Worked on stabilizing GPU minimizer interactions for rigid bodies with rotational degrees of freedom in the glotzerlab/hoomd-blue repository. Addressed a bug by introducing a dedicated scope for the host-side ArrayHandle in C++, ensuring proper lock release before entering anisotropic simulation blocks. This approach prevented GPUArray access errors during rigid body rotations, enhancing the reliability of anisotropic simulations. Focused on GPU computing and CUDA, the work improved simulation stability and reduced runtime failures. The contribution centered on a targeted bug fix rather than new features, demonstrating careful attention to low-level memory management and safe parallel execution in scientific software.