Revolutionizing Combinatorial Optimization: The Free-Energy Machine
In a groundbreaking study, researchers unveil the Free-Energy Machine (FEM), a novel approach that combines statistical physics with machine learning to tackle complex combinatorial optimization probl...
https://news.lavx.hu/article/revolutionizing-combinatorial-optimization-the-free-energy-machine
My colleagues at TU Delft are seeking to hire a postdoc to work on Applied Planning and Scheduling under Uncertainty, with applications in modelling supply chain scenarios for offshore wind farm installation: https://careers.tudelft.nl/job/Delft-Postdoc-in-Applied-Planning-and-Scheduling-under-Uncertainty-2628-CD/814890902/
A benchmark for #quantum optimization: the #traveling salesman #by Richard H. Warren
10.26421/QIC21.7-8-2
Exciting update! My Network Algorithms and Approximations course continues with Lessons 2, 3 & 4 now available! 
Topics covered:
Submodular (Set) Cover – Greedy log-approximation & Group Steiner Tree Maximum Coverage – 1 - 1/e approx, LP relaxations & budgeted coverage Unique Coverage – Log(n)-approximation, NP-hardness & max-cut ties
Watch now:
Lesson 2: https://youtu.be/xi6P3bqy61g Lesson 3: https://youtu.be/jC44JdD74Hw Lesson 4: https://youtu.be/ypzFnl0Wfp4
New lectures premiere every Wednesday at 7PM ET!
Full playlist: https://www.youtube.com/playlist?list=PLx7SjCaKZzEIeJxOlTuXveAE5eY7WOYB9
Subscribe for more: https://www.youtube.com/@hajiaghayi
2/n Very much like this visualisation presented by Frank Phillipson (based on a fig from Caceres-Cruz et al., 2014) of solvers for optimisation problems. Might try to use this or something like it in the course on algorithm for NP-hard problems that I am involved in.
Fully Programmable Spatial Photonic Ising Machine by Focal Plane Division
https://arxiv.org/abs/2410.10689
#IsingMaching #SpinGlasses #CombinatorialOptimization #Photonics #Optics
arXiv:2410.10689
Ising machines are an emerging class of hardware that promises ultrafast and energy-efficient solutions to NP-hard combinatorial optimization problems. Spatial photonic Ising machines (SPIMs) exploit optical computing in free space to accelerate the computation, showcasing parallelism, scalability, and low power consumption. However, current SPIMs can implement only a restricted class of problems. This partial programmability is a critical limitation that hampers their benchmark. Achieving full programmability of the device while preserving its scalability is an open challenge. Here, we report a fully programmable SPIM achieved through a novel operation method based on the division of the focal plane. In our scheme, a general Ising problem is decomposed into a set of Mattis Hamiltonians, whose energies are simultaneously computed optically by measuring the intensity on different regions of the camera sensor. Exploiting this concept, we experimentally demonstrate the computation with high success probability of ground-state solutions of up to 32-spin Ising models on unweighted maximum cut graphs with and without ferromagnetic bias. Simulations of the hardware prove a favorable scaling of the accuracy with the number of spins. Our fully programmable SPIM enables the implementation of many quadratic unconstrained binary optimization problems, further establishing SPIMs as a leading paradigm in non von Neumann hardware.
GHOST v3 is released! It now integrates a complete solver searching for all possible solutions of your combinatorial problems.
The interface also slightly changed, but you need to change one method name only in your current code: Solver::solve becomes Solver::fast_search.
Some functionalities remain to be implemented, like the parallelization of this complete solver. Stay tuned!
https://github.com/richoux/GHOST
#ConstraintProgramming #CombinatorialOptimization #AI #Framework
@fizise @enorouzi @lysander07 @amszmidt and as a “bonus”, this stop of funding happened when I finished my PhD thesis about #DefeasibleReasoning. There was not much of a perspective to continue academic research on this topic at that time and I continued working in industry on #CombinatorialOptimization with a mix of #AI and #OR methods. So I felt quite impacted by this.
#7AYW #Day3 #CombinatorialOptimization #MachineLearning
Nuria Gómez-Vargas @justnuu_ shows a predict-and-optimize approach to guide the training of ML models with performances on the optimization problem and to enhance sparsity in the feature space for decisions explainability. https://t.co/8e0XTTeQPh
#7AYW #Day3 #CombinatorialOptimization #MachineLearning
Léo Baty presents a policy for the dynamic VRPTW,
which ranked first of the competition @EuroNeuripsVRP. It relies on a #DeepLearning pipeline with a prize collecting VRPTW combinatorial optimization layer. https://t.co/ZIze00SGf4
#7AYW #Day1 #CombinatorialOptimization #MachineLearning
Francesco Paolo Saccomanno proposes a reinforcement learning strategy to address #BinPacking, where the agent is trained to imitate a classic heuristic, the “best fit” strategy. https://t.co/9SnLnX4CW8
#7AYW #CombinatorialOptimization & #MachineLearning
Antonio Consolo studies a variant of multivariate randomized regression trees and presents a decomposition training algorithm with a heuristic for the reassignment of the input vectors along the branching nodes of the tree.
#7AYW #CombinatorialOptimization & #MachineLearning
Simone Milanesi introduces the BeMi ensemble, a structured architecture of Binarized Neural Networks based on training a single BNN for each possible pair of classes and applying a Condorcet-inspired majority voting scheme.
How often should you clean your room?
This is the title of this #CombinatorialOptimization #paper. The take on the question is truly original.
https://arxiv.org/abs/1305.1984
Authors went at length to investigate the issue and prove a logarithmic upper bound.
#Combinatorialoptimization - a field of research addressing problems that feature strongly in a wealth of practical and industrial contexts - has been identified as one of the core potential fields of applicability of near-term #quantumcomputers. It is still unclear, however, to what extent variational #quantumalgorithms can actually outperform classical algorithms for this type of problems.
