Unformatted text preview: concave, but the duality gap may not be zero when the primal is not convex. 3) Consider the following problem: min − cx
x s.t . € (c > 0) −x ≤0
x ≤b
(b > 0) Associate with the first constraint the Lagrange multiplier “f” and with the second constraint the Lagrange multiplier “g”. Which statement is true at the optimal feasible point (a.k.a. the solution): A) f=0; g=0 B) f>0; g=0 C) f=0; g>0 D) f>0; g>0 E) None of the above The objective is a slope. A ball on the slope would roll to the right and will be blocked by the constraint x<= b. So that constraint is active while the constraints x>=0 is inactive. Hence, f=0, g>0....
View
Full Document
 Fall '12
 AlexIhler
 Machine Learning, the00

Click to edit the document details