Communication complexity towards lower bounds on circuit depth

Karchmer, Raz, and Wigderson (1995) discuss the circuit depth complexity of n-bit Boolean functions constructed by composing up to d = log n/log log n levels of k = log n-bit Boolean functions. Any such function is in AC¹ . They conjecture that circuit depth is additive under composition, which would imply that any (bounded fan-in) circuit for this problem requires depth. This would separate AC¹ from NC¹. They recommend using the communication game characterization of circuit depth. In order to develop techniques for using communication complexity to prove circuit depth lower bounds, they suggest an intermediate communication complexity problem which they call the Universal Composition Relation. We give an almost optimal lower bound of dk—O(d ²(k log k)^1/2) for this problem. In addition, we present a proof, directly in terms of communication complexity, that there is a function on k bits requiring circuit depth. Although this fact can be easily established using a counting argument, we hope that the ideas in our proof will be incorporated more easily into subsequent arguments which use communication complexity to prove circuit depth bounds. Received: July 30, 1999.     

Graph Complexity and Slice Functions

   Abstract. A graph-theoretic approach to study the complexity of Boolean functions was initiated by Pudlák, R?dl, and Savicky [PRS] by defining models of computation on graphs. These models generalize well-known models of Boolean complexity such as circuits, branching programs, and two-party communication complexity. A Boolean function f is called a 2-slice function if it evaluates to zero on inputs with less than two 1's and evaluates to one on inputs with more than two 1's. On inputs with exactly two 1's f may be nontrivially defined. There is a natural correspondence between 2-slice functions and graphs. Using the framework of graph complexity, we show that sufficiently strong superlinear monotone lower bounds for the very special class of {2-slice functions} would imply superpolynomial lower bounds over a complete basis for certain functions derived from them. We prove, for instance, that a lower bound of n ^1+Ω(1) on the (monotone) formula size of an explicit 2-slice function f on n variables would imply a 2 ^Ω(ℓ) lower bound on the formula size over a complete basis of another explicit function g on l variables, where l=Θ( log n) . We also consider lower bound questions for depth-3 bipartite graph complexity. We prove a weak lower bound on this measure using algebraic methods. For instance, our result gives a lower bound of Ω(( log n) ³ / ( log log n) ⁵ ) for bipartite graphs arising from Hadamard matrices, such as the Paley-type bipartite graphs. Lower bounds for depth-3 bipartite graph complexity are motivated by two significant applications: (i) a lower bound of n ^Ω(1) on the depth-3 complexity of an explicit n -vertex bipartite graph would yield superlinear size lower bounds on log-depth Boolean circuits for an explicit function, and (ii) a lower bound of

The “log rank” conjecture for modular communication complexity

The “log rank” conjecture involves the question of how precisely the deterministic communication complexity of a problem can be described in terms of algebraic invariants of the communication matrix of this problem. We answer this question in the context of modular communication complexity. We show that the modular communication complexity can be exactly characterised in terms of the logarithm of a certain rigidity function of the communication matrix. Thus, we are able to exactly determine the modular communication complexity of several problems, such as, e.g., set disjointness, comparability, and undirected graph connectivity. From the bounds obtained for the modular communication complexity we deduce exponential lower bounds on the size of depth-two circuits having arbitrary symmetric gates at the bottom level and a MOD _m -gate at the top. Received: April 14, 2000.     

Free binary decision diagrams for the computation of EAR n

Free binary decision diagrams (FBDDs) are graph-based data structures representing Boolean functions with the constraint (additional to binary decision diagram) that each variable is tested at most once during the computation. The function EAR_n is the following Boolean function defined for n × n Boolean matrices: EAR_n(M) = 1 iff the matrix M contains two equal adjacent rows. We prove that each FBDD computing EAR_n has size at least and we present a construction of such diagrams of size approximately .     

Graph Complexity and Slice Functions

Abstract. A graph-theoretic approach to study the complexity of Boolean functions was initiated by Pudlák, Rödl, and Savický [PRS] by defining models of computation on graphs. These models generalize well-known models of Boolean complexity such as circuits, branching programs, and two-party communication complexity. A Boolean function f is called a 2-slice function if it evaluates to zero on inputs with less than two 1's and evaluates to one on inputs with more than two 1's. On inputs with exactly two 1's f may be nontrivially defined. There is a natural correspondence between 2-slice functions and graphs. Using the framework of graph complexity, we show that sufficiently strong superlinear monotone lower bounds for the very special class of {2-slice functions} would imply superpolynomial lower bounds over a complete basis for certain functions derived from them. We prove, for instance, that a lower bound of n ^1+Ω(1) on the (monotone) formula size of an explicit 2-slice function f on n variables would imply a 2 ^Ω(?) lower bound on the formula size over a complete basis of another explicit function g on l variables, where l=Θ( log n) . We also consider lower bound questions for depth-3 bipartite graph complexity. We prove a weak lower bound on this measure using algebraic methods. For instance, our result gives a lower bound of Ω(( log n) ³ / ( log log n) ⁵ ) for bipartite graphs arising from Hadamard matrices, such as the Paley-type bipartite graphs. Lower bounds for depth-3 bipartite graph complexity are motivated by two significant applications: (i) a lower bound of n ^Ω(1) on the depth-3 complexity of an explicit n -vertex bipartite graph would yield superlinear size lower bounds on log-depth Boolean circuits for an explicit function, and (ii) a lower bound of $\exp((\log \log n)^{\omega(1)})$ would give an explicit language outside the class Σ ₂ ^cc of the two-party communication complexity as defined by Babai, Frankl, and Simon [BFS]. Our lower bound proof is based on sign-representing polynomials for DNFs and lower bounds on ranks of ±1 matrices even after being subjected to sign-preserving changes to their entries. For the former, we use a result of Nisan and Szegedy [NS] and an idea from a recent result of Klivans and Servedio [KS]. For the latter, we use a recent remarkable lower bound due to Forster [F1].     

Evolutionary Trees and Ordinal Assertions

Sequence data for a group of species is often summarized by a distance matrix M where M[s,t] is the dissimilarity between the sequences of species s and t . An ordinal assertion is a statement of the form ``species a and b are as similar as species c and d ' and is supported by distance matrix M if M[a,b] ≤ M[c,d] . Recent preliminary research suggests that ordinal assertions can be used to reconstruct the evolutionary history of a group of species effectively. However, further research on the mathematical and algorithmic properties of ordinal assertions is needed to facilitate the development and assessment of inference methods that utilize ordinal assertions for reconstructing evolutionary histories. A (weighted ) ordinal representation of a distance matrix M is a (weighted) phylogeny T such that, for all species a , b , c , and d labeling T , d _T (a,b) ≤ d _T (c,d) if and only if M[a,b] ≤ M[c,d], where d _T is the weighted path length when T is weighted, otherwise d _T is the unweighted path length. Hence, an ordinal representation of M is a phylogeny that supports the same ordinal assertions supported by M , and so is the focus of our examination of the mathematical and algorithmic properties of ordinal assertions. As it turns out, ordinal representations are rich in structure. In this paper several results on weighted and unweighted ordinal representations are presented: — The unweighted ordinal representation of a distance matrix is unique. This generalizes the well-known result that no two phylogenies share the same distance matrix [10], [21]. — The unweighted ordinal representation of a distance matrix can be found in O(n ² log ² (n)) time. The algorithm presented improves upon an O(n ³ ) algorithm by Kannan and Warnow [13] that finds binary unweighted ordinal representations of distance matrices. — Under certain conditions, weighted ordinal representations can be found in polynomial time. Received May 11, 1997; revised March 13, 1998.     

Randomness, Stochasticity, and Approximations

Polynomial time unsafe approximations for intractable sets were introduced by Meyer and Paterson [9] and Yesha [19], respectively. The question of which sets have optimal unsafe approximations has been investigated extensively; see, e.g., [1], [5], [15], and [17]. Recently, Wang [15], [17] showed that polynomial time random sets are neither optimally unsafe approximable nor Δ -levelable. In this paper we show that: (1) There exists a polynomial time stochastic set in the exponential time complexity class which has an optimal unsafe approximation. (2) There exists a polynomial time stochastic set in the exponential time complexity class which is Δ -levelable. The above two results answer a question asked by Ambos-Spies and Lutz [2]: What kind of natural complexity property can be characterized by p -randomness but not by p -stochasticity? Our above results also extend Ville's [13] historical result. The proof of our first result shows that, for Ville's stochastic sequence, we can find an optimal prediction function f such that we will never lose our own money betting according to f (except the money we have earned), that is to say, if at the beginning we have only $1 and we always bet $1 that the next selected bit is 1 , then we always have enough money to bet on the next bit. Our second result shows that there is a stochastic sequence for which there is a betting strategy f such that we will never lose our own money betting according to f (except the money we have earned), but there is no such optimal betting strategy. That is to say, for any such betting strategy, we can find another betting strategy which could be used to make money more quickly. Received May 1997, and in final form September 1998.     

Quantum Algorithms for Weighing Matrices and Quadratic Residues

Abstract. In this article we investigate how we can employ the structure of combinatorial objects like Hadamard matrices and weighing matrices to devise new quantum algorithms. We show how the properties of a weighing matrix can be used to construct a problem for which the quantum query complexity is significantly lower than the classical one. It is pointed out that this scheme captures both Bernstein and Vazirani's inner-product protocol, as well as Grover's search algorithm. In the second part of the article we consider Paley's construction of Hadamard matrices, which relies on the properties of quadratic characters over finite fields. We design a query problem that uses the Legendre symbol χ (which indicates if an element of a finite field F _q is a quadratic residue or not). It is shown how for a shifted Legendre function f _s (i)=χ(i+s) , the unknown s ∈ F _q can be obtained exactly with only two quantum calls to f _s . This is in sharp contrast with the observation that any classical, probabilistic procedure requires more than log q + log ((1-ɛ )/2) queries to solve the same problem.     

Limiting negations in non-deterministic circuits

The minimum number of NOT gates in a Boolean circuit computing a Boolean function f is called the inversion complexity of f. In 1958, Markov determined the inversion complexity of every Boolean function and, in particular, proved that log₂(n+1) NOT gates are sufficient to compute any Boolean function on n variables. In this paper, we consider circuits computing non-deterministically and determine the inversion complexity of every Boolean function. In particular, we prove that one NOT gate is sufficient to compute any Boolean function in non-deterministic circuits if we can use an arbitrary number of guess inputs.     

A lower bound on the mod 6 degree of the OR function

We examine the computational power of modular counting, where the modulus m is not a prime power, in the setting of polynomials in Boolean variables over Z _m . In particular, we say that a polynomial P weakly represents a Boolean function f (both have n variables) if for any inputs x and y in {0,1}ⁿ, we have whenever . Barrington et al. (1994) investigated the minimal degree of a polynomial representing the OR function in this way, proving an upper bound of O(n ^{1/ r} ) (where r is the number of distinct primes dividing m) and a lower bound of . Here, we show a lower bound of when m is a product of two primes and in general. While many lower bounds are known for a much stronger form of representation of a function by a polynomial (Barrington et al. 1994, Tsai 1996), very little is known using this liberal (and, we argue, more natural) definition. While the degree is known to be for the generalized inner product because of its high communication complexity (Grolmusz 1995), our bound is the best known for any function of low communication complexity and any modulus not a prime power. received 29 September 1994     

On Capital Investment

We deal with the problem of making capital investments in machines for manufacturing a product. Opportunities for investment occur over time, every such option consists of a capital cost for a new machine and a resulting productivity gain, i.e., a lower production cost for one unit of product. The goal is that of minimizing the total production costs and capital costs when future demand for the product being produced and investment opportunities are unknown. This can be viewed as a generalization of the ski-rental problem and related to the mortgage problem [3]. If all possible capital investments obey the rule that lower production costs require higher capital investments, then we present an algorithm with constant competitive ratio. If new opportunities may be strictly superior to previous ones (in terms of both capital cost and production cost), then we give an algorithm which is O (min{1+log C , 1+log log P , 1+log M }) competitive, where C is the ratio between the highest and the lowest capital costs, P is the ratio between the highest and the lowest production costs, and M is the number of investment opportunities. We also present a lower bound on the competitive ratio of any on-line algorithm for this case, which is Ω (min{log C , log log P / log log log P , log M / log log M }). This shows that the competitive ratio of our algorithm is tight (up to constant factors) as a function of C , and not far from the best achievable as a function of P and M . Received February 6, 1997; revised November 17, 1997.     

An improved complexity hierarchy on the depth of Boolean functions

Summary Circuit depth is an important complexity measure for a Boolean function. Let some Boolean function of n variables have depth k according to an arbitrary binary basis . For each j where [log n]jk we prove the existence of a Boolean function f with the following properties. f depends essentially on n variables and the depth of f according to is exactly j Thus we state the best possible hierarchy result on the depth of all nondegenerate Boolean functions.     

A characterization of span program size and improved lower bounds for monotone span programs

We give a characterization of span program size by a combinatorial-algebraic measure. The measure we consider is a generalization of a measure on covers which has been used to prove lower bounds on formula size and has also been studied with respect to communication complexity.?In the monotone case our new methods yield lower bounds for the monotone span program complexity of explicit Boolean functions in n variables over arbitrary fields, improving the previous lower bounds on monotone span program size. Our characterization of span program size implies that any matrix with superpolynomial separation between its rank and cover number can be used to obtain superpolynomial lower bounds on monotone span program size. We also identify a property of bipartite graphs that is suficient for constructing Boolean functions with large monotone span program complexity. Received: September 30, 2000.     

Choosing,Agreeing, and Eliminating in Communication Complexity

We consider several questions inspired by the direct-sum problem in (two-party) communication complexity. In all questions, there are k fixed Boolean functions f ₁,…,f _k and each of Alice and Bob has k inputs, x ₁,…,x _k and y ₁,…,y _k , respectively. In the eliminate problem, Alice and Bob should output a vector σ₁,…,σ _k such that f _i (x _i , y _i ) ≠ σ _i for at least one i (i.e., their goal is to eliminate one of the 2 ^k output vectors); in the choose problem, Alice and Bob should return (i, f _i (x _i , y _i )), for some i (i.e., they choose one instance to solve), and in the agree problem they should return f _i (x _i , y _i ), for some i (i.e., if all the k Boolean values agree then this must be the output). The question, in each of the three cases, is whether one can do better than solving one (say, the first) instance. We study these three problems and prove various positive and negative results. In particular, we prove that the randomized communication complexity of eliminate, of k instances of the same function f, is characterized by the randomized communication complexity of solving one instance of f.     

The Complexity of Boolean Functions in Different Characteristics

Every Boolean function on n variables can be expressed as a unique multivariate polynomial modulo p for every prime p. In this work, we study how the degree of a function in one characteristic affects its complexity in other characteristics. We establish the following general principle: functions with low degree modulo p must have high complexity in every other characteristic q. More precisely, we show the following results about Boolean functions f : {0, 1}ⁿ → {0, 1} which depend on all n variables, and distinct primes p, q:

On P versus NP $ \cap $ co-NP for decision trees and read-once branching programs

It is known that if a Boolean function f in n variables has a DNF and a CNF of size then f also has a (deterministic) decision tree of size exp(O(log n log² N)). We show that this simulation cannot be made polynomial: we exhibit explicit Boolean functions f that require deterministic trees of size exp where N is the total number of monomials in minimal DNFs for f and ?f. Moreover, we exhibit new examples of explicit Boolean functions that require deterministic read-once branching programs of exponential size whereas both the functions and their negations have small nondeterministic read-once branching programs. One example results from the Bruen—Blokhuis bound on the size of nontrivial blocking sets in projective planes: it is remarkably simple and combinatorially clear. Other examples have the additional property that f is in AC⁰. Received: June 5 1997.     

Recognition of interval Boolean functions

Interval functions constitute a special class of Boolean functions for which it is very easy and fast to determine their functional value on a specified input vector. The value of an n-variable interval function specified by interval [a,b] (where a and b are n-bit binary numbers) is true if and only if the input vector viewed as an n-bit number belongs to the interval [a,b]. In this paper we study the problem of deciding whether a given disjunctive normal form represents an interval function and if so then we also want to output the corresponding interval. For general Boolean functions this problem is co-NP-hard. In our article we present a polynomial time algorithm which works for monotone functions. We shall also show that given a Boolean function f belonging to some class which is closed under partial assignment and for which we are able to solve the satisfiability problem in polynomial time, we can also decide whether f is an interval function in polynomial time. We show how to recognize a “renamable” variant of interval functions, i.e., their variable complementation closure. Another studied problem is the problem of finding an interval extension of partially defined Boolean functions. We also study some other properties of interval functions.      

Finding least-weight subsequences with fewer processors

By restricting weight functions to satisfy the quadrangle inequality or the inverse quadrangle inequality, significant progress has been made in developing efficient sequential algorithms for the least-weight subsequence problem [10], [9], [12], [16]. However, not much is known on the improvement of the naive parallel algorithm for the problem, which is fast but demands too many processors (i.e., it takesO(log² n) time on a CREW PRAM with n³/logn processors). In this paper we show that if the weight function satisfies the inverse quadrangle inequality, the problem can be solved on a CREW PRAM in O(log² n log logn) time withn/log logn processors, or in O(log² n) time withn logn processors. Notice that the processor-time complexity of our algorithm is much closer to the almost linear-time complexity of the best-known sequential algorithm [12].     

Unbounded-error quantum query complexity

This work studies the quantum query complexity of Boolean functions in an unbounded-error scenario where it is only required that the query algorithm succeeds with a probability strictly greater than 1/2. We show that, just as in the communication complexity model, the unbounded-error quantum query complexity is exactly half of its classical counterpart for any (partial or total) Boolean function. Moreover, connecting the query and communication complexity results, we show that the “black-box” approach to convert quantum query algorithms into communication protocols by Buhrman-Cleve—Wigderson [STOC’98] is optimal even in the unbounded-error setting.We also study a related setting, called the weakly unbounded-error setting, where the cost of a query algorithm is given by q+log(1/2(p−1/2)), where q is the number of queries made and p>1/2 is the success probability of the algorithm. In contrast to the case of communication complexity, we show a tight multiplicative Θ(logn) separation between quantum and classical query complexity in this setting for a partial Boolean function. The asymptotic equivalence between them is also shown for some well-studied total Boolean functions.