The SMACOF Algorithm
The stress function can be expanded as follows:
Note that the first term is a constant and the second term is quadratic in X (i.e. for the Hessian matrix V the second term is equivalent to tr) and therefore relatively easily solved. The third term is bounded by:
where has:
- for
and for
and .
Proof of this inequality is by the Cauchy-Schwartz inequality, see Borg (pp. 152–153).
Thus, we have a simple quadratic function that majorizes stress:
The iterative minimization procedure is then:
- at the kth step we set
- stop if otherwise repeat.
This algorithm has been shown to decrease stress monotonically (see de Leeuw).
Read more about this topic: Stress Majorization