arXiv · 2606.21799
Towards a Doubly Efficient IP=PSPACE
Abstract
We show that every language in PSPACE decidable by a Turing machine in time $T(n)=n^{O(\log n)}$ admits a doubly efficient interactive proof system: the prover runs in time polynomial in T(n), and the verifier runs in time polynomial in n. This extends the best previously known regime for such proof systems from $T(n)=n^{O(\sqrt{\log n / \log\log n})}$, established by Berger, Goyal, Hong, and Kalai (FOCS 2025), to $T(n)=n^{O(\log n)}$. Beyond improving the range of T, our protocol is substantially simpler than previous doubly efficient proofs for time-bounded PSPACE. Earlier constructions proceed indirectly: they first build batch interactive proofs and then invoke them as a black box to obtain doubly efficient protocols. In contrast, we give a direct construction. This not only simplifies the proof but also points to a more promising route for future improvements.
Explore related subjects
Keep this discovery
Liyan Chen, Matthew M. Hong, Yael Tauman Kalai, Zoe Xi. 2026-06-19. Towards a Doubly Efficient IP=PSPACE. https://arxiv.org/abs/2606.21799
Cite the original work for its findings. Save a collection to share your selection of sources.