Algorithms resilient to memory corruptions

• Applicant: Gabriel Moruz, Ph.D.
• Subject Area: Theoretical Computer Science
• Term: from 2011 to 2014
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 197026676

Due to intrinsic design characteristics, modern memories are prone to soft memory errors, where bits may flip corrupting the data in the affected memory cells. Such memory corruptions are triggered by a variety of factors ranging from power failures to cosmic rays, and trends point out that the rate they occur will increase in the future.Our aims are three-fold. First, we plan to propose computational models to reflect better the reality than existing models, while still providing guarantees for the robustness of the algorithms. Secondly, we want to study geometric data structures in the already established faulty-memory RAM model. Finally, we will design and implement resilient algorithms for real-world applications such as clustering.

• DFG Programme: Research Grants