@Orly01@mmonaci74 Well, we gave the conjecture, you have to (dis)prove it! Possible features include the obj. value (and the degree of infeasibility if you use the simplex dual), the degree of degeneracy, etc. My #AI friends would even try a #DNN to self-drive along the polyhedron edges
… including easy-to-separate #Benders cuts from linear reformulations of quadratic terms (including McCormick with triangle inequalities). Have a look if you interested in quadratic #MIPs#orms
This remembers me an old idea of mine of using this as a crossover operation in #genetic algorithms for #orms using an autoencoder to compute the latent space of the feasible solutions of a combination optimization problem such as #VRP. Worth a try?
Watching a hybrid talk by @JeffLinderoth on subspace clustering via integer programming. Benders decomposition, column generation, all of our favourites are there; such a nice work by @JeffLinderoth @jim_luedtke Daniel and Akhilesh! #KnightsOfMIP
Watching a hybrid talk by @JeffLinderoth on subspace clustering via integer programming. Benders decomposition, column generation, all of our favourites are there; such a nice work by @JeffLinderoth @jim_luedtke Daniel and Akhilesh! #KnightsOfMIP
Here is a video of the talk I gave this week at the “Optimization without Borders” conference (Sochi, Russia), organized in honor of the 65th birthday of Yurii Nesterov” https://t.co/0hUxsODLMi
My paper on the Benders' decomposition framework in SCIP has been accepted and now available online. This paper aims to be a guide for future researchers implementing the branch-and-cut approach for Benders' decomposition. https://t.co/foeXn0FGn4
Quarantine may show a new side of you ;) After 2+ weeks at home alone, I started this YouTube channel of animations for kids. Are magical things happening also in your house? https://t.co/ujremfoxoj
#fun#sapo#kids#stayathome#iorestoacasa
In their paper, @pononomo, @dominiqs81 and @AndreaOmarT explain how the special ingredients added by the #CPLEX team to their #Benders implementation moved it from 1.5x slower to 5x faster than default MIP! And it's automatic!
https://t.co/pj7qjFut0d
#orms
The presentations made by the #CPLEX development team at #INFORMS2019 are all at https://t.co/1zG5RnDADC. You’ll read about the upcoming 12.10 release, Benders, #MachineLearning, Penalty Variables, infeasible MIPs and the Generic Callback. #orms