PORTAL: Controllable Landscape Generator for Continuous Optimization-Part I: Framework
🤖 AI Summary
Existing continuous optimization test suites suffer from limited flexibility, poor controllability, and insufficient landscape diversity. To address these issues, this paper introduces the first **controllable benchmark generation platform** specifically designed for continuous optimization, enabling **fine-grained, orthogonal control** over key landscape properties—including basin curvature, condition number, variable interaction, and surface roughness. A hierarchical neutralization mechanism mitigates component-scale imbalance, while element-wise and coupled operators—integrated with block-wise configurable structures—support directional deformation, exponential scaling, and sequential composition, ensuring both landscape stability and interpretability. The platform enables multimodal, asymmetric, and heterogeneous roughness modeling, facilitating progressive difficulty scaling and arbitrary multi-component composition. Its open-source implementation has been successfully deployed in algorithm evaluation, meta-learning, and pedagogical settings, significantly enhancing the generality and customization capability of benchmark construction.
Application Category
📝 Abstract
Benchmarking is central to optimization research, yet existing test suites for continuous optimization remain limited: classical collections are fixed and rigid, while previous generators cover only narrow families of landscapes with restricted variability and control over details. This paper introduces PORTAL (Platform for Optimization Research, Testing, Analysis, and Learning), a general benchmark generator that provides fine-grained, independent control over basin curvature, conditioning, variable interactions, and surface ruggedness. PORTAL's layered design spans from individual components to block-wise compositions of multi-component landscapes with controllable partial separability and imbalanced block contributions. It offers precise control over the shape of each component in every dimension and direction, and supports diverse transformation patterns through both element-wise and coupling operators with compositional sequencing. All transformations preserve component centers and local quadratic structure, ensuring stability and interpretability. A principled neutralization mechanism prevents unintended component domination caused by exponent or scale disparities, which addresses a key limitation of prior landscape generators. On this foundation, transformations introduce complex landscape characteristics, such as multimodality, asymmetry, and heterogeneous ruggedness, in a controlled and systematic way. PORTAL enables systematic algorithm analysis by supporting both isolation of specific challenges and progressive difficulty scaling. It also facilitates the creation of diverse datasets for meta-algorithmic research, tailored benchmark suite design, and interactive educational use. The complete Python and MATLAB source code for PORTAL is publicly available at [https://github.com/EvoMindLab/PORTAL].
Problem
Research questions and friction points this paper is trying to address.
Generates diverse continuous optimization landscapes with fine-grained control
Addresses limitations of rigid test suites and narrow prior generators
Enables systematic algorithm analysis and tailored benchmark design
Innovation
Methods, ideas, or system contributions that make the work stand out.
PORTAL provides fine-grained control over landscape characteristics.
It uses layered design with controllable partial separability.
Neutralization mechanism prevents unintended component domination.
🔎 Similar Papers
No similar papers found.
