On Lagrangian duality gap of quadratic fractional programming with a two-sided quadratic constraint (Q2174902): Difference between revisions
From MaRDI portal
Created claim: Wikidata QID (P12): Q129316328, #quickstatements; #temporary_batch_1724922362815 |
Normalize DOI. |
||
Property / DOI | |||
Property / DOI: 10.1007/s11590-018-1320-4 / rank | |||
Property / DOI | |||
Property / DOI: 10.1007/S11590-018-1320-4 / rank | |||
Normal rank |
Latest revision as of 08:45, 17 December 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | On Lagrangian duality gap of quadratic fractional programming with a two-sided quadratic constraint |
scientific article |
Statements
On Lagrangian duality gap of quadratic fractional programming with a two-sided quadratic constraint (English)
0 references
27 April 2020
0 references
For the minimization of the ratio of two quadratic functions under a two-sided quadratic constraint, assuming the Slater condition and that the denominator of the objective function is bounded from below by a strictly positive number on the feasible set, the authors present a semi-definite programming reformulation. Using this reformulation, they prove a strong Lagrangian duality theorem for a scaled version of the original problem obtained by replacing the quadratic constraints by the ones resulting from dividing them by the denominator of the objective function. To illustrate that strong duality may fail without rescaling, they examine a variant of a problem considered in [\textit{A. Beck} et al., SIAM J. Matrix Anal. Appl. 28, No. 2, 425--445 (2006; Zbl 1115.65065)], namely the minimization of \(\frac{\left\Vert Ax-b\right\Vert ^{2}}{\left\Vert x\right\Vert ^{2}+1}\) subject to \(\alpha \leq \left\Vert x\right\Vert ^{2}\leq \beta \), for which they find a necessary and sufficient condition for the duality gap to be strictly positive.
0 references
fractional programming
0 references
quadratic programming
0 references
Lagrangian duality
0 references
total least squares
0 references
0 references
0 references
0 references