    work page

  41. [41]

    Air Transportation R&D Symposium 2026 , year=

    A Flow-Centric Approach for Network-Level ATFM Delay Optimization and Hotspot Resolution Using Hierarchical Monte Carlo Tree Search , author=. Air Transportation R&D Symposium 2026 , year=

    work page 2026

  42. [42]

    Handbook of metaheuristics , pages=

    Simulated annealing: From basics to applications , author=. Handbook of metaheuristics , pages=. 2018 , publisher=

    work page 2018

  43. [43]

    Aerospace , VOLUME =

    Lavandier, Julien and Islami, Arianit and Delahaye, Daniel and Chaimatanan, Supatcha and Abecassis, Amir , TITLE =. Aerospace , VOLUME =. 2021 , NUMBER =

    work page 2021