1 / 13

Program Checkers for NP and Black-box separations

Program Checkers for NP and Black-box separations. Mohammad Mahmoody School on Black-Box Impossibilities July 2014. Main Message. Open for 25 years: Do all NP languages have “program checkers”? [ Manuel Blum,  Sampath Kannan : Designing Programs That Check Their Work. STOC 1989 ]

khalil
Download Presentation

Program Checkers for NP and Black-box separations

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Program Checkers for NPand Black-box separations Mohammad Mahmoody School on Black-Box Impossibilities July 2014

  2. Main Message • Open for 25 years: Do all NP languages have “program checkers”? [ Manuel Blum, SampathKannan: Designing Programs That Check Their Work. STOC 1989 ] • Message: If NP not checkable  black-box impossibly results follow • Examples: • No NP-hard one-way function [HoMX10] •  No NP-hard hash functions [HaMX10] • No black-box -round ZK for NP from OWPs with negligible soundness error [GWXY] • No black-box 3-round ZK for NP from OWFs with -bit verifier messages [MP12]

  3. Plan • Part I: Short intro to program checkers • Part II: Applications to separations

  4. Plan • Part I: Short intro to program checkers • Part II: Applications to separations

  5. Definition • Program claims to solve language • A program checker gets and as input and runs “safely” : • If P correct : • If Pbuggy:or “I found a bug”. • Both need to hold with “high” probability. • Example: graph isomorphism (or non-isomorphism)

  6. Checking SAT ? • Is SATis checkable? Open since [BK89,FRS89]. • If P(x) says “x satisfiable”  make sure by self-reducibility • What if P(x) = “x not satisfiable”? should still convince the checker… • would be a PCP for Moreover: each query efficiently computable using NP oracle

  7. PCP Two provers • [FRS89]: SAT is checkable iffcoNP is provable with two provers in • Proof: on board! • So: proving coNP with a singleprover in is a stronger tasklets call it : strongprogram checkers for NP. • Known proof systems for coNP require #P –complex provers…

  8. Recalling the Results • NP-hard one-way function  NP checkable • NP-hard hash functions  NP strongly checkable • black-box -round ZK for NP from OWPs with negligible soundness error • No black-box 3-round ZK for NP from OWFs with -bit verifier messages NP has strongprogram checker

  9. Plan • Part I: Short intro to program checkers • Part II: Applications [Just OWF]

  10. Ruling out P  NP Crypto • Open question: Ruling out black-box reductions that prove “NP BPP  OWF exists”(under complexity assumptions) • Potentially easier to rule out stronger primitives (e.g. public-key) Prior works: • [FF 91, BT 04]: non-adaptivereduction • [Brassard 79]: general reduction, but for one-way permutations • [BL 13]: general reduction, but for homomorphic encryption

  11. NP-hard OWF  SAT checkable • Theorem: If R solves NP given any weakly inverting oracle for f There is two prover proof system for coNP with prover complexity • Proof: • Prover 1: emulates the reduction • Prover 2 either: • re-answer one of P1’s answers • or invert P1 rand of R Simulation : (ym,a1),…(ym,am) VER x yi / f(u) ai / u’ P2

  12. Proof Intuition 1) Only one query from P2  it is an “oracle”. 2) If P2 inverts with prob caught with prob 3) P1 should match oracle P2 (or gets caught with prob) 3) Soundness error is high?Use sequential repetition! P1 rand of R Simulation : (ym,a1),…(ym,am) VER x yi / f(u) ai / u’ P2

  13. Direct proof using PCPs • On the board!

More Related