Global multi-parametric optimal value bounds and solution estimates for separable parametric programs (Q804476): Difference between revisions

This is a discussion of a parametric method for finding global solutions for nonconvex separable programs. The problem is put in the form \[ \text{ minimize } f(x,\epsilon)=\sum^{n}_{p=1}f_ p(x_ p)\quad s.t.\quad g_ i(x,\epsilon)=\sum^{n}_{p=i}g_{ip}(x_ p)- \epsilon_ i\geq 0;\quad i=1,...,m. \] Upper and lower bounds for the solution are obtained for a specific value of \(\epsilon\), using branch- and-bound methods, leading to a new value for \(\epsilon\), and the process is iterated until a desired tolerance level is attained.