Synthetic Optimization Problem Generation: Show Us the Correlations!

Abstract

I n many computational experiments, correlation is induced between certain types of coefficients in synthetic (or simulated) instances of classical optimization problems. Typically, the correlations that are induced are only qualified-that is, described by their presumed intensity. We quantify the population correlations induced under several strategies for simulating 0-1 knapsack problem instances and also for correlation-induction approaches used to simulate instances of the generalized assignment, capital budgeting (or multidimensional knapsack), and set-covering problems. We discuss implications of these correlation-induction methods for previous and future computational experiments on simulated optimization problems.

A decision support tool for generating shipping data for the Hunter Valley coal chain

“…An exception is the recent paper by Hall and Posner [6] which succinctly captures the importance of data generation: "In many experiments, the methods chosen to generate synthetic data can significantly affect the results of an experiment." A different, but equally important, perspective is provided by Reilly [10] who emphasizes the importance of understanding and properly inducing correlations between characteristics of data, which is one of the major drivers and, at the same time, one of the major challenges in our stem generation research.…”

Generating new test instances by evolving in instance space

“…Instead, a handful of test instances are usually the focus of algorithm development efforts, from which inferences about performance on other instances are optimistically made. Obtaining a sufficient number of real-world or real-world-like instances for statistical inference can be difficult, especially since synthetically generated instances of problems often have quite different properties and underlying structure to real-world instances [4][5][6]. Sampling from the set of possible test instances, ensuring that the selected instances are real-world-like, and that no selection bias has been introduced that would affect the conclusions, is a significant challenge affecting algorithmic testing in both industrial and academic environments.…”

Generative Adversarial Construction of Parallel Portfolios

“…Actually, the above two cases are not rare and have been discussed in different areas in the literature. For example, it has been reported that in combinatorial optimization, some commonly used benchmark instances are not necessarily challenging [20], narrowly defined [21], and distinct from real-world instances [22]; in research areas closely related to real-world applications, such as logistics, there are also concerns that the instances proposed decades ago already could not represent the real-world cases of today due to the constant growth of big cities [23], [24].…”