Tuesday, 14:15 - 14:40 h, Room: MA 005


Antonio Morsi
Solving MINLPs on loosely coupled networks

Coauthors: Bjoern Geissler, Alexander Martin, Lars Schewe


Considering MINLPs defined on a network structure, such as nonlinearly-constrained network flow problems, we obtain dual bounds on the overall problem by a decomposition of the underlying graph into its biconnected and triconnected components and by the relaxation of the coupling constraints between these components. The dual bounds are further tightened by branching on violated nonconvex constraints. Branching candidates are obtained from an approximate primal solution to the master problem, which is solved by a bundle method. To solve the subproblems, in the case of factorable MINLPs, we use Chebyshev approximation to compute univariate piecewise linearizations (or piecewise polynomials) of the arising nonlinearities in advance. These approximations lead to MILP relaxations, or mixed integer polynomial relaxations, of the subproblems. We conclude with computational results of our approach for two real-world applications, water and gas network optimization.


Talk 3 of the invited session Tue.2.MA 005
"Advances in MINLP" [...]
Cluster 14
"Mixed-integer nonlinear programming" [...]


  The system of instant Virginia Payday Loans allows any adult U.S. citizen. But it is worth noting that these tests were carried out on the blood cells. Therefore, it's too early to say about scientific evidence of Viagra influence on blood clots.