OptQuest 10.0.0

Documentation (HTML)

• OptQuest API Documentation
• OptQuest 10 API
• OptQuest 10 Legacy API

Release Notes:

New Unified API Architecture

OptQuest 10 delivers a comprehensive declarative API that provides consistent functionality across Java, C#, C++, C, and Python environments. This unified approach provides developers the same interface regardless of their preferred programming language.

For Java developers, the legacy OptQuest 9 API continues to be fully supported with no deprecation timeline, ensuring backward compatibility for existing projects. C#, C++, C, and Python developers who want to use OptQuest 10 will need to use the new API.

Additional language support is available for platforms that support C-style calling conventions, including MATLAB, R, Julia, Rust, Go, and IDL. A beta Windows COM and Linux C and C++ interfaces are also available upon request for specialized integration scenarios.

Comprehensive API documentation is available above with detailed implementation examples and best practices.

Enhanced Metaheuristic Algorithms

OptQuest continues to advance its world-class metaheuristic algorithms for black-box optimization with significant improvements in version 10:

Diversification Enhancement: Refined diversification heuristics now identify promising solution regions more efficiently, reducing computation time while improving solution quality.

Multi-Objective Frontier Exploration: A new specialized heuristic dramatically improves frontier exploration capabilities for multi-objective optimization problems, providing better coverage of the Pareto front.

Advanced Goal-Finding: For multi-objective scenarios, when users provide targeted ranges for objective values, enhanced goal-finding mechanisms focus the optimization in the specified regions of the Pareto front.

Multi-Dimensional Variable Support

OptQuest 10 introduces powerful new variable types that expand optimization capabilities into spatial domains:

Geolocation Variables: Native support for optimizing locations on Earth’s surface, incorporating geographical constraints and distance calculations assuming a spherical earth model.

Tuple Variables: Support for arbitrary Euclidean space optimization, enabling efficient search across multi-dimensional coordinate systems.

These new variable types leverage specialized search algorithms that understand the geometric properties of spatial domains, resulting in more efficient convergence to optimal locations compared to treating coordinates as independent variables.

Expanded Expression Functions

The mathematical expression engine has been significantly enhanced with additional functions that simplify the formulation of complex constraints and objectives. These new functions not only make optimization problems easier to express but also improve evaluation performance. New supported functions include clamp, signum, finite, eq, ne, lt, le, gt, ge, and if.

The expanded function library enables users to tackle a broader range of optimization scenarios without requiring custom evaluation code. See the “Functions Available In Expressions” section of the API documentation (here) for the complete list of supported mathematical operations.

Advanced Nonlinear Constraint Handling

A new non-linear constraint-handling heuristic significantly accelerates the discovery of feasible solutions in heavily constrained optimization problems. This improvement is particularly beneficial for:

• Problems with multiple nonlinear constraints
• Scenarios where the feasible region represents a small portion of the search space
• Applications requiring rapid identification of constraint-satisfying solutions

The enhanced algorithm reduces the time required to find initial feasible points, allowing more computational resources to focus on optimization within the feasible region.

Adaptive Sampling and Metamodel Creation

OptQuest 10 introduces adaptive sampling as a powerful complement to traditional optimization approaches. While optimization seeks specific optimal points, adaptive sampling comprehensively characterizes the entire response surface across the design space.

Adaptive Sampling Process: The engine generates trial solutions across the domain, building an accurate representation of system behavior with minimal evaluations. This approach excels at helping an analysts understand global trade-offs, identify behavioral patterns, or map response surface characteristics.

Metamodel Generation: With data from an adaptive sampling or optimization run, OptQuest can construct metamodels—sophisticated predictive models that approximate the response surface based on evaluated solutions. These metamodels enable rapid estimation of metric values at any point within the domain without requiring additional costly simulations.

Note: Metamodel creation is currently only available through the legacy Java API, with support for other languages planned for future releases.

Enhanced Localization Support

Resolved: Fixed a parsing issue affecting enumeration variables in locales that use commas as decimal separators. The string parsing for enumeration variable lists now correctly respects the current system locale, ensuring proper functionality across international deployments.

OptQuest 9.1.2

Documentation

• OptQuest Java/.NET Documentation (HTML)
• Java/.NET API (HTML)
• Release Notes