Software for Data-Based Stochastic Programming Using Estimation

Article Computer Science, Interdisciplinary Applications

Efficient Stochastic Programming in Julia

Martin Biel et al.

Summary: Stochastic Programs.jl is a user-friendly and powerful open-source framework for stochastic programming written in Julia. It allows efficient modeling with expressive syntax and interactive analysis of models. The framework seamlessly scales to distributed environments, enabling running small instances locally and larger instances on supercomputers or clouds using parallel optimization algorithms.

INFORMS JOURNAL ON COMPUTING (2022)

Add to Collection

Article Management

From Data to Decisions: Distributionally Robust Optimization Is Optimal

Bart P. G. van Parys et al.

Summary: This study focuses on stochastic programs where decision makers can only access a finite set of samples from unknown distributions. By proposing a meta-optimization problem, the research aims to find the least conservative predictors and prescriptors, which serve as near-optimal candidates for decision-making. Leveraging large deviations theory, the unique solution to this problem involves solving a distributionally robust optimization problem over distributions within a given relative entropy distance from the empirical data distribution.

MANAGEMENT SCIENCE (2021)

Add to Collection

Article Computer Science, Interdisciplinary Applications

SDDP.j1: A Julia Package for Stochastic Dual Dynamic Programming

Oscar Dowson et al.

Summary: SDDP.j1 is an open-source library for solving multistage stochastic programming problems using the stochastic dual dynamic programming algorithm. It is built on JuMP, an algebraic modeling language in Julia, and provides a solver-agnostic, user-friendly interface. Leveraging unique features of Julia, such as multiple dispatch, it offers an extensible framework for practitioners to build on.

INFORMS JOURNAL ON COMPUTING (2021)

Add to Collection

Article Mathematics, Applied