A New Multipoint Symmetric Secant Method with a Dense Initial Matrix

In large-scale optimization, when either forming or storing Hessian matrices are prohibitively expensive, quasi-Newton methods are often used in lieu of Newton’s method because they only require first-order information to approximate the true Hessian.  Multipoint symmetric secant (MSS) methods can be thought of as generalizations of quasi-Newton methods in that they attempt to impose additional requirements on their approximation of the … Read more