Knowledge workers increasingly rely on multi-agent systems to solve complex problems. While these systems offer valuable support, they often obscure which agents contributed to a response, leading to a lack of transparency that may result in errors and reduced trust. To address this, we propose avatars that make agents’ expertise and contributions transparent. We iteratively co-designed avatars representing distinct expertise areas and validated them in an experiment (N=100). Building on this, we developed four multi-agent prototypes varying in explanation modality (text vs. avatars) and resolution (low vs. high). We then conducted a mixed-methods evaluation with an online experiment (N=124) and follow-up interviews (N=20). Qualitative results suggest that avatars foster clearer mental models, improve perceived explainability, and support users’ trust calibration without increasing cognitive load, although no significant quantitative differences were found. Our research contributes validated avatar designs, insights into explanation strategies, and design implications for explainable multi-agent systems.

Microstructures forming during ternary eutectic directional solidification processes have significant influence on the macroscopic mechanical properties of metal alloys. For a realistic simulation, we use the well established thermodynamically consistent phase-field method and improve it with a new grand potential formulation to couple the concentration evolution. This extension is very compute intensive due to a temperature dependent diffusive concentration. We significantly extend previous simulations that have used simpler phase-field models or were performed on smaller domain sizes. The new method has been implemented within the massively parallel HPC framework waLBerla that is designed to exploit current supercomputers efficiently. We apply various optimization techniques, including buffering techniques, explicit SIMD kernel vectorization, and communication hiding. Simulations utilizing up to 262,144 cores have been run on three different supercomputing architectures and weak scalability results are shown. Additionally, a hierarchical, mesh-based data reduction strategy is developed to keep the I/O problem manageable at scale.