Error-correcting Codes and Computation

Error-correcting codes are a method for a redundant representation of data, that enables one to can recover the original data even in the presence of some noise or corruption.

In addition to their wide practical applicability, error-correcting codes are also supported by a rich theory, with connections to diverse disciplines in mathematics, science, and engineering.

A particularly fruitful such connection has been the interplay with the theory of computation, where on the one hand, computational methods were used for the design of error-correcting codes admitting efficient error-correction algorithms, and in the reverse direction, such codes were useful for a variety of applications within the theory of computation.The current project aims to significantly enhance the aforementioned connections.

Our first main objective is to design error-correcting codes that on the one hand, achieve the best possible information-theoretic trade-off between their redundancy and error-resilience, and on the other hand, admit super-fast error-correction algorithms.

We further believe that such codes are a powerful tool that can be used for boosting the efficiency of various fundamental computational tasks, and our second main objective is to explore such potential applications.

We outline a couple of such potential applications within the theory of computation, to obtaining highly-efficient proof systems, fine-grained inapproximability results, fast derandomization, and code-based cryptography with low overhead.The goals we plan to pursue are fundamental and long-standing, and even a partial progress on them would be groundbreaking, with theoretical, and potentially also practical, impact.

Despite the significant challenge, there has recently been an exciting progress towards these goals (including by the PI), and we consequently believe that we are now in a unique position for pursuing these goals.