Inside parsing is a best parse parsing method based on the Inside algorithm that is often used in estimating probabilistic parameters of stochastic context free grammars. It gives a best parse in O(N3G3) time where N is the input size and G is the grammar size. Earley algorithm can be made to return best parses with the same complexity in N.

By way of experiments, we show that Inside parsing can be more efficient than Earley parsing with sufficiently large grammar and sufficiently short input sentences. For instance, Inside parsing is better with sentences of 16 or less words for a grammar containing 429 states. In practice, parsing can be made efficient by employing the two methods selectively.

The redundancy of Inside algorithm can be reduced by the topdown filtering using the chart produced by Earley algorithm, which is useful in training the probabilistic parameters of a grammar. Extensive experiments on Penn Tree corpus show that the efficiency of Inside computation can be improved by up to 55%.

Let f be a de Morgan read-once function of n variables. Let fε be the random restriction obtained by independently assigning to each variable of f, the value 0 with probability (1 -ε)/2, the value 1 with the same probability, and leaving it unassigned with probability ε. We show that fε depends, on the average, on only O(εαn + εn1/α) variables, where . This result is asymptotically the tightest possible. It improves a similar result obtained recently by Håstad, Razborov and Yao.

A k×n array with entries from the q-letter alphabet {0, 1, …, q − 1} is said to be t-covering if each k × t submatrix has (at least one set of) qt distinct rows. We use the Lovász local lemma to obtain a general upper bound on the minimal number K = K(n, t, q) of rows for which a t-covering array exists; for t = 3 and q = 2, we are able to match the best-known such bound. Let Kλ = Kλ(n, t, q), (λ ≥ 2), denote the minimum number of rows that guarantees the existence of an array for which each set of t columns contains, amongst its rows, each of the qt possible ‘words’ of length t at least λ times. The Lovász lemma yields an upper bound on Kλ that reveals how substantially fewer rows are needed to accomplish subsequent t-coverings (beyond the first). Finally, given a random k × n array, the Stein–Chen method is employed to obtain a Poisson approximation for the number of sets of t columns that are deficient, i.e. missing at least one word.

Long regressive sequences in well-quasi-ordered sets contain ascendingsubsequences of length n. The complexity of the corresponding function H(n) is studied in the Grzegorczyk-Wainer hierarchy. An extension to regressive canonical colourings is indicated.

Suppose that each vertex of a graph independently chooses a colour uniformly from the set {1, …, k}; and let Si be the random set of vertices coloured i. Farr shows that the probability that each set Si is stable (so that the colouring is proper) is at most the product of the k probabilities that the sets Si separately are stable. We give here a simple proof of an extension of this result.

For any positive integer k and ε > 0, there exist nk,ε, ck, e > 0 with the following property. Given any system of n > nk,ε points in the plane with minimal distance at least 1 and any t1, t2…, tk ≥ 1, the number of those pairs of points whose distance is between ti and for some 1 ≤ i ≤ k, is at most (n2/2) (1 − 1/(k+1)+ε). This bound is asymptotically tight.

Kuratowski's Theorem, perhaps the most famous result in graph theory, states that K5 and K3,3 are the only non-planar graphs for which both G\e, the deletion of the edge e, and G/e, the contraction of the edge e, are planar for all edges e of G. We characterize the almost-planar graphs, those non-planar graphs for which G\e or G/e is planar for all edges e of G. This paper gives two characterizations of the almost-planar graphs: an explicit description of the structure of almost-planar graphs; and an excluded minor criterion. We also give a best possible bound on the number of edges of an almost-planar graph.

We derive an explicit formula for the difference χ(G;λ) − χ(G|X;λ)χ(Tx(G); λ)/(λ − 1), where χ(G;λ) is the characteristic polynomial of a simple matroid G, G|X is the restriction of G to a flat X in G, and Tx(G) is the complete principal truncation of G at the flat X. Two counting proofs of this formula are given. The first uses the critical problem and the second uses the broken-circuit complex. We also derive several inequalities involving Whitney numbers of the first kind and other numerical invariants.

The square lattice site percolation model critical probability is shown to be at most .679492, improving the best previous mathematically rigorous upper bound. This bound is derived by extending the substitution method to apply to site percolation models.

We prove that every connected graph of order n ≥ 2 has an induced subgraph with all degrees odd of order at least cn/log n, where cis a constant. We also give a bound in terms of chromatic number, and resolve the analogous problem for random graphs.

Let ex(n, K3,3) denote the maximum number of edges of a K3,3-free graph on n vertices. Improving earlier results of Kővári, T. Sós and Turán on Zarankiewicz' problem, we obtain that Brown's example for a maximal K3,3-free graph is asymptotically optimal. Hence .

Let ℝ+(ℋn),ℤ(ℋn),ℤ+(ℋn) be, respectively, the cone over ℝ, the lattice and the cone over ℤ, generated by all cuts of the complete graph on n nodes. For i ≥ 0, let has exactly i realizations in ℤ+(ℋn)}. We show that is infinite, except for the undecided case and empty and for i = 0, n ≤ 5 and for i ≥ 2, n ≤ 3. The set contains 0,1,∞ nonsimplicial points for n ≤ 4, n = 5, n ≥ 6, respectively. On the other hand, there exists a finite number t(n) such that t(n)d ∈ ℤ+(ℋn) for any ; we also estimate such scales for classes of points. We construct families of points of and ℤ+(ℋn), especially on a 0-lifting of a simplicial facet, and points d ∈ ℝ+(ℋn) with di, n = t for 1 ≤ i ≤ n − 1.

Two natural classes of polymatroids can be associated with hypergraphs: the so-called Boolean and hypergraphic polymatroids. Boolean polymatroids carry virtually all the structure of hypergraphs; hypergraphic polymatroids generalize graphic matroids. This paper considers algorithmic problems associated with recognizing members of these classes. Let k be a fixed positive integer and assume that the k-polymatroid ρ is presented via a rank oracle. We present an algorithm that determines in polynomial time whether ρ is Boolean, and if it is, finds the hypergraph. We also give an algorithm that decides in polynomial time whether ρ is the hypergraphic polymatroid associated with a given hypergraph. Other structure-theoretic results are also given.

The perceptron learning algorithm quite naturally yields an algorithm for finding a linearly separable boolean function consistent with a sample of such a function. Using the idea of a specifying sample, we give a simple proof that, in general, this algorithm is not efficient.

In the Penney Ante game two players choose one binary string of length k each in turn, and toss a coin repeatedly. If at some stage the last k outcomes match one of their strings, the player with that string wins. The case k ≤ 4 is somewhat exceptional and in any case easily done. For k ≥ 5, Guibas and Odlyzko proved that the second player's optimal strategy is to choose the first k − 1 digits of the first player's string prefixed by 0 or 1. They conjectured that these two choices are never equally good. We prove that this conjecture is correct. Then we prove that 01…100 (with k − 1 ones) is an optimal strategy for the first player, and find all the strategies that are equally good.

Using a group-theoretic approach, we derive some Erdős-Ko-Rado-type results for certain Sperner families of chains and antichains in partial orders. In particular, we establish Bollobás-type inequalities for arbitrary Sperner families of intersecting affine subspaces, and special intersecting Sperner families in generalized Boolean algebras.

We describe a time randomized algorithm that estimates the number of feasible solutions of a multidimensional knapsack problem within 1 ± ε of the exact number. (Here r is the number of constraints and n is the number of integer variables.) The algorithm uses a Markov chain to generate an almost uniform random solution to the problem.

An analytic study is made of the correlation structure of Tausworthe and linear congruential random number generators. The former case is analyzed by the bit mask correlations recently introduced by Compagner. The latter is studied first by an extension to word masks, which include spectral test coefficients as special cases, and then by the bit mask procedure. Although low order bit mask coefficients vanish in both cases, the Tausworthe generator appears to produce a substantially smaller non-vanishing correlation set for large masks – but with larger correlation values – than does the linear congruential.