Separate linear-time classes for multitape versus random-access models

Determine whether the linear-time class for one-dimensional multitape Turing machines differs from the linear-time class for random-access machine models by proving a language that is solvable in linear time on one model but not on the other.

Background

The paper’s simulation techniques exploit locality in multitape models and highlight obstacles for random-access models. The authors note that a negative answer to the extension question would yield a separation of linear-time classes.

They emphasize this separation as a longstanding open question in complexity, linking it to foundational model-robustness issues.