Packing arrays and packing designs (Q697642)

For parameters for which orthogonal arrays do not exist, one may study related packing arrays. A packing array is a \(k \times b\) array with \(q\)-ary entries and with the property that in every \(t \times b\) subarray, every \(t \times 1\) column vector appears at most once. One usually wants to maximize \(b\) for given parameters. Without further constraints this equals the problem of finding \(q\)-ary error-correcting codes of length \(k\) and Hamming distance \(k-t+1\), so one may wish to impose yet another parameter, \(n\), meaning that the array must contain a set of \(n\) columns that are at Hamming distance \(k\) from each other. Obviously \(n \leq q\). The authors consider packing arrays with \(t=2\). In particular, the situation when \(n\) is close to \(q\) is studied. Several lower and upper bounds on the value of \(b\) are given for such packing arrays.