Probabilistically Checkable Arguments for all NP

ePrint Report: Probabilistically Checkable Arguments for all NP

Shany Ben-David

A probabilistically checkable argument (PCA) is a computational relaxation of PCPs, where soundness is guaranteed to hold only for false proofs generated by a computationally bounded adversary. The advantage of PCAs is that they are able to overcome the limitations of PCPs. A succinct PCA has a proof length that is polynomial in the witness length (and is independent of the non-deterministic verification time), which is impossible for PCPs, under …

adversary argument ben computational eprint report generated limitations report