Generalized goal programming: polynomial methods and applications
read more
Citations
Steepest descent methods for multicriteria optimization
Gap-free computation of pareto-points by quadratic scalarizations
Distance rationalization of voting rules
Dominating Sets for Convex Functions with some Applications
References
Numerical Methods for Least Squares Problems
Convex analysis and minimization algorithms
Optimal Estimation of Executive Compensation by Linear Programming
Facility Location: A Survey of Applications and Methods
Related Papers (5)
Frequently Asked Questions (5)
Q2. What is the feasible region of the gauge?
The feasible region K is then defined byK = { x = (xf )f∈F ∈ (IRn)F | xf = af for all f ∈ A } , (10)while the gauge γ is taken as the l1 norm, so that one minimizes the sum of all interactions between the facilities,inf xf=af ∀f∈A ∑ (f,g)∈E γ(f,g)(xf − xg).
Q3. what is the self-concordant barrier for b?
to construct a self-concordant barrier for the set B̃, one can use selfconcordant barriers b+i with self-concordancy parameter ϑ + i for the cones epi(γi) and a self-concordant barrier b̃ with self-concordancy parameter ϑ̃ for the unit ball of γ̃ to defineb̂(u1, t1, . . . , uk, tk) := b̃(t1, . . . , tk) + k∑ i=1 b+i (ui, ti),a self-concordant barrier for B̂ with self-concordancy parameter ϑ̃+ ∑k i=1 ϑ + i .
Q4. what is the optimal solution for (P)?
In other words, x̄ is optimal for (P ) iff there exists some ū ∈ IRm satisfyingC>ū+ d ∈ E∗, (28)γ◦(ū) ≤ 1, (29)γ(Cx̄+ c) + d>x̄ = ū>(Cx̄+ c) + d>x̄, (30)= ū>c+ inf xM∈M,xE∈E ū>C(xM + xE) + d >(xM + xE). (31)But (30) holds iffCx̄+ c ∈ NB◦(ū).
Q5. what is the standard logarithmic barrier for the polytope B?
Of course, the standard logarithmic barrier bB(x) = − ∑k i=1 ln(gi − a>i x) for the polytope B can be used todefine b+B(x, t) = − ∑k i=1 ln(git−a>i x), a self-concordant barrier for the epigraph of γ with self-concordancy parameter ϑ+B = k. 2Example 10 Let γ be a gauge, A ∈ IRn×n be a regular matrix and c ∈ IRn be a vector with γ◦(A>c) < 1. Then γ̃(x) := γ(Ax) + c>x defines a gauge [35].