[BA] Enabling Variability of Parallelization and Distribution for Solving Expensive Blackbox Optimization Problems

Expensive Blackbox Optimization~(EBBO) problems are widely-spread in industry and research in domains such as: machine learning, search-based software engineering, operation research and many others.
Enhancing the solving approaches with parallelization and distribution capabilities is a common technique to accelerate the optimization process.
A recent research at the chair of software technology showed that improving variability of a solving approach, also improves optimization quality.
State-of-the-art EBBO solvers are bound to a single parallelization and distribution mechanism, lacking variability in this respect.
The goal of this thesis is to enable variability of parallelization and distribution in EBBO and to study its effect on the optimization process.
To reach this goal the following tasks need to be accomplished:

• Examine state-of-the-art EBBO solvers and derive a set of techniques for parallelization and distribution of EBBO.
• Propose an approach for variability modeling of parallelization and distribution based on the conducted review.
• Integrate the proposed approach into BRISE-MPL for EBBO.
• Evaluate the effect of enabling variability of parallelization and distribution on the optimization process.

Betreuer: Dmytro Pukhkaiev