Skip to content

Bibliography

This bibliography collects every external source cited across this site’s methodology chapters, plus background references that the methodology rests on without quoting. Chapters cite an entry in short form, author and year, linked to the section below that holds it.

For domain terms, see Glossary.


  • Benders, J.F. (1962). Partitioning procedures for solving mixed-variables programming problems. Numerische Mathematik, 4(1), 238–252. doi:10.1007/BF01386316 The original Benders decomposition paper. Foundation for the L-shaped method and SDDP. Background reference for SDDP Algorithm, Cut Management, What Novomodelo Solves.

  • Pereira, M.V.F. & Pinto, L.M.V.G. (1991). Multi-stage stochastic optimization applied to energy planning. Mathematical Programming, 52(1–3), 359–375. doi:10.1007/BF01582895 The original SDDP paper. Foundational for the entire algorithm and for the hydrothermal-dispatch application that motivates Novomodelo. Background reference for SDDP Algorithm, What Novomodelo Solves.

  • Birge, J.R. (1985). Decomposition and partitioning methods for multistage stochastic linear programs. Operations Research, 33(5), 989–1007. doi:10.1287/opre.33.5.989 Nested decomposition for multistage stochastic linear programs: each stage passes its decisions forward and receives cuts from its successors, the scheme SDDP applies over sampled scenarios. Background reference for SDDP Algorithm.

  • Birge, J.R. & Louveaux, F.V. (1988). A multicut algorithm for two-stage stochastic linear programs. European Journal of Operational Research, 34(3), 384–392. doi:10.1016/0377-2217(88)90159-2 The multicut L-shaped method, which keeps one cut per outcome where the single-cut method aggregates the outcomes into one cut. The single-cut formulation of Cut Management is contrasted with it. Cited in Cut Management §3, SDDP Algorithm §6.

  • Birge, J.R. & Louveaux, F.V. (2011). Introduction to Stochastic Programming, 2nd edition. Springer. doi:10.1007/978-1-4614-0237-4 Standard textbook reference for stochastic programming theory and decomposition methods. Background reference for SDDP Algorithm.

  • Philpott, A.B. & Guan, Z. (2008). On the convergence of stochastic dual dynamic programming and related methods. Operations Research Letters, 36(4), 450–455. doi:10.1016/j.orl.2008.01.013 Convergence theory for SDDP under finitely many scenarios. Background reference for Cut Management.

  • Shapiro, A. (2011). Analysis of stochastic dual dynamic programming method. European Journal of Operational Research, 209(1), 63–72. doi:10.1016/j.ejor.2010.08.007 Convergence analysis, complexity bounds, and risk-averse extensions for SDDP. Cited in Risk Measures §10.

  • Dowson, O. (2020). The policy graph decomposition of multistage stochastic programming problems. Networks, 76(1), 3–23. doi:10.1002/net.21932 Defines the policy graph: nodes with local subproblems and cut pools, linked by probability-weighted arcs, a cycle among them representing an infinite horizon. Cited in Policy Graphs §4, Horizon Modes §6.


  • de Matos, V.L., Philpott, A.B. & Finardi, E.C. (2015). Improving the performance of Stochastic Dual Dynamic Programming. Journal of Computational and Applied Mathematics, 290, 196–208. doi:10.1016/j.cam.2015.04.048 Cut selection strategies for SDDP, including the Level-1 active-cut criterion. Cited in Cut Management §7, §8.

  • Guigues, V. & Römisch, W. (2012). Sampling-based decomposition methods for multistage stochastic programs based on extended polyhedral risk measures. SIAM Journal on Optimization, 22(2), 286–312. doi:10.1137/100811696 Risk-averse sampling-based decomposition methods, SDDP among them, under extended polyhedral risk measures, a class that contains CVaR, with a convergence proof. Cited in Cut Management §4 (Tier 2), The SDDP Framework in One Page §2.

  • Guigues, V. (2016). Convergence analysis of sampling-based decomposition methods for risk-averse multistage stochastic convex programs. SIAM Journal on Optimization, 26(4), 2468–2494. doi:10.1137/140983136 Convergence of risk-averse sampling-based decomposition methods, SDDP among them, for multistage stochastic convex programs. Cited in Cut Management §4 (Tier 2), The SDDP Framework in One Page §2.

  • Guigues, V. (2017). Dual dynamic programing with cut selection: Convergence proof and numerical experiments. European Journal of Operational Research, 258(1), 47–57. doi:10.1016/j.ejor.2016.10.047 Defines Limited-Memory Level 1 (LML1) cut selection, which keeps at each trial point only the oldest of the cuts with the highest value there, and proves the convergence of dual dynamic programming with it. Cited in Cut Management §6, §7, §9.

  • Bandarra, M. & Guigues, V. (2021). Single cut and multicut stochastic dual dynamic programming with cut selection for multistage stochastic linear programs: convergence proof and numerical experiments. Computational Management Science, 18(2), 125–148. doi:10.1007/s10287-021-00387-8. Preprint: arXiv:1902.06757 Convergence proof for single-cut and multicut SDDP with Level 1 and LML1 cut selection: with finitely many scenarios, an optimal first-stage solution after finitely many iterations with probability 1. Cited in Cut Management §9.


  • Rockafellar, R.T. & Uryasev, S. (2000). Optimization of conditional value-at-risk. Journal of Risk, 2(3), 21–41. doi:10.21314/JOR.2000.038 Definition of CVaR and the linearisation that allows it to be embedded in linear programmes — the basis for risk-averse cut aggregation in SDDP. Background reference for Risk Measures.

  • Philpott, A.B. & de Matos, V.L. (2012). Dynamic sampling algorithms for multi-stage stochastic programs with risk aversion. European Journal of Operational Research, 218(2), 470–483. doi:10.1016/j.ejor.2011.10.056 Dynamic sampling under risk aversion with Markovian scenario transitions. Background reference for Risk Measures.

  • Philpott, A.B., de Matos, V.L. & Finardi, E.C. (2013). On solving multistage stochastic programs with coherent risk measures. Operations Research, 61(4), 957–970. doi:10.1287/opre.2013.1175 Time-consistent risk-averse SDDP with CVaR. Dual representation and aggregation weights for risk-averse cut generation, a convergence proof for the risk-averse method, and an inner approximation of the cost-to-go from which a deterministic upper bound is evaluated. Cited in Risk Measures §9, §10, Cut Management §4 (Tier 2), The SDDP Framework in One Page §2, Upper Bound Evaluation (Appendix, References).


  • Costa, B.F.P. & Leclère, V. (2023). Duality of upper bounds in stochastic dynamic programming. Optimization Online. Preprint: optimization-online.org/?p=23738 Duality framework for inner-approximation upper bounds, which builds on the inner approximation of Philpott, de Matos & Finardi (2013). Basis for the SIDP inner-approximation estimator described in Upper Bound Evaluation. Cited in Upper Bound Evaluation (Appendix, References).

  • Diniz, A.L. & Maceira, M.E.P. (2008). A four-dimensional model of hydro generation for the short-term hydrothermal dispatch problem considering head and spillage effects. IEEE Transactions on Power Systems, 23(3), 1298–1308. doi:10.1109/TPWRS.2008.922253 The piecewise-linear hydro production function (FPHA): generation as a function of storage, turbined flow and spillage, with the head computed from the forebay and tailrace levels, and the nonconcave regions of the exact function approximated by convexification and regression. Origin of the approach fitted in Hydro Production Function Models §2. Novomodelo fits its planes over storage and turbined flow at spillage = 0 and carries the spillage effect by a per-plane lateral-flow secant rather than a spillage axis in the fit. Cited in Hydro Production Function Models §2.

  • Box, G.E.P. & Jenkins, G.M. (1976). Time Series Analysis: Forecasting and Control, revised edition. Holden-Day, San Francisco. Foundational textbook for ARMA / autoregressive time-series modelling and the Yule-Walker estimation method that underlies the PAR(p) fitting procedure. Its sample autocovariance estimator uses the population divisor 1/N1/N. Cited in PAR(p) Inflow Model §3.2. Background reference for Scenario Generation.

  • Hipel, K.W. & McLeod, A.I. (1994). Time Series Modelling of Water Resources and Environmental Systems. Elsevier, Amsterdam. Chapter 14 is the canonical presentation of periodic models: the PAR model definition, the periodic autocovariance/ACF conventions (the more recent observation names the season), the periodic Yule-Walker equations, the lag-0 variance identity, the periodic PACF with its ±1.96/Nm\pm 1.96/\sqrt{N_m} significance band, and the periodic-stationarity condition. Novomodelo’s fitting procedure is this formulation written in correlation form over the sms_m-standardized series. Cited in PAR(p) Inflow Model §3.5.

  • Maceira, M.E.P. & Damázio, J.M. (2006). Use of the PAR(p) model in the stochastic dual dynamic programming optimization scheme used in the operation planning of the Brazilian hydropower system. Probability in the Engineering and Informational Sciences, 20(1), 143–156. doi:10.1017/S0269964806060098 The periodic autoregressive PAR(p) model as fitted inside SDDP for the Brazilian system. Source of the iterative AR-order-reduction procedure that keeps composed lag contributions non-negative. Cited in PAR(p) Inflow Model §3.6, §7.6.

  • Treistman, F., Maceira, M.E.P., Damázio, J.M. & Cruz, C.B. (2020). Periodic time series model with annual component applied to operation planning of hydrothermal systems. In 2020 International Conference on Probabilistic Methods Applied to Power Systems (PMAPS), Liège, Belgium, 1–6. doi:10.1109/PMAPS47429.2020.9183472 The PAR(p)-A model: the periodic autoregressive model augmented with an annual component, a regression term on the rolling annual average (the mean of the twelve previous monthly inflows), which extends the model’s memory to long dry and wet periods. Cited in PAR(p) Inflow Model §7.

  • Larroyd, P.V., Pedrini, R., Beltrán, F., Teixeira, G., Finardi, E.C. & Picarelli, L.B. (2022). Dealing with Negative Inflows in the Long-Term Hydrothermal Scheduling Problem. Energies, 15(3), 1115. doi:10.3390/en15031115 Penalty, truncation and truncation-combined-with-penalty strategies for negative PAR(p) inflows in SDDP: under each, the PAR state keeps the raw realization and a correction adds water to the inflow the water balance receives, bounded by the truncation amount under the truncation strategies. Its §2.5 shows the stage problems’ values falling below the master problems built from the cuts under truncation, so the lower bound can exceed an upper bound evaluated on the stage problems. Cited in Inflow Non-Negativity Solution Methods §6.2, §7, §9.

  • Maceira, M.E.P., Terry, L.A., Costa, F.S., Damázio, J.M. & Melo, A.C.G. (2002). Chain of optimization models for setting the energy dispatch and spot price in the Brazilian system. In Proceedings of the 14th Power Systems Computation Conference (PSCC), Seville, Spain. The NEWAVE / DECOMP / GEVAZP optimization chain for the Brazilian system, in which the medium-term study takes its terminal cost-to-go from the cuts of the long-term policy. Source of the DECOMP-style scenario tree — a deterministic trunk with branching at the final stage — expressed as a terminal fan of sibling nodes traversed by enumerated selection, and the published counterpart of the chained studies of Post-Study Boundary §5. Cited in Scenario Generation §6, Post-Study Boundary §5.


  • Birge, J.R. & Zhao, G. (2007). Successive linear approximation solution of infinite-horizon dynamic stochastic programs. SIAM Journal on Optimization, 18(4), 1165–1186. doi:10.1137/060665506 Infinite-horizon discounted stochastic linear programs solved by cutting-plane approximation of a stationary value function, with a convergence proof. Cited in Horizon Modes §6.

  • Shapiro, A. & Ding, L. (2020). Periodical multistage stochastic programs. SIAM Journal on Optimization, 30(3), 2083–2102. doi:10.1137/19M129406X Infinite-horizon multistage programs with periodic data: the periodic analogue of the discounted Bellman equations, whose value functions repeat with the period, and the SDDP algorithm that shares cuts across the stages of one period. Cited in Horizon Modes §6.

  • Costa, B.F.P., Calixto, A.O., Sousa, R.F.S., Figueiredo, R.T., Penna, D.D.J., Khenayfis, L.S. & Oliveira, A.M.R. (2025). Boundary conditions for hydrothermal operation planning problems: the infinite horizon approach. Proceeding Series of the Brazilian Society of Computational and Applied Mathematics, 11(1), 1–7. doi:10.5540/03.2025.011.01.0355 Periodic policy graph and infinite-horizon SDDP formulation for hydrothermal operation planning. The reserved cyclic design that Horizon Modes describes follows its presentation of the season function τ(t)\tau(t), the cycle convergence inequality, the season-indexed cut pool with its cut-sharing equation, and the fixed-point Bellman operator. Cited in Horizon Modes §6.


  • Diniz, A.L. & Maceira, M.E.P. (2013). Multi-lag Benders decomposition for power generation planning with nonanticipativity constraints on the dispatch of LNG thermal plants. In H.I. Gassmann & W.T. Ziemba (eds.), Stochastic Programming: Applications in Finance, Energy, Planning and Logistics, World Scientific Series in Finance, 443–464. doi:10.1142/9789814407519_0016 Thermal dispatch decided stages before delivery, carried as state of the stage problems of SDDP: the published counterpart of the anticipated-thermal commitments of State Augmentation §5. Cited in State Augmentation §5.

  • Dowson, O. & Kapelevich, L. (2021). SDDP.jl: A Julia Package for Stochastic Dual Dynamic Programming. INFORMS Journal on Computing, 33(1), 27–33. doi:10.1287/ijoc.2020.0987. Documentation: sddp.dev Reference SDDP implementation in Julia. Influenced the sampling-scheme abstractions, the convex-combination risk-measure structure and the notation conventions in Novomodelo. Its EAVaR measure puts the weight λ\lambda on the expectation; Risk Measures §3 gives the mapping to Novomodelo’s λ\lambda. Cited in Notation Conventions, Scenario Generation §3, Risk Measures §3, Policy Graphs §4.

  • Huangfu, Q. & Hall, J.A.J. (2018). Parallelizing the dual revised simplex method. Mathematical Programming Computation, 10(1), 119–142. doi:10.1007/s12532-017-0130-5 HiGHS dual simplex implementation. HiGHS is Novomodelo’s default LP solver. Background reference for LP Warm-Start.


  • Tomlin, J.A. (1975). On scaling linear programming problems. In Computational Practice in Mathematical Programming, Mathematical Programming Studies 4, 146–166. doi:10.1007/BFb0120718 Compares scaling methods for linear programs, among them geometric-mean scaling, which sets each row or column factor from the largest and smallest magnitudes in it. Cited in LP Layout and Scaling §2.2.

  • Higham, N.J. (1988). Computing a nearest symmetric positive semidefinite matrix. Linear Algebra and its Applications, 103, 103–118. doi:10.1016/0024-3795(88)90223-6 The nearest symmetric positive-semidefinite matrix in the Frobenius norm to a symmetric matrix is the matrix with its negative eigenvalues set to zero: the eigenvalue clipping applied when factorising the spatial correlation matrix. Cited in PAR(p) Inflow Model §6, Scenario Generation §2.1.

  • Elble, J.M. & Sahinidis, N.V. (2012). Scaling linear optimization problems prior to application of the simplex method. Computational Optimization and Applications, 52(2), 345–371. doi:10.1007/s10589-011-9420-4 Computational comparison of the scaling methods applied before the simplex method, among them geometric-mean scaling and equilibration. Cited in LP Layout and Scaling §2.2.

  • Higham, N.J. (2002). Computing the nearest correlation matrix — a problem from finance. IMA Journal of Numerical Analysis, 22(3), 329–343. doi:10.1093/imanum/22.3.329 The nearest correlation matrix problem, which adds to the positive-semidefinite projection the unit-diagonal constraint that eigenvalue clipping does not keep. Cited in PAR(p) Inflow Model §6.


  • CEPEL — Centro de Pesquisas de Energia Elétrica (n.d.). Documentação Técnica dos modelos para Planejamento da Operação do SIN – Ambiente Libs. Online manual: see.cepel.br/manual/libs/latest/ Official documentation of the NEWAVE / DECOMP / DESSEM suite of stochastic-dispatch models operated for the Brazilian system. Cited only for practitioner terms and notation: the equivalent-terms tables of the glossary, the GEVAZP residual distribution (section Distribuição Lognormal 3 parâmetros) and the PAR(p) notation (section Modelo Autorregressivo Periódico - Par(p)). The methods those models implement are credited to their primary articles above: FPHA → Diniz & Maceira (2008); PAR(p) and iterative order reduction → Maceira & Damázio (2006); PAR(p)-A → Treistman et al. (2020); DECOMP-style scenario tree → Maceira et al. (2002). Novomodelo’s dead-volume filling model is its own and is not attributed here. Cited in Glossary — Brazilian Power-System Ecosystem and Glossary — Equivalent terms in other planning tools.