Simulation Strategies for an Efficient Local Search to solve Stochastic Scheduling Problems
Pith reviewed 2026-05-25 03:56 UTC · model grok-4.3
The pith
A t-test stopping rule for simulations lets local search solve stochastic scheduling problems with fewer runs while matching the quality of fixed large simulation counts.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The t-test based simulation stopping rule is the most effective among the techniques studied for reducing the number of simulations inside local search on both the Stochastic Parallel Machine Scheduling Problem and the Stochastic Electric Vehicle Scheduling Problem, delivering lower runtimes while keeping solution quality comparable to the quality obtained with a fixed large number of simulations per iteration.
What carries the argument
The t-test stopping rule, which halts additional simulations once statistical evidence indicates the current sample mean is sufficiently accurate.
If this is right
- Local search can be applied to stochastic scheduling problems with substantially lower per-iteration simulation cost on the two tested problem classes.
- The t-test method outperforms the other simulation-reduction techniques examined in both case studies.
- Solution quality remains comparable to exhaustive-simulation baselines while runtime decreases.
Where Pith is reading between the lines
- The same stopping rule could be inserted into other local-search or metaheuristic frameworks that rely on Monte Carlo estimates of stochastic objectives.
- Combining the t-test rule with variance-reduction sampling methods might produce further efficiency gains on the same problems.
- Repeating the experiments on instances with heavier-tailed or correlated duration distributions would test whether the observed advantage persists.
Load-bearing premise
The t-test stopping rule produces estimates accurate enough for the local search to converge to solutions whose true expected performance is comparable to those obtained with a fixed large number of simulations.
What would settle it
Evaluating the final schedules from both the t-test local search and a fixed-large-simulation local search with an independent, much larger number of simulations and finding that the t-test solutions have statistically higher true expected costs.
read the original abstract
In scheduling problems, deterministic task durations are often assumed. This usually does not capture reality and may lead to schedules that are not robust to (small) changes to these task lengths. The use of stochastic task durations therefore seems preferable. Including these in local search, which is the way to find good solutions for difficult scheduling problems, is not straightforward, though. The objective value becomes stochastic then too, and computing the expected value is often not possible. One way out of this it to approximate this value by using simulation. This is quite easy to implement in a local search algorithm, but it may require many simulations each iteration to get a reliable estimate. Hence such an approach comes with a performance penalty. In this paper, we study techniques to limit the number of simulations. Besides comparing known techniques, we propose our own method for this, which is based on $t$-tests. We evaluate these techniques on the Stochastic Parallel Machine Scheduling Problem and the Stochastic Electric Vehicle Scheduling Problem. In these case studies, we show the effectiveness of using such methods to reduce runtime while retaining solution quality. Our method using $t$-tests turns out to be most effective in both problems.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper addresses the challenge of incorporating stochastic task durations into local search for scheduling problems by approximating expected objective values via simulation. It compares existing techniques for limiting simulation counts per iteration and proposes a new method based on t-tests for early stopping. The approach is evaluated on two case studies—the Stochastic Parallel Machine Scheduling Problem and the Stochastic Electric Vehicle Scheduling Problem—where the t-test method is reported to be most effective at reducing runtime while retaining solution quality.
Significance. If the t-test stopping rule produces final solutions whose true expected performance is comparable to those found under fixed large simulation budgets, the method could meaningfully improve the practicality of simulation-based local search for stochastic combinatorial optimization problems without sacrificing solution robustness.
major comments (2)
- [Case studies / experimental evaluation] The central empirical claim—that the t-test method retains solution quality while reducing runtime—requires that final reported solutions from variable-simulation trajectories have their true expected costs verified under an independent high-fidelity estimator never used inside the search. The manuscript provides no indication that such a re-evaluation step was performed on the two case studies; without it, apparent quality retention may reflect bias in the stopping rule rather than comparable true performance.
- [Evaluation on Stochastic Parallel Machine Scheduling Problem and Stochastic Electric Vehicle Scheduling Problem] The comparison of techniques relies on reported solution quality and runtime, but the load-bearing assumption that the t-test stopping rule produces estimates accurate enough for the local search to converge to equivalent basins is not tested via any cross-validation or statistical comparison of true expectations across methods.
minor comments (1)
- [Abstract] The abstract states that 'our method using t-tests turns out to be most effective in both problems' but does not define the precise metrics (e.g., average expected cost, best-found cost, statistical significance) used to declare superiority.
Simulated Author's Rebuttal
We thank the referee for the constructive comments, which highlight important aspects of the empirical validation. We address each major comment below and will revise the manuscript to incorporate additional verification steps.
read point-by-point responses
-
Referee: [Case studies / experimental evaluation] The central empirical claim—that the t-test method retains solution quality while reducing runtime—requires that final reported solutions from variable-simulation trajectories have their true expected costs verified under an independent high-fidelity estimator never used inside the search. The manuscript provides no indication that such a re-evaluation step was performed on the two case studies; without it, apparent quality retention may reflect bias in the stopping rule rather than comparable true performance.
Authors: We agree that an independent high-fidelity re-evaluation of final solutions strengthens the claim of retained solution quality. The current experiments assess quality via the simulation estimates used during search, which could introduce bias. We will add a post-processing step in the revised manuscript: all final solutions from every method will be re-evaluated using a fixed, large simulation budget (independent of the search) to compute true expected costs, with results reported alongside the original metrics. revision: yes
-
Referee: [Evaluation on Stochastic Parallel Machine Scheduling Problem and Stochastic Electric Vehicle Scheduling Problem] The comparison of techniques relies on reported solution quality and runtime, but the load-bearing assumption that the t-test stopping rule produces estimates accurate enough for the local search to converge to equivalent basins is not tested via any cross-validation or statistical comparison of true expectations across methods.
Authors: We acknowledge that the manuscript does not include explicit cross-validation or statistical tests comparing true expected values across methods. To address this, the revised version will include a statistical comparison (e.g., paired t-tests or Wilcoxon tests) of the independently re-evaluated expected costs across all techniques on both case studies, confirming whether the t-test method converges to statistically equivalent basins. revision: yes
Circularity Check
No circularity: purely empirical comparison with no derivations or self-referential fits
full rationale
The paper contains no equations, derivations, or claimed first-principles predictions. It describes an empirical study comparing simulation-reduction techniques (including a proposed t-test method) on two scheduling case studies, reporting runtime and solution quality. No load-bearing step reduces to a fit, self-citation chain, or definitional equivalence; the central claim rests on direct experimental outcomes rather than any internal construction that would force the reported superiority.
Axiom & Free-Parameter Ledger
Reference graph
Works this paper leans on
-
[1]
(2002) Simulation-based optimization using simulated annealing with ranking and selection
Ahmed M.A., Alkhamis T.M. (2002) Simulation-based optimization using simulated annealing with ranking and selection. Computers & Operations Research 29(4):387--402. doi:10.1016/S0305-0548(00)00073-3
-
[2]
van den Akker M., van Blokland K., Hoogeveen H. (2013) Finding Robust Solutions for the Stochastic Job Shop Scheduling Problem by Including Simulation in Local Search . In: Bonifaci V., Demetrescu C., Marchetti-Spaccamela A. (eds) Experimental Algorithms, 12th International Symposium, SEA 2013, Rome, Italy, June 5-7, 2013. Proceedings, Lecture Notes in Co...
-
[3]
(2004) Simulation-based optimization using simulated annealing with confidence interval
Alkhamis T., Ahmed M. (2004) Simulation-based optimization using simulated annealing with confidence interval. In: Proceedings of the 2004 Winter Simulation Conference, 2004., vol 1. IEEE, p 519, doi:10.1109/WSC.2004.1371356
-
[4]
(1999) Simulated annealing for discrete optimization with estimation
Alkhamis T.M., Ahmed M.A., Tuan V.K. (1999) Simulated annealing for discrete optimization with estimation. European Journal of Operational Research 116(3):530--544. doi:10.1016/S0377-2217(98)00112-X
-
[5]
Alrefaei M.H., Andradóttir S. (1999) A Simulated Annealing Algorithm with Constant Temperature for Discrete Stochastic Optimization . Management Science 45(5):748--764. doi:10.1287/mnsc.45.5.748
-
[6]
(2016) Simulation optimization: a review of algorithms and applications
Amaran S., Sahinidis N.V., Sharda B., Bury S.J. (2016) Simulation optimization: a review of algorithms and applications. Annals of Operations Research 240(1):351--380. doi:10.1007/s10479-015-2019-x
-
[7]
SIAM Review53(3), 464–501 (2011)
Bertsimas D., Brown D.B., Caramanis C. (2011) Theory and Applications of Robust Optimization . SIAM Review 53(3):464--501. doi:10.1137/080734510
-
[8]
(2009) A survey on metaheuristics for stochastic combinatorial optimization
Bianchi L., Dorigo M., Gambardella L.M., Gutjahr W.J. (2009) A survey on metaheuristics for stochastic combinatorial optimization. Natural Computing 8(2):239--287. doi:10.1007/s11047-008-9098-4
-
[9]
(1997) Introduction to Stochastic Programming
Birge J.R., Louveaux F. (1997) Introduction to Stochastic Programming . Springer New York, doi:10.1007/978-1-4614-0237-4
-
[10]
(2026) Scheduling electric vehicles by simulated annealing with recombination through ILP
ten Bosch W., Hoogeveen J.A., van Kooten Niekerk M.E., de Bruin P. (2026) Scheduling electric vehicles by simulated annealing with recombination through ILP . Public Transport doi:10.1007/s12469-026-00424-2
-
[11]
(2023) Scheduling Electric Buses with Stochastic Driving Times
de Bruin P., van den Akker M., Hoogeveen H., van Kooten Niekerk M. (2023) Scheduling Electric Buses with Stochastic Driving Times . Schloss Dagstuhl – Leibniz-Zentrum für Informatik, doi:10.4230/OASICS.ATMOS.2023.14
-
[12]
(2000) Simulation budget allocation for further enhancing the efficiency of ordinal optimization
Chen C., Lin J., Y \" u cesan E., Chick S.E. (2000) Simulation budget allocation for further enhancing the efficiency of ordinal optimization. Discrete Event Dynamic Systems 10(3):251--270. doi:10.1023/A:1008349927281
-
[13]
Clapper Y., Berkhout J., Bekker R. (2024) Adaptive budget allocation in simheuristics applied to stochastic home healthcare routing and scheduling. Computers & Industrial Engineering 198:110651. doi:10.1016/j.cie.2024.110651
-
[14]
Ghiani G., Legato P., Musmanno R., Vocaturo F. (2007) A combined procedure for discrete simulation-optimization problems based on the simulated annealing framework. Computational Optimization and Applications 38(1):133--145. doi:10.1007/s10589-007-9010-7
-
[15]
(1999) An explanation of ordinal optimization: Soft computing for hard problems
Ho Y.C. (1999) An explanation of ordinal optimization: Soft computing for hard problems. Information Sciences 113(3):169--192. doi:10.1016/s0020-0255(98)10056-7
-
[16]
Juan A.A., Faulin J., Grasman S.E., Rabe M., Figueira G. (2015) A review of simheuristics: Extending metaheuristics to deal with stochastic combinatorial optimization problems. Operations Research Perspectives 2:62--72. doi:10.1016/j.orp.2015.03.001
-
[17]
(2022) Simheuristics: An Introductory Tutorial
Juan A.A., Li Y., Ammouriova M., Panadero J., Faulin J. (2022) Simheuristics: An Introductory Tutorial . In: 2022 Winter Simulation Conference (WSC), pp 1325--1339, doi:10.1109/WSC57314.2022.10015318
-
[18]
(2006) Chapter 17 Selecting the Best System
Kim S.H., Nelson B.L. (2006) Chapter 17 Selecting the Best System . In: Henderson S.G., Nelson B.L. (eds) Simulation, Handbooks in Operations Research and Management Science, vol 13. North-Holland, p 501--534, doi:10.1016/s0927-0507(06)13017-0
-
[19]
(2003) Optimization by direct search: New perspectives on some classical and modern methods
Kolda T.G., Lewis R.M., Torczon V. (2003) Optimization by direct search: New perspectives on some classical and modern methods. SIAM Review 45(3):385--482. doi:10.1137/S003614450242889
-
[20]
(2015) Simulation Modeling and Analysis , 5th edn
Law A.M. (2015) Simulation Modeling and Analysis , 5th edn. McGraw-Hill
work page 2015
-
[21]
Lee L.H., Chen C.H., Chew E.P., Li J., Pujowidianto N.A., Zhang S. (2010) A review of optimal computing budget allocation algorithms for simulation optimization problem. International Journal of Operations Research 7(2):19--31
work page 2010
-
[22]
(2025) Robustness Measures for Stochastic Parallel Machine Scheduling and Train Unit Shunting
Loman C., Pascual L., van den Akker M., van den Broek R., Hoogeveen H. (2025) Robustness Measures for Stochastic Parallel Machine Scheduling and Train Unit Shunting . doi:10.48550/arXiv.2512.15471
-
[23]
Passage G., van den Akker M., Hoogeveen H. (2025) A new, efficient approach to speed up local search by estimating the solution quality: an application to stochastic, parallel machine scheduling. Journal of Heuristics 31(3). doi:10.1007/s10732-025-09562-5
-
[24]
(2022) Electric bus planning & scheduling: A review of related problems and methodologies
Perumal S.S.G., Lusby R.M., Larsen J. (2022) Electric bus planning & scheduling: A review of related problems and methodologies. European Journal of Operational Research 301(2):395--413. doi:10.1016/j.ejor.2021.10.058
-
[25]
(1978) On two-stage selection procedures and related probability-inequalities
Rinott Y. (1978) On two-stage selection procedures and related probability-inequalities. Communications in Statistics - Theory and Methods 7(8):799--811. doi:10.1080/03610927808827671
-
[26]
Tasoglu G., Yildiz G. (2019) Simulated annealing based simulation optimization method for solving integrated berth allocation and quay crane scheduling problems. Simulation Modelling Practice and Theory 97:101948. doi:10.1016/j.simpat.2019.101948
-
[27]
(2005) The Use of Buffers in Project Management: The Trade-Off between Stability and Makespan
Van de Vonder S., Demeulemeester E., Herroelen W., Leus R. (2005) The Use of Buffers in Project Management: The Trade-Off between Stability and Makespan . International Journal of Production Economics 97(2):227--240. doi:10.1016/j.ijpe.2004.08.004
-
[28]
(1984) A Table for Rinott 's Selection Procedure
Wilcox R.R. (1984) A Table for Rinott 's Selection Procedure . Journal of Quality Technology 16(2):97--100. doi:10.1080/00224065.1984.11978896
-
[29]
Yang H.a., Lv Y., Xia C., Sun S., Wang H. (2014) Optimal Computing Budget Allocation for Ordinal Optimization in Solving Stochastic Job Shop Scheduling Problems . Mathematical Problems in Engineering 2014:1--10. doi:10.1155/2014/619254
-
[30]
(2019) Considering sample means in Rinott 's procedure with a Bayesian approach
Yoon M., Bekker J. (2019) Considering sample means in Rinott 's procedure with a Bayesian approach. European Journal of Operational Research 273(1):249--258. doi:10.1016/j.ejor.2018.06.040
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.