POISSON APPROXIMATIONS FOR 2-DIMENSIONAL PATTERNS

JAMES C. FU1 AND MARKOS V. KOUTRAS2

1 Department of Statistics, The University of Manitoba,
Winnipeg, Manitoba, Canada R3T 2N2

2 Department of Mathematics, Section of Statistics and O.R., University of Athens,
Panepistemiopolis, Athens 157 10, Greece

(Received November 5, 1992; revised March 18, 1993)

Abstract.    Let X = (Xij)n×n be a random matrix whose elements are independent Bernoulli random variables, taking the values 0 and 1 with probability qij and pij (pij + qij = 1) respectively. Upper and lower bounds for the probabilities of m non-overlapping occurrences of a square submatrix with all its elements being equal to 1, are obtained. Some Poisson convergence theorems are established for n \to \infty. Numerical results indicate that the proposed bounds perform very well, even for moderate and small values of n.

Key words and phrases:    Random matrix, Bernoulli random variables, Poisson approximation, patterns, consecutive-k-out-of-n:F system.

Source ( TeX , DVI , PS )