P/poly
Generated by GPT-5-miniExpansion Funnel Raw 40 → Dedup 0 → NER 0 → Enqueued 0
| P/poly | |
|---|---|
| Name | P/poly |
| Field | Computational complexity theory |
| Introduced | 1980s |
| Notable people | Richard Karp, Michael Sipser, Leonid Levin, Stephen Cook, Mikhail Rabin, Jurgen Schmidhuber, Alexander Razborov, Steven Rudich, Noam Nisan, Michael Sipser |
| Related concepts | NP (complexity), PSPACE, BPP, EXP (complexity), Circuit complexity |
P/poly P/poly is a non-uniform complexity class describing decision problems solvable by polynomial-time machines with polynomial-bounded advice or equivalently by polynomial-size Boolean circuits. It arises in the study of Richard Karp-style reductions, Stephen Cook-style completeness, and circuit lower bounds, and is central to discussions connecting Leonid Levin’s universal machines, Michael Sipser’s textbook exposition, and structural questions about NP (complexity) and randomized classes such as BPP.
Definition
P/poly consists of languages L for which there exists a polynomial p(n) and a family of Boolean circuits {C_n} with size at most p(n) that decide membership: for every string x of length n, C_n(x)=1 iff x∈L. Equivalently, there exists a deterministic polynomial-time Turing machine M and an advice function a(n) with |a(n)| ≤ p(n) such that M(x,a(|x|)) accepts exactly the strings in L. These characterizations connect to work by Richard Karp, Stephen Cook, Leonid Levin, and later formalizations in texts by Michael Sipser and papers by Noam Nisan.
Examples and properties
Canonical examples include all unary languages that are decidable, many promise problems studied in Peter Shor-adjacent quantum complexity literature, and artificially crafted tally sets used in reductions in papers by Alexander Razborov and Steven Rudich. Properties: P/poly contains P (complexity), is closed under union and intersection, and is closed under polynomial-time many-one reductions when advice is appropriately transformed. It contains sparse languages considered by Lance Fortnow and Avi Wigderson in relation to derandomization, and interacts with probabilistic classes like BPP and counting classes related to Valiant’s theorem studied by Leslie Valiant.
Non-uniformity and advice functions
Non-uniformity is formalized via advice strings a(n) studied in complexity surveys by Michael Sipser and in the context of circuit lower bounds by Noam Nisan and Alexander Razborov. Advice functions are arbitrary strings depending only on input length, an idea appearing in complexity dialogues involving Richard Karp and Leonid Levin about uniformity versus non-uniformity. Advice-based definitions enable constructions in hardness versus randomness frameworks used by Oded Goldreich, Salil Vadhan, and Rafael Impagliazzo to relate derandomization to circuit lower bounds. Trade-offs between advice length and time resources are treated in works by Lance Fortnow and Bill McCreight.
Relationships to other complexity classes
P/poly strictly contains P (complexity) and contains classes such as BPP under widely believed derandomization hypotheses discussed by Oded Goldreich and Noam Nisan. It is contained in EXP (complexity) and relates to NP (complexity), with the Karp–Lipton theorem tying a collapse of the Polynomial hierarchy—studied by Jurgen Schmidhuber and Arora and Barak—to NP⊆P/poly. Results connecting MA and AM to non-uniform classes appear in papers by Scott Aaronson and Silverman?; the role of P/poly in separations with PSPACE was explored by Fortnow and Sipser.
Completeness and circuit characterizations
P/poly lacks natural complete languages under uniform reductions because of non-uniformity, but many circuit families capture its behavior, a theme central to the circuit complexity program led by Alexander Razborov and Steven Rudich. Circuit lower bounds separating P/poly from richer classes were pursued in landmark results by Valiant and later by researchers including Razborov and Karpinski. Levin-style universal search and Kolmogorov complexity perspectives from Leonid Levin and Ray Solomonoff inform completeness-like constructions via promise problems and non-uniform reductions, while structural theorems by Noam Nisan clarify monotone circuit characterizations.
Implications and consequences in complexity theory
Inclusion statements such as NP⊆P/poly imply major collapses like PH collapse (the Karp–Lipton theorem), linking seminal results of Richard Karp, Stephen Cook, and Arora and Barak. P/poly’s role in hardness versus randomness frameworks underlies derandomization programs initiated by Oded Goldreich, Noam Nisan, and Salil Vadhan, and influences cryptographic assumptions discussed by Ron Rivest, Adi Shamir, and Leonard Adleman. Circuit lower bounds against P/poly inform separation attempts for NP (complexity), PSPACE, and EXP (complexity), and motivate research programs in boolean circuit complexity advanced by Alexander Razborov and Steven Rudich. The class also appears in discussions of completeness notions, non-uniform reductions, and structural complexity results surveyed by Michael Sipser and Lance Fortnow.