Pith. sign in

REVIEW 3 major objections 6 minor 32 references

End-to-End Supply Chain Planning in the Paper Industry Via Column Generation and Benders Decomposition

T0 review · 3 major / 6 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read Aggregate supply profiles make integrated paper supply chain planning tractable at industrial scale, cutting eight-week total cost by 24.4% and returning implementable plans in hours.

desk verdict A sound and genuinely useful decomposition architecture for integrated paper-mill planning, with a real structural insight in the aggregate-supply interface; the empirical headline claims are plausible but rest on proprietary data and an untested integrality assumption. read the letter →

arxiv 2607.16618 v1 pith:5PAXE63R submitted 2026-07-18 math.OC

Changkun Guan , Amira Hijazi , Pascal Van Hentenryck This is my paper · ORCID
classification math.OC MSC 90C1190C3990B0690B30
keywords supplychainplanningpaperindustrycolumngenerationBendersdecompositioncuttingstockvehicleloadingmake-to-stockrollinghorizon
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

This paper attempts to establish that the paper industry's end-to-end planning problem—production scheduling, trimming, vehicle loading, and multi-period fulfillment of make-to-order and make-to-stock demand—can be solved exactly at industrial scale if the downstream fulfillment problem is driven only by an aggregate supply profile. That profile, indexed by product, customer, and period, is the paper's central structural interface. The proposed method, BDCG-DP, couples column generation with exact dynamic-programming pricing on the supply side and Benders decomposition on the demand side, and reports that on eight-week instances from a major manufacturer it lowers total cost by 24.4% relative to a standard column-generation baseline and closes the root-LP gap from about 28% to 4%. If true, this makes integrated rolling-horizon planning practical where sequential planning was the norm, and it identifies the supply-profile decomposition as the reason the integrated problem becomes tractable.

What carries the argument

The aggregate supply profile S_{p,c,t}—the quantity of product p shipped to customer c in period t, aggregated over all production runs and vehicle types—is the paper's central interface. Once fixed, the downstream MTS problem separates by product-customer pair (Proposition 1), enabling Benders optimality cuts that are linear in the supply profile. Lemma 1 supplies the integrality bridge: on integral supply profiles, each downstream chain LP solves the original discrete inventory problem exactly, because the incremental formulation has a consecutive-ones, totally unimodular constraint matrix with integral breakpoints. The supply-side pricing reduces to exact integer knapsacks (one-dimensiona

What would settle it

Take a single product-customer chain with fractional safety-stock penalty slopes (or any non-integral breakpoint), fix an integral supply profile, and compute the chain's LP value and integer value: if they differ, Lemma 1 fails and Phase 2 no longer evaluates downstream cost exactly. On the proprietary T8 instances, one could round all penalty breakpoints to integers and rerun the full method; if the root-LP bound or final gap changes materially, the reported 24.4% improvement depends on the integrality convention.

Watch

Extended reading notes

Core claim

The central claim is that downstream make-to-stock fulfillment depends on upstream production, trimming, and loading decisions only through the aggregate supply profile—the quantity of each product delivered to each customer in each period. Once that profile is fixed, the downstream problem decomposes exactly by product-customer chain, and each chain is a small multi-period inventory problem. The paper proves this decomposition (Proposition 1), and shows that when the supply profile is integral, the LP relaxation of each chain coincides with the original integer problem because the incremental formulation has a totally unimodular matrix (Lemma 1). This yields a two-phase procedure: Phase 1 p

Load-bearing premise

The load-bearing premise is that all supply profiles, demands, safety-stock targets, and penalty breakpoints are integral, so that the LP relaxation of each downstream inventory chain provably equals its integer optimum; the proprietary instances cannot be audited, so the 4.12% gap is a certificate only under that data convention.

Editorial extensions

If this is right

  • If the central claim is right, integrated planning across production, trim, loading, and fulfillment becomes tractable at industrial scale over eight-week horizons, eliminating the need for the sequential planning that is still common in practice.
  • The aggregate supply profile becomes a reusable modeling interface: any production-distribution setting where downstream service constraints depend only on cumulative per-customer supply over time can adopt the same decomposition.
  • Rolling-horizon planners can run what-if scenarios within hours instead of days, since early feasible incumbents with root-LP gaps below 10% are available well before the time budget expires.
  • The reported 24.4% cost reduction comes almost entirely from lowering make-to-stock penalty costs, indicating that better alignment of production timing with inventory targets—not lower production or transportation spend—drives the gain.
  • Exact dynamic-programming pricing yields most of the short-horizon speedup (e.g., 16x at one week), while the Benders decomposition is the main contributor to the long-horizon quality improvement on eight-week instances.

Reading between the lines

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

  • Beyond the paper, the same supply-profile decomposition likely transfers to other multi-plant, multi-period production-distribution systems—e.g., steel, consumer goods, or building materials—where downstream fulfillment only sees aggregate supply per customer and period.
  • A testable extension would be to replace the single-deckle trimming knapsack with two-stage or two-dimensional cutting, and to check whether the Benders cuts still yield tight root-LP bounds; the paper's pricing DP would need to handle the extra combinatorial state.
  • The integrality bridge (Lemma 1) depends on a data convention that penalty breakpoints, demands, and safety-stock targets are integral. If any of these are fractional in a real instance, the Phase 2 evaluation of the downstream chain is no longer exact, and the reported 4.12% gap would not be a certificate for the true integer chain cost.
  • A natural stress test is to run the same two-phase framework on a small public instance with fractional penalty slopes and compare the certified root-LP bound and final gap against a direct MIP solve; this would delineate how much of the reported performance rests on the integrality assumption.
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 / 6 minor

Summary. The paper studies an integrated paper-manufacturing planning problem combining production scheduling, trimming, vehicle loading, and multi-period make-to-stock/make-to-order fulfillment. It proposes an exact pattern-based formulation exploiting the aggregate supply profile as the interface between upstream supply decisions and downstream fulfillment, and a two-phase hybrid algorithm (BDCG-DP) that combines column generation with exact dynamic-programming pricing for trimming/loading subproblems and Benders decomposition for MTS inventory chains. On 58 proprietary instances from a North American paper manufacturer, the method is reported to reduce median runtimes by factors of 8–16 on short horizons and, on six eight-week instances, to lower the median final root-LP gap from 28.43% (CG-DP) to 4.12% and reduce the median final objective by 24.4%, with integer-feasible plans available within 2.0–5.7 hours.

Significance. The aggregate-supply-profile decomposition is a clean structural idea with potential applicability beyond paper manufacturing. The mathematical development—Proposition 1 (supply-profile decomposition), Lemma 2 (productwise reduction of trimming pricing), Proposition 2 (LP optimality of Phase 1), Proposition 3 (finite termination), and the exactness argument in EC.3.4—uses standard TU and Benders machinery and is internally coherent. The DP-based exact pricing and the supporting-hyperplane Benders cuts are standard and correctly derived. If the empirical claims are reproducible, the paper constitutes a substantial practical advance: it shows that integrated rolling-horizon planning with industrial-scale combinatorial detail is computationally feasible. The main caveats are the unverified integrality precondition for Lemma 1, a citation mismatch regarding the root-LP certificate, and the confidentiality of the benchmark instances, which together limit independent verification of the headline numbers.

major comments (3)
  1. [Section 4.1.4 and EC.3.5 (Lemma 1)] The exactness of the Phase-2 chain evaluation and the validity of the root-LP lower bound rest on the integrality of initial inventory, demands, safety-stock targets, and penalty breakpoints. The paper asserts this as a data convention, but no verification is presented for the 58 confidential instances. If any instance contains a fractional demand, safety-stock target, or penalty breakpoint, the TU argument fails; the LP chain value can understate the true integer chain cost, and the Benders cuts built from these LP values would not certify the actual integer objective. Consequently the reported 4.12% median gap and 24.4% cost reduction are not established as bounds for the true integer problem. Please verify integrality (or describe an exact scaling to integer data) and provide sensitivity evidence.
  2. [Section 6.2 and Appendix EC.2] Section 6.2 cites 'Appendix Table EC.2' as the certificate of reduced-cost optimality of the common root-LP bound, but EC.2 reports only Phase-1 wall-clock times and contains no certificate—no final root-LP values, no maximum reduced costs, no convergence flag. Since GapLP is measured against this root LP, the missing certificate undermines the reported gap metric. Provide the certificate per instance or an explicit statement of how reduced-cost optimality was verified and where the data are archived.
  3. [Data Ethics & Reproducibility Note] The empirical contribution rests entirely on 58 confidential instances; no code or data are provided. Given that the integrality precondition in Lemma 1 is unverifiable on these instances, I request a reproducibility supplement: a synthetic instance generator calibrated to the reported dimensions, per-instance data-integrity summaries (e.g., the maximum denominator of all inventory, demand, and safety-stock values), or an independent certification from the industrial partner. Without this, the falsifiability of the headline computational claims is limited.
minor comments (6)
  1. [Abstract and Section 1.1] The 24.4% cost reduction is a median across six T8 instances; the abstract should state 'median' rather than implying a uniform reduction on every instance.
  2. [Section 6.2, GapLP definition] The formula GapLP uses max{1,|Z_LP_root|}; clarify the units and the reason for the max (presumably to avoid division by zero).
  3. [Table 4] The T8 row reports 24.7% 'Obj. reduction vs CG-MIP' while the text quotes 24.4% 'vs CG-DP'; the two numbers should be reconciled and labeled consistently.
  4. [EC.3.4 and Section 4.2] The exactness proof relies on the auxiliary lane-level cap (EC.3.1), which is omitted from the implemented master. Since this is a technical finiteness device, its omission and the effect of positive vehicle costs should be mentioned when the pattern-based model is introduced in the main text.
  5. [Section 5.5.1, Algorithm 1] The tolerance epsilon_viol is not specified in the computational study; please state its value and any sensitivity of the results to it.
  6. [Figure 6] The figure shows feasible-objective progress curves; clarify whether the curves continue after the first feasible BDCG-DP solution and how the end-of-budget point is defined.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the headline cost reduction is an optimizer output on fixed cost data, and no load-bearing step reduces to a fit or to a self-citation.

full rationale

The paper's central claims are computational: BDCG-DP is compared against CG-DP on the same proprietary instances with a common initial pattern pool (EC.1.1), a common 8-hour budget, and identical cost inputs. The 24.4% cost reduction and 4.12% root-LP gap are outputs of the optimizer, not parameters fitted to produce those outcomes. Phase 1 Benders cuts are valid supporting hyperplanes of the downstream LP value function (Proposition 1, Eq. 5.3), and Phase 2 evaluates chain LPs at integral supply profiles; no equation is shown to be equivalent to its own input by construction. There are no self-citations to prior work by the same authors, so no self-citation chain is load-bearing. The closest potential concern is Section 6.5, where the MTS-penalty reduction is attributed to demand-side Benders cuts; that is an ablation explanation of an algorithmic mechanism, not a case of the output being assumed in the input. The stated integrality convention (Section 4.1.4 and Lemma 1, EC.3.5) is an unverified data assumption, and Section 6.2's reference to a reduced-cost certificate does not match Table EC.2, which reports phase times; these are reproducibility/verification gaps, not circular reductions. Under the hard rule that circularity requires a quoted reduction of a claim to its own inputs, none is present.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The contribution is algorithmic, not a fitted scientific law. There are no invented physical entities; the 'aggregate supply profile' S is an interface variable, not a new conserved quantity or particle. The derivation relies on standard LP duality, Benders decomposition, column-generation optimality, DP exactness for integer knapsacks, and total unimodularity, plus domain assumptions about paper-mill operations. The benchmark hyperparameters listed above are protocol choices that shape the reported numbers but are not fitted to a target result.

free parameters (4)
  • Terminal MIP gap target = 2% relative gap
    Benchmark protocol setting for all methods; final objectives in Table 4 are measured when this target is hit or the 8-hour budget expires. Not fitted to data, but it shapes the headline numbers.
  • Wall-clock budgets = 8h total, 4h Phase 1
    Time limits define the T4/T8 comparisons and the 'within hours' claims; changing them could change which method is declared the winner.
  • Initial trimming-pattern budget = 30,000 patterns
    Shared starting pool cap described in EC.1.1; affects solution quality and the fairness of cross-method comparison.
  • Phase-2 cut tolerance = epsilon_viol (not specified numerically)
    Controls when the Benders refinement loop stops in Algorithm 1; leaving it unspecified makes exact reproduction impossible.
assumptions (6)
  • standard math LP duality and supporting hyperplanes characterize the convex MTS chain value function; Benders optimality cuts are globally valid.
    Used in Sections 5.2 and 5.3 for cut construction; standard for convex piecewise-linear subproblems.
  • standard math The consecutive-ones incremental formulation is totally unimodular, so LP and IP chain costs coincide on integral supply profiles.
    Lemma 1 and EC.3.5; load-bearing for the exactness of Phase 2 cost evaluation.
  • standard math Exact dynamic programming solves the one- and two-dimensional integer knapsack pricing subproblems.
    Section 5.4.3; assumes bounded integral capacity coefficients.
  • domain assumption Downstream MTS fulfillment depends on supply only through aggregate S_{p,c,t} and decomposes by (p,c); no downstream constraints couple chains.
    Section 5.1, Proposition 1; if cross-customer constraints existed, the Benders chain decomposition would fail.
  • domain assumption All jumbo rolls in a run are processed; no jumbo-roll inventory; one vehicle serves one run and one destination; MTO orders cannot backlog.
    Section 3; these industrial-practice assumptions are baked into the model and are not relaxed or validated against alternative operating rules.
  • domain assumption All demand, inventory, safety-stock breakpoints, and capacity coefficients are integral, so the DP and total-unimodularity arguments apply.
    Sections 4 and EC.3.5; stated as a data convention, but unverifiable on the proprietary instances.

how reviews work

0 comments
Cite this review

Pith. "Pith review of End-to-End Supply Chain Planning in the Paper Industry Via Column Generation and Benders Decomposition." pith.science (2026). https://pith.science/paper/5PAXE63R

@misc{pith2026260716618,
  author       = {Pith},
  title        = {Pith review of: End-to-End Supply Chain Planning in the Paper Industry Via Column Generation and Benders Decomposition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5PAXE63R}},
  note         = {Machine review of arXiv:2607.16618}
}
read the original abstract

Problem definition: The paper studies an integrated end-to-end planning problem in large-scale paper manufacturing, where production scheduling, trimming decisions, vehicle loading, and multi-period fulfillment of make-to-order and make-to-stock demand must be coordinated over time. In practice, these decisions are often optimized sequentially, leading to material waste, inefficient transportation, and degraded service levels. Solving the fully integrated problem at industrial scale remains computationally challenging due to its combinatorial structure. Methodology/results: A key structural feature of the problem is that downstream fulfillment decisions depend on upstream production and logistics choices only through aggregate supply availability over time. By exploiting this structure, the paper develops an exact mathematical formulation and proposes a two-phase hybrid framework (BDCG-DP) that integrates column generation (CG) using exact dynamic-programming (DP) for supply-side decisions with Benders decomposition (BD) for downstream fulfillment. Computational experiments on proprietary instances from a major North American paper manufacturer show that BDCG-DP lowers total costs by 24.4% compared to a traditional CG-DP on challenging eight-week planning problems. Median runtime for four-week planning problems decreases from over five hours using CG-DP to under one hour using BDCG-DP. Managerial implications: This paper provides the first exact model that integrates production, trimming, load planning, and multi-period fulfillment at an industrial scale. The proposed approach returns integer-feasible plans within 2.3 to 6 hours for the most complex planning problems, enabling planners to access high-quality implementable schedules within hours, a capability that was previously unavailable in practice.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 16 canonical work pages

  1. [1]

    Operations Research Letters , volume=

    On the two-dimensional knapsack problem , author=. Operations Research Letters , volume=. 2004 , publisher=

    2004

  2. [2]

    Leao and Marcos M

    Aline A.S. Leao and Marcos M. Furlan and Franklina M.B. Toledo , keywords =. Decomposition methods for the lot-sizing and cutting-stock problems in paper industries , journal =. 2017 , issn =. doi:https://doi.org/10.1016/j.apm.2017.04.010 , url =

    work page doi:10.1016/j.apm.2017.04.010 2017

  3. [3]

    An analysis of the integrated lot-sizing and cutting-stock problem formulation , journal =

    Livia Maria Pierini and Kelly Cristina Poldi , keywords =. An analysis of the integrated lot-sizing and cutting-stock problem formulation , journal =. 2021 , issn =. doi:https://doi.org/10.1016/j.apm.2021.06.009 , url =

    work page doi:10.1016/j.apm.2021.06.009 2021

  4. [4]

    A coupling cutting stock-lot sizing problem in the paper industry , url =

    Poltroniere, S. A coupling cutting stock-lot sizing problem in the paper industry , url =. Annals of Operations Research , number =. 2008 , bdsk-url-1 =. doi:10.1007/s10479-007-0200-6 , id =

    work page doi:10.1007/s10479-007-0200-6 2008

  5. [5]

    Matheuristic for the lot-sizing and scheduling problem in integrated pulp and paper production , journal =

    Marcos Furlan and Bernardo Almada-Lobo and Maristela Santos and Reinaldo Morabito , keywords =. Matheuristic for the lot-sizing and scheduling problem in integrated pulp and paper production , journal =. 2024 , issn =. doi:https://doi.org/10.1016/j.cie.2024.110183 , url =

    arXiv 2024

  6. [6]

    Enforcing minimum run length in the cutting stock problem , journal =

    Adolf Diegel and Garth Miller and Edouard Montocchio and Sias. Enforcing minimum run length in the cutting stock problem , journal =. 2006 , issn =. doi:https://doi.org/10.1016/j.ejor.2004.09.039 , url =

    work page doi:10.1016/j.ejor.2004.09.039 2006

  7. [7]

    Progressive Selection Method for the Coupled Lot-Sizing and Cutting-Stock Problem , journal =

    Wu, Tao and Akartunal. Progressive Selection Method for the Coupled Lot-Sizing and Cutting-Stock Problem , journal =. 2017 , doi =. https://doi.org/10.1287/ijoc.2017.0746 , abstract =

    arXiv 2017

  8. [8]

    Gilmore, P. C. and Gomory, R. E. , title =. Operations Research , volume =. 1961 , doi =. https://doi.org/10.1287/opre.9.6.849 , abstract =

    doi:10.1287/opre.9.6.849 1961

Show all 32 references
  1. [9]

    Gramani and Paulo M

    Maria Cristina N. Gramani and Paulo M. França , keywords =. The combined cutting stock and lot-sizing problem in industrial processes , journal =. 2006 , issn =. doi:https://doi.org/10.1016/j.ejor.2004.12.019 , url =

    2006 doi

  2. [10]

    International Journal of Mathematics in Operational Research , volume =

    Witthayapraphakorn, Aphisak and Jaijit, Sasarose and Charnsethikul, Peerayuth , title =. International Journal of Mathematics in Operational Research , volume =. 2021 , doi =. https://www.inderscienceonline.com/doi/pdf/10.1504/IJMOR.2021.113578 , abstract =

    2021

  3. [11]

    Kantorovich, L. V. , title =. Management Science , volume =. 1960 , doi =. https://doi.org/10.1287/mnsc.6.4.366 , abstract =

    1960 doi

  4. [12]

    Beraldi and M.E

    P. Beraldi and M.E. Bruni and D. Conforti , keywords =. The stochastic trim-loss problem , journal =. 2009 , issn =. doi:https://doi.org/10.1016/j.ejor.2008.04.042 , url =

    2009 doi

  5. [13]

    The one-dimensional cutting stock problem with usable leftovers – A survey , journal =

    Adriana Cristina Cherri and Marcos Nereu Arenales and Horacio Hideki Yanasse and Kelly Cristina Poldi and Andréa Carla. The one-dimensional cutting stock problem with usable leftovers – A survey , journal =. 2014 , issn =. doi:https://doi.org/10.1016/j.ejor.2013.11.026 , url =

    2014 doi

  6. [14]

    Cutting stock problem with usable leftovers: A review , journal =

    Victor Senergues and Nadjib Brahimi and Adriana Cristina Cherri and François Klein and Olivier Péton , keywords =. Cutting stock problem with usable leftovers: A review , journal =. 2026 , issn =. doi:https://doi.org/10.1016/j.ejor.2025.03.014 , url =

    2026 doi

  7. [15]

    Vossen , keywords =

    Harald Reinertsen and Thomas W.M. Vossen , keywords =. The one-dimensional cutting stock problem with due dates , journal =. 2010 , issn =. doi:https://doi.org/10.1016/j.ejor.2009.03.042 , url =

    2010 doi

  8. [16]

    Belov and G

    G. Belov and G. Scheithauer , keywords =. A cutting plane algorithm for the one-dimensional cutting stock problem with multiple stock lengths , journal =. 2002 , issn =. doi:https://doi.org/10.1016/S0377-2217(02)00125-X , url =

    2002 doi

  9. [17]

    Belov and G

    G. Belov and G. Scheithauer , keywords =. A branch-and-cut-and-price algorithm for one-dimensional stock cutting and two-dimensional two-stage cutting , journal =. 2006 , issn =. doi:https://doi.org/10.1016/j.ejor.2004.08.036 , url =

    2006 doi

  10. [18]

    , title =

    Haessler, Robert W. , title =. Management Science , volume =. 1971 , doi =. https://doi.org/10.1287/mnsc.17.12.B793 , abstract =

    1971 doi

  11. [19]

    Inventory Management with Advance Booking Information: The Case of Surgical Supplies and Elective Surgeries , journal =

    Chan, Jacky and G\". Inventory Management with Advance Booking Information: The Case of Surgical Supplies and Elective Surgeries , journal =. 2024 , doi =. https://doi.org/10.1287/msom.2021.0063 , abstract =

    2024

  12. [20]

    Manufacturing & Service Operations Management , volume =

    Mandl, Christian and Minner, Stefan , title =. Manufacturing & Service Operations Management , volume =. 2023 , doi =. https://doi.org/10.1287/msom.2020.0890 , abstract =

    2023

  13. [21]

    Manufacturing & Service Operations Management , volume =

    DeValve, Levi and Wei, Yehua and Wu, Di and Yuan, Rong , title =. Manufacturing & Service Operations Management , volume =. 2023 , doi =. https://doi.org/10.1287/msom.2021.0981 , abstract =

    2023

  14. [22]

    The Impact of Committing to Customer Orders in Online Retail , journal =

    Figueira, Gon. The Impact of Committing to Customer Orders in Online Retail , journal =. 2023 , doi =. https://doi.org/10.1287/msom.2022.1124 , abstract =

    2023

  15. [23]

    Manufacturing & Service Operations Management , volume =

    Basciftci, Beste and Ahmed, Shabbir and Gebraeel, Nagi , title =. Manufacturing & Service Operations Management , volume =. 2024 , doi =. https://doi.org/10.1287/msom.2023.0157 , abstract =

    2024

  16. [24]

    Manufacturing & Service Operations Management , volume =

    Yu, Yi and Deng, Tianhu and Song, Jing-Sheng , title =. Manufacturing & Service Operations Management , volume =. 2024 , doi =. https://doi.org/10.1287/msom.2023.0520 , abstract =

    2024

  17. [25]

    Annals of Operations Research , volume=

    A coupling cutting stock-lot sizing problem in the paper industry , author=. Annals of Operations Research , volume=. 2008 , publisher=

    2008

  18. [26]

    Journal of Mathematics and Statistics , publisher =

    Optimization Based Heuristic Approaches for Solving an Integrated One-dimensional Cutting Stock-Transportation Problem , author =. Journal of Mathematics and Statistics , publisher =. 2007 , month =. doi:10.3844/jmssp.2007.142.150 , url =

    2007 doi

  19. [27]

    Operations Research , volume =

    Keskinocak, Pinar and Wu, Frederick and Goodwin, Richard and Murthy, Sesh and Akkiraju, Rama and Kumaran, Santhosh and Derebail, Annap , title =. Operations Research , volume =. 2002 , doi =. https://doi.org/10.1287/opre.50.2.249.425 , abstract =

    2002 doi

  20. [28]

    Absi, Nabil and Kedad-Sidhoum, Safia , title =. RAIRO. Operations Research , pages =. 2007 , publisher =. doi:10.1051/ro:2007014 , mrnumber =

    2007 doi

  21. [29]

    The multi-item capacitated lot-sizing problem with setup times and shortage costs , journal =

    Nabil Absi and Safia Kedad-Sidhoum , keywords =. The multi-item capacitated lot-sizing problem with setup times and shortage costs , journal =. 2008 , issn =. doi:https://doi.org/10.1016/j.ejor.2006.01.053 , url =

    2008 doi

  22. [30]

    , date =

    Loparic, Marko and Pochet, Yves and Wolsey, Laurence A. , date =. The uncapacitated lot-sizing problem with sales and safety stocks , url =. Mathematical Programming , number =. 2001 , bdsk-url-1 =. doi:10.1007/PL00011411 , id =

    2001 doi

  23. [31]

    and Wolfe, Philip , title =

    Dantzig, George B. and Wolfe, Philip , title =. Operations Research , volume =. 1960 , doi =

    1960

  24. [32]

    Benders, J. F. , title =. Numerische Mathematik , volume =. 1962 , doi =

    1962

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.