SDDP.jl: A Julia Package for Stochastic Dual Dynamic Programming

Abstract

We present SDDP.jl, an open-source library for solving multistage stochastic programming problems using the stochastic dual dynamic programming algorithm. SDDP.jl is built on JuMP, an algebraic modeling language in Julia. JuMP provides SDDP.jl with a solver-agnostic, user-friendly interface. In addition, we leverage unique features of Julia, such as multiple dispatch, to provide an extensible framework for practitioners to build on our work. SDDP.jl is well tested, and accessible documentation is available at https://github.com/odow/SDDP.jl .

Aversão a Risco E Política Ótima De Investimentos E Financiamentos De Uma Corporação: Uma Abordagem via Programação Dinâmica Estocástica

“…Nesta seção descreveremos os resultados da simulação do modelo descrito no Capítulo 3 e as premissas descritas na seção 4.1. Para a resolução e simulação dos modelos PDDE utilizamos o Framework desenvolvido por Dowson e Kapelevich [45] para a linguagem de programação Julia. O solver utilizado foi o Gurobi, versão 7.5.2.…”

“…O problema do caso base para λ = 0.5 foi resolvido em 187 segundos utilizando um MacBook Pro com processador Intel Core i7 de 2.7 Ghz e 16 GB de memória. O modelo foi simulado em 50 estágios nas instâncias padrão do Framework desenvolvido por Dowson e Kapelevich [45].…”

A Framework for Assessing the Impacts of Network Formulations in the Operation of Hydrothermal Power Systems

“…To develop a tool that can be used by both researchers and industry professionals, we take advantage of the Julia language [15] and two main packages: PowerModels.jl [16], which implements power flow models for electrical dispatch, SDDP.jl [18], which implements the stochastic dual dynamic programming algorithm. Both PowerModels.jl and SDDP.jl handle their respective inconsistent policy, derived as a more realistic chained planning-and-implementation process (as per Algorithm 2 or approximation schemes such as Algorithm 3) offers more accurate estimates for future decisions than the related planning policy.…”

“…More implementations of SDDP are also available in Julia [49] [50], but SDDP.jl [18] has proven an easy to learn, efficient version of SDDP that is flexible enough for the purposes of the HydroPowerModels.jl package. -Build the multistage, hydrothermal steady-state power network optimization problem.…”

“…This tool enables not only the assessment of the results found in this work but also allows researchers in the electrical industry to test new ideas, leveraging state-of-the-art solution methods and mathematical formulations for the economic dispatch problem. It combines two other opensource packages developed in Julia Language [15], namely, PowerModels.jl, developed by Coffrin [16,17] to solve steady-state power network optimization problems, and SDDP.jl, developed by Oscar Dowson [18,19] to solve multistage convex stochastic optimization problems. This combination puts together two state-of-the-art open-source packages sharing the same layer for modeling mathematical optimization in Julia called JuMP.jl [20].…”