Pith. sign in

REVIEW 3 major objections 5 minor 167 references

Optimizing only delivery decisions, subject to preparation feasibility, beats joint optimization of both stages in dynamic order fulfillment.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-11 21:59 UTC pith:YWAHWHN5

load-bearing objection Solid OR methods paper: a genuine policy-level MDP decomposition for two-stage fulfillment, with clean finite-set theory and consistent real-data gains; the fixed-prep-cost premise is the only load-bearing soft spot. the 3 major comments →

arxiv 2607.04056 v1 pith:YWAHWHN5 submitted 2026-07-04 math.DS cs.LGmath.COmath.OC

A Policy Decomposition Framework for Dynamic Order Fulfillment Operations

Gal Neria , Michal Tzur , Marlin W. Ulmer This is my paper
classification math.DS cs.LGmath.COmath.OC MSC 90B0690C3990C4090C59
keywords dynamic order fulfillmentpolicy decompositionMarkov decision processapproximate dynamic programmingvalue function approximationlarge neighborhood searchintegrated preparation and deliverysynchronization constraints
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Many modern fulfillment systems have two tightly coupled stages—preparing an order and delivering it—under continuous random arrivals. Variable costs almost always sit in delivery (routing, delay, time-window violations), while preparation capacity is largely fixed by staffing. The paper defines the Dynamic Order Fulfillment Problem to capture this structure and shows that the right way to solve it is not to optimize both stages together. Instead, search over delivery policies and treat preparation only as a filter that accepts or rejects candidate deliveries. The resulting Decomposition-Driven Framework with Value Function Approximation (DDF-VFA) uses large-neighborhood search on partial delivery plans, a fast preparation scheduler that issues feasibility cuts, and a neural value-function estimate of future cost. On two real-data variants (warehouse picking and production with setups), the method consistently reduces cost relative to FIFO, joint myopic optimization, and joint anticipatory search. The practical claim is that, whenever preparation costs are fixed, concentrating search effort on delivery yields better real-time decisions than treating both stages symmetrically.

Core claim

When preparation costs are fixed and independent of operational choices, an optimal dynamic policy can be recovered by optimizing only over delivery-stage policies that admit at least one compatible preparation policy. The paper proves that an iterative master–subproblem loop over this reduced policy space terminates at a joint optimum (Claim 1) and shows that a practical realization of the loop (DDF-VFA) outperforms both sequential and joint benchmarks on two real-world problem variants.

What carries the argument

Policy-level decomposition of the DOFP Markov decision process: a delivery-stage master problem that proposes policies (or partial decisions completed by Trip Assignment and Scheduling) and a preparation-stage subproblem (Synchronized Preparation Scheduling) that either returns a compatible preparation decision or generates feasibility cuts that prune the master space, with a neural value-function approximation estimating the cost-to-go.

Load-bearing premise

Preparation-stage costs are fixed and do not depend on operational decisions, so the optimal joint policy is completely determined by the best delivery policy that still has a feasible preparation counterpart.

What would settle it

Construct a DOFP instance in which preparation decisions themselves carry material variable cost (or in which the Synchronized Preparation Scheduling heuristic systematically rejects feasible preparations) and check whether DDF-VFA still recovers the jointly optimal policy and still beats the joint anticipatory benchmark.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Fulfillment systems whose variable costs concentrate in delivery can safely treat preparation as a feasibility filter rather than a co-optimized stage, reducing the real-time decision space.
  • The same master–subproblem split extends, by modular add-ons, to batched preparation, multi-stage preparation, heterogeneous resources, and soft or hard time windows without redesigning the overall search.
  • When time windows or setup matrices shrink the feasible delivery space, anticipation via value-function approximation yields smaller gains; the decomposition itself remains the dominant source of improvement.
  • Component-level ablations show that objective-aligned neighborhood operators and anticipatory preparation scheduling are both necessary for the claimed gains; removing either degrades travel time or consolidation.

Where Pith is reading between the lines

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

  • If preparation capacity itself becomes decision-dependent (overtime, flexible staffing, energy-priced machines), the clean master–subproblem split would have to be replaced by a true bi-level or multi-objective search.
  • The same cost-asymmetry argument may apply to other two-stage dynamic systems outside logistics—e.g., diagnosis-then-treatment in healthcare or compute-then-transmit in edge networks—whenever the second stage carries almost all variable cost.
  • Because the preparation subproblem only needs feasibility, any faster exact or approximate scheduler (constraint programming, learned heuristics) can be swapped in without touching the delivery master or the value-function network.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper defines the Dynamic Order Fulfillment Problem (DOFP) as an MDP class that couples preparation and delivery under stochastic arrivals, with state/decision spaces partitioned by stage and linked by synchronization constraints. Exploiting the modeling premise that variable costs arise only in delivery, it proposes a policy-level decomposition into a delivery master problem and a preparation feasibility subproblem (Claim 1, finite-policy optimality). The implemented DDF-VFA solves the master via LNS over partial trip sequences completed by TAS, solves the subproblem via the SPS heuristic (or feasibility cuts), and evaluates post-decision states with a neural-network VFA. On two real-data variants (Picking from Meituan; Production from ORTEC), averaged over 100 realizations, DDF-VFA reports 3–87% cost reductions versus FIFO, Integrated, AI, and DDF-G, with component ablations and sensitivity checks on penalties and setup times.

Significance. If the fixed-preparation-cost premise holds for the intended applications, the work supplies a clean problem class, a policy-level decomposition that is formally correct under finiteness, and a practical LNS+TAS+SPS+VFA pipeline that measurably outperforms both sequential and joint non-decomposed baselines on two real datasets. The component ablations (operator removal, myopic SPS, VFA off) and multi-parameter sensitivities strengthen the empirical claim. The appendices sketch modular extensions (batching, multi-stage preparation, perishability, heterogeneous resources), which, if fully validated, would broaden impact across e-commerce, production-routing, and meal delivery.

major comments (3)
  1. §3.1 item 6 and Eq. (4)/Claim 1 rest the optimality of the master/subproblem split on preparation costs being fixed and independent of operational decisions. The paper cites supporting literature but does not quantify how often this holds, nor does it report a controlled experiment in which preparation incurs material variable cost (e.g., overtime, energy, or sequence-dependent labor). Without that check, the claimed superiority of delivery-only optimization remains conditional on an untested modeling premise that is load-bearing for the central claim.
  2. §5.2.1 and Claim 1: the theoretical guarantee assumes an exact subproblem solver that either returns a compatible preparation decision or correctly certifies non-existence. The implemented SPS is a heuristic (EDD/SPT with limited lookahead; Appendix E variants). The paper never claims online optimality of the heuristic policy, yet the abstract and §7.3 present DDF-VFA as realizing the decomposition framework. A short experiment that replaces SPS by an exact MILP (or reports the rate of false-infeasible cuts) is needed to bound the gap between the theory and the reported numbers.
  3. §7.2–7.3 and Fig. 3: the strongest empirical claim is consistent outperformance of AI (joint LNS+VFA without decomposition). Both methods share the same VFA architecture and LNS iteration budget, yet the paper does not report wall-clock times, number of feasible candidates evaluated, or a matched-compute ablation. Without that, it remains possible that the gains are partly due to a more focused search rather than the decomposition principle itself; a runtime-normalized comparison would make the claim robust.
minor comments (5)
  1. Figure 2 is referenced as missing in the source text; the workflow diagram is essential for understanding the MP–SP feedback loop and should be restored.
  2. §4 introduces the acronym DTS-OSP without definition; it appears to be a residual from an earlier draft and should be replaced by DOFP.
  3. Notation for post-decision value V(S^x_k) versus the SP value V_SP(S_k, x^{(2)}_k) is dense; a short clarifying sentence after Eq. (10) would help.
  4. Appendix G tables report additional KPIs (earliness, setup) that strengthen the Production-variant story; a one-sentence pointer in §7.3 would improve accessibility.
  5. Typos: “penanlty” (Eq. 26), “hetrogeneous” (§6.1), “standart deviation” (Appendix H).

Circularity Check

0 steps flagged

No significant circularity: policy decomposition and empirical gains are self-contained against independent benchmarks.

full rationale

The paper's derivation chain is algorithmic and empirical rather than a first-principles prediction that collapses to its inputs. The DOFP class and MDP (Sec. 3) are defined from structural observations (cost asymmetry, synchronization); the policy master/subproblem split (Eqs. 4–10, Claim 1) follows directly from the fact that variable costs appear only in the delivery stage, with a finite-set termination argument proved in-place. DDF-VFA (Sec. 5) implements that split via LNS+TAS+SPS+NN-VFA; the VFA architecture and one benchmark (AI) reuse components from the authors' prior meal-delivery work, but those components are re-applied, not re-derived as load-bearing uniqueness theorems, and the reported cost reductions (Fig. 3, Table 1, 100 realizations) are measured on held-out real-data instances against FIFO/Integrated/AI/DDF-G. No equation equates a claimed prediction to a fitted parameter by construction, no uniqueness result is imported solely via self-citation to forbid alternatives, and no known empirical pattern is merely renamed. The fixed-cost modeling premise is an explicit assumption, not a circular step. Score 0 is therefore the correct, proportionate finding.

Axiom & Free-Parameter Ledger

5 free parameters · 5 axioms · 3 invented entities

The central performance claim rests on the DOFP structural axioms (especially fixed preparation cost and non-preemption), the finiteness hypothesis of Claim 1, the heuristic completeness of SPS/TAS, and a collection of free parameters (penalty weights, setup matrices, NN architecture, LNS operators) that are either taken from prior work or chosen for the experiments. No new physical entities are postulated; the invented objects are algorithmic constructs.

free parameters (5)
  • delay/earliness penalty factors ρ, ρe, ρl
    Chosen by the authors for each variant and sensitivity experiment (ρ ∈ {0,1,10,25,50} in Table 1); they directly scale the objective and therefore the reported cost reductions.
  • sequence-dependent setup-time matrix sij
    Hand-specified 4-class matrix (Table H.7) and its variants (3-5-5-7, 5-5-5-5); controls feasibility of preparation sequences and measured violations.
  • neural-network VFA architecture and training hyperparameters
    Referenced to Neria et al. (2025) and retrained per configuration; the quality of ˆV directly affects which delivery decisions are selected.
  • LNS neighborhood operators and iteration budget
    Problem-specific operators (swap, insertion, reordering, time-window sorts) and the number of LNS iterations are design choices that determine the explored portion of XMP.
  • click-to-door promise τ and vehicle capacity κ
    Instance parameters (τ=60 min, κ=145) taken from data or set by authors; they bound the feasible region of every policy.
axioms (5)
  • domain assumption Preparation-stage operational costs are fixed (independent of the specific preparation decisions x(1)); all variable cost is incurred by delivery decisions x(2).
    Stated as defining property 6 of the DOFP class (§3.1) and used to justify optimizing only π(2) subject to existence of a compatible π(1) (Eq. 4).
  • domain assumption Decisions are non-preemptable: in-progress preparation tasks and en-route vehicles cannot be interrupted.
    Property 7 of the DOFP class (§3.1); shapes the post-decision state and the feasible action sets X(1), X(2).
  • standard math The sets of feasible stage policies Π(1) and Π(2) are finite, guaranteeing termination of the iterative master/subproblem loop at an optimal policy.
    Explicit hypothesis of Claim 1 (§4); without finiteness the iterative refinement argument does not guarantee optimality.
  • ad hoc to paper A compatible preparation decision, when one exists, can be found (or correctly declared nonexistent) by the SPS heuristic.
    SPS is a fast EDD/SPT heuristic with limited lookahead (§5.2.1); false infeasibility would incorrectly prune the master policy space.
  • domain assumption Orders arrive according to a (possibly non-stationary) Poisson process and processing/travel times are deterministic once realized.
    Used for instance generation (§7.1, App. H) and for the transition function of the MDP; stochastic processing times are left to future work.
invented entities (3)
  • Dynamic Order Fulfillment Problem (DOFP) class no independent evidence
    purpose: Unifies previously separate preparation-delivery problems under a common MDP with partitioned state/decision spaces and synchronization constraints.
    Defined in §3; no independent empirical existence outside the modeling framework of this paper.
  • DDF-VFA algorithm (policy master/subproblem loop + LNS + TAS + SPS + NN-VFA) no independent evidence
    purpose: Practical solution method that realizes the policy decomposition for real-time decisions.
    Introduced in §5; performance is measured only inside the paper’s own experimental loop.
  • Trip Assignment and Scheduling (TAS) and Synchronized Preparation Scheduling (SPS) procedures no independent evidence
    purpose: Complete partial delivery decisions and generate (or cut) compatible preparation decisions inside the master/subproblem iteration.
    Heuristic constructs of §5.1.2 and §5.2.1; correctness relative to the true feasible sets is not formally guaranteed.

pith-pipeline@v1.1.0-grok45 · 37846 in / 3735 out tokens · 37643 ms · 2026-07-11T21:59:27.408893+00:00 · methodology

0 comments
read the original abstract

Modern supply chains span diverse operational environments, ranging from e-commerce distribution networks to customized production-to-order manufacturing lines. Across these settings, operational efficiency depends on coordinating two highly interdependent stages: order preparation and downstream delivery. Although these stages are traditionally managed in isolation, real-world fulfillment systems must satisfy stringent delivery expectations under dynamic stochastic order arrivals. To bridge this gap, we introduce the Dynamic Order Fulfillment Problem (DOFP), a new problem class unifying logistical challenges previously studied separately. We model DOFP as a Markov decision process whose state and decision spaces are partitioned into preparation and delivery sub-spaces, linked by synchronization constraints. While recent approaches attempt to optimize both fulfillment stages simultaneously over myopic rolling horizons, our framework isolates and optimizes the downstream delivery policy, treating preparation strictly as a state-level constraint filter. To solve this, we develop the Decomposition-Driven Framework with Value Function Approximation (DDF-VFA), which utilizes a novel policy-level decomposition. This design partitions the search into a delivery-stage master problem and a preparation-stage compatibility subproblem, iteratively refined via feedback loops. DDF-VFA executes this strategy by combining a large-neighborhood search over partial delivery decisions with a neural-network value function approximation for the cost-to-go. Numerical illustrations on two example variants using real-world datasets show that DDF-VFA consistently outperforms benchmarks that optimize the two stages independently or jointly without decomposition. Finally, the framework naturally scales to accommodate additional real-world complexities such as batched or multi-stage preparation.

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

167 extracted references · 1 canonical work pages

  1. [1]

    2024 , url =

    Statista , title =. 2024 , url =

    2024

  2. [2]

    Transportation Science , volume=

    The restaurant meal delivery problem with ghost kitchens , author=. Transportation Science , volume=

  3. [3]

    Materials on Demand , howpublished =

  4. [4]

    Manufacturing & Service Operations Management , volume=

    On-demand delivery from stores: Dynamic dispatching and routing with random demand , author=. Manufacturing & Service Operations Management , volume=. 2023 , publisher=

    2023

  5. [5]

    Dispatch — Building Products Delivery , howpublished =

  6. [6]

    Operations Research , volume=

    An exact solution framework for multitrip vehicle-routing problems with time windows , author=. Operations Research , volume=. 2020 , publisher=

    2020

  7. [7]

    Operations Research , volume=

    Branch-price-and-cut algorithms for the vehicle routing problem with stochastic and correlated travel times , author=. Operations Research , volume=. 2021 , publisher=

    2021

  8. [8]

    Computers & Operations Research , volume=

    Off-line approximate dynamic programming for the vehicle routing problem with a highly variable customer basis and stochastic demands , author=. Computers & Operations Research , volume=. 2023 , publisher=

    2023

  9. [9]

    European Journal of Operational Research , volume=

    Deep Q-learning for same-day delivery with vehicles and drones , author=. European Journal of Operational Research , volume=. 2022 , publisher=

    2022

  10. [10]

    European Journal of Operational Research , volume=

    Benders decomposition for the hazmat transport network design problem , author=. European Journal of Operational Research , volume=. 2018 , publisher=

    2018

  11. [11]

    Operations Research , volume=

    Decomposition branching for mixed integer programming , author=. Operations Research , volume=. 2022 , publisher=

    2022

  12. [12]

    Transportation Research Part B: Methodological , volume=

    The value of stochastic crowd resources and strategic location of mini-depots for last-mile delivery: A Benders decomposition approach , author=. Transportation Research Part B: Methodological , volume=. 2022 , publisher=

    2022

  13. [13]

    Operations Research , volume =

    Ignacio Erazo and Alejandro Toriello , title =. Operations Research , volume =

  14. [14]

    Operations Research , volume=

    A convex reformulation and an outer approximation for a large class of binary quadratic programs , author=. Operations Research , volume=. 2023 , publisher=

    2023

  15. [15]

    Errico, Fausto and Crainic, Teodor Gabriel and Malucelli, Federico and Nonato, Maddalena , journal=. A. 2017 , publisher=

    2017

  16. [16]

    Omega , volume=

    Increasing schedule reliability in the multiple depot vehicle scheduling problem with stochastic travel time , author=. Omega , volume=. 2024 , publisher=

    2024

  17. [17]

    Operations Research , volume=

    Exact algorithms for electric vehicle-routing problems with time windows , author=. Operations Research , volume=. 2016 , publisher=

    2016

  18. [18]

    Operations Research , volume=

    The humanitarian pickup and distribution problem , author=. Operations Research , volume=. 2019 , publisher=

    2019

  19. [19]

    Operations Research , volume=

    Vessel service planning in seaports , author=. Operations Research , volume=. 2022 , publisher=

    2022

  20. [20]

    Exact Two-Step

    Celik, Sifa and Martin, Layla and Schrotenboer, Albert H and van Woensel, Tom , year=. Exact Two-Step

  21. [21]

    A converging

    van der Laan, Niels and Romeijnders, Ward , journal=. A converging. 2024 , publisher=

    2024

  22. [22]

    Rahmaniani, Ragheb and Ahmed, Shabbir and Crainic, Teodor Gabriel and Gendreau, Michel and Rei, Walter , journal=. The. 2020 , publisher=

    2020

  23. [23]

    Operations Research , volume=

    Branch-and-price for prescriptive contagion analytics , author=. Operations Research , volume=. 2025 , publisher=

    2025

  24. [24]

    Logistics Research , volume=

    Delivery deadlines in same-day delivery , author=. Logistics Research , volume=. 2017 , publisher=

    2017

  25. [25]

    Data-Centric Engineering , volume=

    Model and algorithm for pharmaceutical distribution routing problem considering customer priority and carbon emissions , author=. Data-Centric Engineering , volume=. 2024 , publisher=

    2024

  26. [26]

    Transportation Science , year=

    Enhancing Online Food Delivery with Transfer Points: A Data-Driven Decompose-Then-Optimize Approach , author=. Transportation Science , year=

  27. [27]

    Computers & Operations Research , volume=

    Home health care routing and scheduling: A review , author=. Computers & Operations Research , volume=. 2017 , publisher=

    2017

  28. [28]

    European Journal of Operational Research , volume=

    Combined vehicle routing and scheduling with temporal precedence and synchronization constraints , author=. European Journal of Operational Research , volume=. 2008 , publisher=

    2008

  29. [29]

    Operations Research , volume=

    A branch-and-repair method for three-dimensional bin selection and packing in e-commerce , author=. Operations Research , volume=. 2023 , publisher=

    2023

  30. [30]

    Operations Research , volume=

    A stochastic integer programming approach to air traffic scheduling and operations , author=. Operations Research , volume=. 2020 , publisher=

    2020

  31. [31]

    Production and Operations Management , volume=

    Coordinated charging station search in stochastic environments: A multiagent approach , author=. Production and Operations Management , volume=. 2023 , publisher=

    2023

  32. [32]

    Transportation Science , volume=

    The electric vehicle routing problem with capacitated charging stations , author=. Transportation Science , volume=. 2022 , publisher=

    2022

  33. [33]

    Transportation Science , volume=

    Cooperative learning for smart charging of shared autonomous vehicle fleets , author=. Transportation Science , volume=. 2023 , publisher=

    2023

  34. [34]

    Transportation Science , volume=

    Electric vehicle routing with public charging stations , author=. Transportation Science , volume=. 2021 , publisher=

    2021

  35. [35]

    Transportation Science , volume=

    Robust charging network planning for metropolitan taxi fleets , author=. Transportation Science , volume=. 2024 , publisher=

    2024

  36. [36]

    2011 , publisher=

    Introduction to Stochastic Programming , author=. 2011 , publisher=

    2011

  37. [37]

    European Journal of Operational Research , volume=

    Long-run optimal pricing in electricity markets with non-convex costs , author=. European Journal of Operational Research , volume=. 2023 , publisher=

    2023

  38. [38]

    Computation of convex hull prices in electricity markets with non-convexities using

    Andrianesis, Panagiotis and Bertsimas, Dimitris and Caramanis, Michael C and Hogan, William W , journal=. Computation of convex hull prices in electricity markets with non-convexities using. 2021 , publisher=

    2021

  39. [39]

    European Journal of Operational Research , volume=

    Benders decomposition without separability: A computational study for capacitated facility location problems , author=. European Journal of Operational Research , volume=. 2016 , publisher=

    2016

  40. [40]

    European Journal of Operational Research , volume=

    A survey on bilevel optimization under uncertainty , author=. European Journal of Operational Research , volume=. 2023 , publisher=

    2023

  41. [41]

    Perspectives on Using

    Hewitt, Mike and Rei, Walter , booktitle=. Perspectives on Using. 2024 , publisher=

    2024

  42. [42]

    Annals of Operations Research , volume=

    An overview of bilevel optimization , author=. Annals of Operations Research , volume=. 2007 , publisher=

    2007

  43. [43]

    Rahmaniani, Ragheb and Crainic, Teodor Gabriel and Gendreau, Michel and Rei, Walter , journal=. The. 2017 , publisher=

    2017

  44. [44]

    2023 , url =

    Statista , title =. 2023 , url =

    2023

  45. [45]

    Management Science , volume=

    Courier dispatch in on-demand delivery , author=. Management Science , volume=. 2024 , publisher=

    2024

  46. [46]

    European Journal of Operational Research , volume=

    Recent advances in integrating demand management and vehicle routing: A methodological review , author=. European Journal of Operational Research , volume=. 2023 , publisher=

    2023

  47. [47]

    2024 , note=

    Meituan , title=. 2024 , note=

    2024

  48. [48]

    Journal of Cleaner Production , volume=

    A green multi-objective integrated scheduling of production and distribution with heterogeneous fleet vehicle routing and time windows , author=. Journal of Cleaner Production , volume=. 2020 , publisher=

    2020

  49. [49]

    Materials On Demand Website , howpublished =

  50. [50]

    2024 , howpublished =

    BPDS , title =. 2024 , howpublished =

    2024

  51. [51]

    Dantzig, G. B. and Wolfe, P. , title =. Operations Research , volume =

  52. [52]

    Benders, J. F. , title =. Numerische Mathematik , volume =

  53. [53]

    2014 , publisher=

    Integer and combinatorial optimization , author=. 2014 , publisher=

    2014

  54. [54]

    Artificial Intelligence , volume=

    Between MDPs and semi-MDPs: A framework for temporal abstraction in reinforcement learning , author=. Artificial Intelligence , volume=. 1999 , publisher=

    1999

  55. [55]

    arXiv preprint arXiv:1703.01161 , year=

    Feudal networks for hierarchical reinforcement learning , author=. arXiv preprint arXiv:1703.01161 , year=

    Pith/arXiv arXiv

  56. [56]

    Advances in Neural Information Processing Systems , volume=

    Data-efficient hierarchical reinforcement learning , author=. Advances in Neural Information Processing Systems , volume=

  57. [57]

    Advances in Neural Information Processing Systems , volume=

    Multi-agent actor-critic for mixed cooperative-competitive environments , author=. Advances in Neural Information Processing Systems , volume=

  58. [58]

    2009 , publisher=

    Singh, Kavinesh J and Philpott, Andy B and Wood, R Kevin , journal=. 2009 , publisher=

    2009

  59. [59]

    A framework for generalized

    Bolusani, Suresh and Ralphs, Ted K , journal=. A framework for generalized. 2022 , publisher=

    2022

  60. [60]

    Handbook of reinforcement learning and control , pages=

    Multi-agent reinforcement learning: A selective overview of theories and algorithms , author=. Handbook of reinforcement learning and control , pages=. 2021 , publisher=

    2021

  61. [61]

    2021 , publisher=

    Lectures on stochastic programming: modeling and theory , author=. 2021 , publisher=

    2021

  62. [62]

    Advances in Neural Information Processing Systems , pages=

    Parr, Ronald and Russell, Stuart J , title=. Advances in Neural Information Processing Systems , pages=

  63. [63]

    Mathematics of Operations Research , volume=

    The complexity of decentralized control of Markov decision processes , author=. Mathematics of Operations Research , volume=

  64. [64]

    Artificial Intelligence , volume=

    Collaborative multiagent decision making using influence diagrams , author=. Artificial Intelligence , volume=. 2005 , publisher=

    2005

  65. [65]

    Discrete Event Dynamic Systems , volume=

    Recent advances in hierarchical reinforcement learning , author=. Discrete Event Dynamic Systems , volume=. 2003 , publisher=

    2003

  66. [66]

    Geoffrion, A. M. , title =. Journal of Optimization Theory and Applications , volume =

  67. [67]

    2017 , publisher=

    A method and IT-framework for on-demand delivery in make-to-order construction supply chains , author=. 2017 , publisher=

    2017

  68. [68]

    European Journal of Operational Research , year=

    Due date-oriented picker routing, an efficient exact solution algorithm, and its application to pick-from-store omnichannel retailing , author=. European Journal of Operational Research , year=

  69. [69]

    Management Science , volume=

    Integrated scheduling of production and distribution operations , author=. Management Science , volume=. 2005 , publisher=

    2005

  70. [70]

    Swarm and Evolutionary Computation , volume=

    Modelling and scheduling integration of distributed production and distribution problems via black widow optimization , author=. Swarm and Evolutionary Computation , volume=. 2022 , publisher=

    2022

  71. [71]

    European Journal of Operational Research , volume=

    Demand management for attended home delivery—A literature review , author=. European Journal of Operational Research , volume=. 2023 , publisher=

    2023

  72. [72]

    Transportation Science , volume=

    The Dynamic Pickup and Allocation with Fairness Problem , author=. Transportation Science , volume=. 2024 , publisher=

    2024

  73. [73]

    European Journal of Operational Research , volume=

    Stochastic modeling of integrated order fulfillment processes with delivery time promise: Order picking, batching, and last-mile delivery , author=. European Journal of Operational Research , volume=. 2024 , publisher=

    2024

  74. [74]

    European Journal of Operational Research , volume=

    Minimizing the number of tardy jobs for m parallel machines , author=. European Journal of Operational Research , volume=. 1995 , publisher=

    1995

  75. [75]

    Environment, Development and Sustainability , pages=

    A new coordinating model for green supply chain and batch delivery scheduling with satisfaction customers , author=. Environment, Development and Sustainability , pages=. 2022 , publisher=

    2022

  76. [76]

    Applied Soft Computing , volume=

    Variable neighborhood-based cuckoo search for production routing with time window and setup times , author=. Applied Soft Computing , volume=. 2022 , publisher=

    2022

  77. [77]

    European Journal of Operational Research , volume=

    A review on integrated scheduling and outbound vehicle routing problems , author=. European Journal of Operational Research , volume=. 2023 , publisher=

    2023

  78. [78]

    Integration of order picking and vehicle routing in a

    Moons, Stef and Ramaekers, Katrien and Caris, An and Arda, Yasemin , journal=. Integration of order picking and vehicle routing in a. 2018 , publisher=

    2018

  79. [79]

    International Journal of Production Research , volume=

    An adaptive large neighbourhood search for the operational integrated production and distribution problem of perishable products , author=. International Journal of Production Research , volume=. 2015 , publisher=

    2015

  80. [80]

    Computers & Industrial Engineering , volume=

    Hybrid metaheuristics for the integrated and detailed scheduling of production and delivery operations in no-wait flow shop systems , author=. Computers & Industrial Engineering , volume=. 2022 , publisher=

    2022

Showing first 80 references.