New Benchmark Suite for Linear Ordering Problem Uses Modern Economic Data
Outdated economic benchmarks? This paper introduces fresh LOP instances and multi-solution metrics.
The Linear Ordering Problem (LOP) is a core combinatorial optimization task used to triangulate economic input-output tables, revealing which industries are critical to an economy. However, most existing LOP algorithms rely on benchmarks from outdated macroeconomic data that no longer reflect modern economies. Worse, LOP instances often yield many distinct global optima that differ substantially, causing instability for applications that depend on a single solution.
In a new preprint, researchers Fabrizio Fagiolo, Marco Baioletti, and Valentino Santucci introduce a novel benchmark suite built from contemporary real-world economic data. They also propose an algorithmic scheme using state-of-the-art LOP metaheuristics to produce diverse sets of high-quality solutions, along with metrics to evaluate both quality and diversity. Experiments validate the suite under traditional single-solution and new multi-scenario settings, promising more reliable economic analysis and decision-making.
- Existing LOP benchmarks rely on outdated macroeconomic data that misrepresent modern economies.
- The paper introduces a new benchmark suite using up-to-date real-world economic data.
- A scheme using metaheuristics generates diverse high-quality solutions with metrics for quality and diversity.
Why It Matters
Enables more accurate economic input-output analysis and robust solution selection for critical industries.