A-optimal partially balanced fractional \(2^{m_ 1+m_ 2}\) factorial designs of resolution V, with \(4\leq m_ 1+m_ 2\leq 6\) (Q1107241)

This paper deals with the problem of constructing fractional factorial (FF) designs of \(2^{m_ 1+m_ 2}\) series such that the covariance matrix of the estimates of effects up to and including two-factor interactions is invariant under any permutation within each \(m_ 2\), \(m_ 1\) factors. These designs of resolution \(V\) are obtained by using the concept of partially balanced array (PB-array). By using extended triangular multi-dimensional partially balanced association schemes, an explicit expression for the characteristic polynomial of the information matrix of a \(2^{m_ 1+m_ 2}\) design of resolution \(V\) is derived. The following series of A-optimal PBFF designs of resolution \(V\) are then \(presented:\) \(-m_ 1=m_ 2=2\), \(11\leq N\leq 28\); \(m_ 1=2\), \(m_ 2=3\), \(16\leq N\leq 32\); \(m_ 1=2\), \(m_ 2=4\), \(22\leq N\leq 40\); \(m_ 1=m_ 2=3\), \(22\leq N\leq 40\); here \(N\) denotes the total number of treatment-combinations.