The polyadic structure of factorable function tensors with applications to high-order minimization techniques (Q1069447)

Factorable functions are shown to have arrays of Nth-order derivatives (tensors) which are naturally computed as sums of generalized outer product matrices (polyads). The computational implications of this for high-order minimization techniques (such as Halley's method of tangent hyperbolas) are investigated. A direct derivation of these high-order techniques is also given.