ABSTRACT
On a small computer, the principal objection to the Nelder–Mead and variable metric methods is their requirement for a working space proportional to n 2, where n is the number of parameters in the function to be minimised. The parameters b and gradient g require only n elements each, so it is tempting to consider algorithms which make use of this information without the requirement that it be collected in a matrix. In order to derive such a method, consider once again the quadratic form ) S ( b ) = 1 2 b T H b − c T b + (any scalar) https://s3-euw1-ap-pe-df-pch-content-public-p.s3.eu-west-1.amazonaws.com/9781315139784/ed4048a6-e22b-4325-b5dd-51851ea3ba32/content/eq680.tif"/>
