Marouane Felloussi
@marouanefl.bsky.social
phd candidate in discrete optimization at mines st-étienne/univ. clermont auvergne
https://marouane-f.github.io
https://marouane-f.github.io
A special issue of the INFORMS Journal on Optimization is soon open for submissions.
July 3, 2025 at 4:23 PM
A special issue of the INFORMS Journal on Optimization is soon open for submissions.
And that's a wrap on #EuroMIP25! Three packed days of great talks and inspiring research.
Huge thanks to @sophie.huiberts.me , @matbesancon.bsky.social, the rest of the MIP committee and the local team for making this first edition a real success!
Huge thanks to @sophie.huiberts.me , @matbesancon.bsky.social, the rest of the MIP committee and the local team for making this first edition a real success!
July 3, 2025 at 4:22 PM
And that's a wrap on #EuroMIP25! Three packed days of great talks and inspiring research.
Huge thanks to @sophie.huiberts.me , @matbesancon.bsky.social, the rest of the MIP committee and the local team for making this first edition a real success!
Huge thanks to @sophie.huiberts.me , @matbesancon.bsky.social, the rest of the MIP committee and the local team for making this first edition a real success!
The talk wraps up with future directions, including a Branch-and-Price implementation to handle integrality, and the potential of combining Simplicial and Dantzig-Wolfe decompositions. Lucas also briefly mentions leveraging quantum optimization for tackling QUBO pricing subproblems.
July 3, 2025 at 4:11 PM
The talk wraps up with future directions, including a Branch-and-Price implementation to handle integrality, and the potential of combining Simplicial and Dantzig-Wolfe decompositions. Lucas also briefly mentions leveraging quantum optimization for tackling QUBO pricing subproblems.
Lucas also draws nice connections to graph theory, where these structural links help identify when and how a problem can be broken down effectively.
July 3, 2025 at 4:07 PM
Lucas also draws nice connections to graph theory, where these structural links help identify when and how a problem can be broken down effectively.
Finally, Lucas hints at the promise of block decompositions when the problem allows it. Numerical evidence suggests that when the model breaks into a handful of blocks (typically less than five), block-wise decomposition strategies can be effective.
July 3, 2025 at 4:06 PM
Finally, Lucas hints at the promise of block decompositions when the problem allows it. Numerical evidence suggests that when the model breaks into a handful of blocks (typically less than five), block-wise decomposition strategies can be effective.
A key takeaway is that injecting quadratic constraints in the pricing leads to better dual bounds but a harder pricing subproblem (as is the case for general DW in MIP).
July 3, 2025 at 4:04 PM
A key takeaway is that injecting quadratic constraints in the pricing leads to better dual bounds but a harder pricing subproblem (as is the case for general DW in MIP).
The focus then shifts to non-convex binary quadratic problems, with reformulations that blend Dantzig-Wolfe decomposition with convex quadratic reformulations.
July 3, 2025 at 4:03 PM
The focus then shifts to non-convex binary quadratic problems, with reformulations that blend Dantzig-Wolfe decomposition with convex quadratic reformulations.
Lucas begins with a simplicial decomposition framework inspired by Carathéodory’s theorem, one that doesn’t rely on duality. Several strategies are proposed for handling the master, taking ideas from the Frank-Wolfe method and enhanced with cutting and sifting techniques for the pricing step.
July 3, 2025 at 3:57 PM
Lucas begins with a simplicial decomposition framework inspired by Carathéodory’s theorem, one that doesn’t rely on duality. Several strategies are proposed for handling the master, taking ideas from the Frank-Wolfe method and enhanced with cutting and sifting techniques for the pricing step.
Lucas Létocart’s wraps up with a dive into Decomposition Methods for Quadratic Programming problems, both convex and non-convex.
#EuroMIP25 #orms
#EuroMIP25 #orms
July 3, 2025 at 3:55 PM
Lucas Létocart’s wraps up with a dive into Decomposition Methods for Quadratic Programming problems, both convex and non-convex.
#EuroMIP25 #orms
#EuroMIP25 #orms
The talk concludes with a discussion on refining the bounds and exploring related problems.
July 3, 2025 at 3:49 PM
The talk concludes with a discussion on refining the bounds and exploring related problems.
The proofs/extensions to higher dimensions use clever arguments involving tiling and pigeonhole principles.
July 3, 2025 at 3:47 PM
The proofs/extensions to higher dimensions use clever arguments involving tiling and pigeonhole principles.
Second, when no exact sparse solution exists, approximate solutions can still be found, with feasibility errors shrinking exponentially as sparsity increases. By allowing more nonzeros, one gets exponentially better approximations.
July 3, 2025 at 3:45 PM
Second, when no exact sparse solution exists, approximate solutions can still be found, with feasibility errors shrinking exponentially as sparsity increases. By allowing more nonzeros, one gets exponentially better approximations.
Stefan presents two main results: first, a new upper bound with at most m+f(Δ) nonzeros, where f(Δ) depends only on the matrix’s subdeterminant. A nice geometric interpretation is given.
July 3, 2025 at 3:42 PM
Stefan presents two main results: first, a new upper bound with at most m+f(Δ) nonzeros, where f(Δ) depends only on the matrix’s subdeterminant. A nice geometric interpretation is given.
Stefan Kuhlmann’s talk focused on Sparse Integer Solutions and Approximations. This is an important topic both theoretically and in applications where nonzero entries can represent costly decisions or changes.
#EuroMIP25 #orms
#EuroMIP25 #orms
July 3, 2025 at 3:39 PM
Stefan Kuhlmann’s talk focused on Sparse Integer Solutions and Approximations. This is an important topic both theoretically and in applications where nonzero entries can represent costly decisions or changes.
#EuroMIP25 #orms
#EuroMIP25 #orms
Sophie’s block‑coordination finds first feasible solutions faster than existing methods.
She then wraps up by outlining future directions regarding convergence and other applications, while also emphasizing the role of ethical OR, tying back to Nadia Brauner’s opening talk.
She then wraps up by outlining future directions regarding convergence and other applications, while also emphasizing the role of ethical OR, tying back to Nadia Brauner’s opening talk.
July 3, 2025 at 3:33 PM
Sophie’s block‑coordination finds first feasible solutions faster than existing methods.
She then wraps up by outlining future directions regarding convergence and other applications, while also emphasizing the role of ethical OR, tying back to Nadia Brauner’s opening talk.
She then wraps up by outlining future directions regarding convergence and other applications, while also emphasizing the role of ethical OR, tying back to Nadia Brauner’s opening talk.
Load shifting in water networks is hard even for obtaining a feasible solution. Instead of working on each pump, the method shifts focus to water‑tank levels (for sensitivity reasons) and penalizes coupling constraints between time blocks.
July 3, 2025 at 3:29 PM
Load shifting in water networks is hard even for obtaining a feasible solution. Instead of working on each pump, the method shifts focus to water‑tank levels (for sensitivity reasons) and penalizes coupling constraints between time blocks.
Drinking‑water networks add frictional resistance and discharge pressure. For fixed node potentials, computing flows is easy, but optimizing is hard.
Sophie casts the deterministic supply–demand problem as a hidden bilevel model via strong duality.
Sophie casts the deterministic supply–demand problem as a hidden bilevel model via strong duality.
July 3, 2025 at 3:26 PM
Drinking‑water networks add frictional resistance and discharge pressure. For fixed node potentials, computing flows is easy, but optimizing is hard.
Sophie casts the deterministic supply–demand problem as a hidden bilevel model via strong duality.
Sophie casts the deterministic supply–demand problem as a hidden bilevel model via strong duality.
Fenchel duality shows how dual inequalities (and equality conditions) help derive stationarity.
July 3, 2025 at 3:23 PM
Fenchel duality shows how dual inequalities (and equality conditions) help derive stationarity.
Steady‑state equilibrium follows from network properties, assumptions, and KKT conditions. Sophie presents three reformulations via strong duality and KKT N&S that inject convexity into subproblems.
July 3, 2025 at 3:21 PM
Steady‑state equilibrium follows from network properties, assumptions, and KKT conditions. Sophie presents three reformulations via strong duality and KKT N&S that inject convexity into subproblems.
Nonconvex MINLP with sequence and space dependencies makes every decision interconnected (pumps, pipes, tanks etc.). Rather than solving end‑to‑end, the goal is to find one feasible solution by disconnecting the problem into scalable and solvable parts.
July 3, 2025 at 3:18 PM
Nonconvex MINLP with sequence and space dependencies makes every decision interconnected (pumps, pipes, tanks etc.). Rather than solving end‑to‑end, the goal is to find one feasible solution by disconnecting the problem into scalable and solvable parts.
Sophie Demassey @sofdem.bsky.social kicks off the third and final day of #EuroMIP25 with a talk on Block Coordination of Nonlinear Network and Discrete Optimization. She focuses on scheduling nonlinear flow networks, with an application to load shifting in drinking water systems.
#EuroMIP25 #orms
#EuroMIP25 #orms
July 3, 2025 at 3:11 PM
Sophie Demassey @sofdem.bsky.social kicks off the third and final day of #EuroMIP25 with a talk on Block Coordination of Nonlinear Network and Discrete Optimization. She focuses on scheduling nonlinear flow networks, with an application to load shifting in drinking water systems.
#EuroMIP25 #orms
#EuroMIP25 #orms
He concludes the talk by addressing ongoing challenges related to degeneracy.
July 3, 2025 at 12:10 AM
He concludes the talk by addressing ongoing challenges related to degeneracy.
The resulting hybridization combines inner and outer approximations of the polyhedral sets. François shares practical insights on applying their method effectively.
July 3, 2025 at 12:09 AM
The resulting hybridization combines inner and outer approximations of the polyhedral sets. François shares practical insights on applying their method effectively.
After a brief introduction of the two decomposition frameworks, François highlights the unboundedness of the solution space for the Fenchel subproblem, thus the need for normalization.
July 3, 2025 at 12:07 AM
After a brief introduction of the two decomposition frameworks, François highlights the unboundedness of the solution space for the Fenchel subproblem, thus the need for normalization.
François Lamothe explores strengthening linear relaxations for MIPs by combining Dantzig-Wolfe and Fenchel decompositions.
#EuroMIP25 #orms
#EuroMIP25 #orms
July 3, 2025 at 12:02 AM
François Lamothe explores strengthening linear relaxations for MIPs by combining Dantzig-Wolfe and Fenchel decompositions.
#EuroMIP25 #orms
#EuroMIP25 #orms