Proof Complexity Lower Bounds from Algebraic Circuit Complexity
نویسندگان
چکیده
We give upper and lower bounds on the power of subsystems of the Ideal Proof System (IPS), the algebraic proof system recently proposed by Grochow and Pitassi [GP14], where the circuits comprising the proof come from various restricted algebraic circuit classes. This mimics an established research direction in the boolean setting for subsystems of Extended Frege proofs, where proof-lines are circuits from restricted boolean circuit classes. Except one, all of the subsystems considered in this paper can simulate the well-studied Nullstellensatz proof system, and prior to this work there were no known lower bounds when measuring proof size by the algebraic complexity of the polynomials (except with respect to degree, or to sparsity). We give two general methods of converting certain algebraic lower bounds into proof complexity ones. Our methods require stronger notions of lower bounds, which lower bound a polynomial as well as an entire family of polynomials it defines. Our techniques are reminiscent of existing methods for converting boolean circuit lower bounds into related proof complexity results, such as feasible interpolation. We obtain the relevant types of lower bounds for a variety of classes (sparse polynomials, depth-3 powering formulas, read-once oblivious algebraic branching programs, and multilinear formulas), and infer the relevant proof complexity results. We complement our lower bounds by giving short refutations of the previously-studied subset-sum axiom using IPS subsystems, allowing us to conclude strict separations between some of these subsystems. Our first method is a functional lower bound, a notion of Grigoriev and Razborov [GR00], which is a function f̂ : {0, 1} → F such that any polynomial f agreeing with f̂ on the boolean cube requires large algebraic circuit complexity. For our classes of interest, we develop functional lower bounds where f̂(x) equals 1/p(x) where p is a constant-degree polynomial, which in turn yield corresponding IPS lower bounds for proving that p is nonzero over the boolean cube. In particular, we show super-polynomial lower bounds for refuting variants of the subset-sum axiom in various IPS subsystems. Our second method is to give lower bounds for multiples, that is, to give explicit polynomials whose all (nonzero) multiples require large algebraic circuit complexity. By extending known techniques, we are able to obtain such lower bounds for our classes of interest, which we then use to derive corresponding IPS lower bounds. Such lower bounds for multiples are of independent interest, as they have tight connections with the algebraic hardness versus randomness paradigm. ∗Email: [email protected]. Department of Computer Science, Princeton University. Supported by the Princeton Center for Theoretical Computer Science. †Email: [email protected]. Department of Computer Science, Tel Aviv University, Tel Aviv, Israel. The research leading to these results has received funding from the European Community’s Seventh Framework Programme (FP7/2007-2013) under grant agreement number 257575. ‡Email: [email protected]. Department of Computer Science, Royal Holloway, University of London, UK. §Email: [email protected]. Institute for Advanced Study, Princeton. This research was partially supported by NSF grant CCF-1412958. 1 ar X iv :1 60 6. 05 05 0v 1 [ cs .C C ] 1 6 Ju n 20 16
منابع مشابه
Towards an algebraic natural proofs barrier via polynomial identity testing
We observe that a certain kind of algebraic proof—which covers essentially all known algebraic circuit lower bounds to date—cannot be used to prove lower bounds against VP if and only if what we call succinct hitting sets exist for VP. This is analogous to the Razborov–Rudich natural proofs barrier in Boolean circuit complexity, in that we rule out a large class of lower bound techniques under ...
متن کاملAlgebra in Computational Complexity
At its core, much of Computational Complexity is concerned with combinatorial objects and structures. But it has often proven true that the best way to prove things about these combinatorial objects is by establishing a connection to a more well-behaved algebraic setting. Indeed, many of the deepest and most powerful results in Computational Complexity rely on algebraic proof techniques. The Ra...
متن کاملGenerating Matrix Identities and Proof Complexity
Motivated by the fundamental lower bounds questions in proof complexity, we initiate the study of matrix identities as hard instances for strong proof systems. A matrix identity of d × d matrices over a field F, is a non-commutative polynomial f(x1, . . . , xn) over F such that f vanishes on every d × d matrix assignment to its variables. We focus on arithmetic proofs, which are proofs of polyn...
متن کاملSeminar 16411 Algebraic and Combinatorial Methods in Computational Complexity
Computational Complexity is concerned with the resources that are required for algorithms to detect properties of combinatorial objects and structures. It has often proven true that the best way to argue about these combinatorial objects is by establishing a connection (perhaps approximate) to a more well-behaved algebraic setting. Indeed, many of the deepest and most powerful results in Comput...
متن کاملCircuit Complexity and Multiplicative Complexity of Boolean Functions
In this note, we use lower bounds on Boolean multiplicative complexity to prove lower bounds on Boolean circuit complexity. We give a very simple proof of a 7n/3− c lower bound on the circuit complexity of a large class of functions representable by high degree polynomials over GF(2). The key idea of the proof is a circuit complexity measure assigning different weights to XOR and AND gates.
متن کامل