The behavior of iterative solvers is less predictable comparing to direct methods like Gaussian elimination. It depends on various factors like matrix properties, desired precision, choice of initial solution vector. At the same time, they are far more effective at solving large linear systems that are crucial in many scientific and engineering problems. Since these computations are often both resource-intensive and time-consuming, the development of useful solution strategies is essential.

Previous research has extensively explored how different matrix properties influence the efficiency of linear solvers. However, comparatively little attention has been paid to the role of the right-hand side (RHS) of the system.
This thesis looks into the possibilities of effect of the RHS on convergence behavior.
It analyzes the correlation between RHS properties and performance of iterative methods.

The goal is to identify patterns that can inform improvements to existing solver frameworks and numerical libraries. By incorporating RHS characteristics into solver selection strategies, it may be possible to enhance the efficiency of automated solver recommendations.