June 2006
Intermediate to advanced
515 pages
16h 22m
English
1 Both terms are open to attack by a purist, for different reasons; but the interpretation is correctly understood.
2 Gradients of other vector quantities, appearing later in this chapter, follow the same protocol.
3 Exercise 6 of Chap. 4 gives an example of an efficient processing of such active set updates.
4 Mostly, we expect μi(a) > 0. But, in degenerate cases, μi(a) = 0 along with gi(a) (x*) is also possible. How do you interpret such a case?
6 See exercise 9 of Chap. 4, for the projection formula.
7 also called the method of multipliers
8 Exercise 5 of Chap. 8 exhibits the smoothening of the eigenstructure of Hϕ(λ) by the penalty term.
Read now
Unlock full access