Optimization Basics Lab
Decision variables, objectives and constraints, feasible and optimal solutions, ill-structured problems and the search dilemma.
About the book
From toy problems to real-world solutions
Optimization is everywhere: in how ants find food, how parcels are routed, how airlines schedule crews and how cities manage traffic. Most real problems are ill-structured: large, dynamic, uncertain and without a neat mathematical model, so classic derivative-based solvers struggle. This book covers the algorithms built for them.
Each algorithm is introduced through its source of inspiration, pseudocode and parameters, then worked by hand on a small example before Python solves a scaled-up, real-world version. Routing, scheduling, assignment, packing, districting and control problems appear throughout.
Every chapter has Python notebooks in the book’s GitHub repository, ready to run in Google Colab, and an interactive lab to explore in the browser.
Contents
Five parts
Interactive labs
A lab for every chapter
Simulations of the book’s examples, experiments you run in the browser, links to the chapter notebooks in Colab, and from-scratch Python companions.
Problems and Algorithms Lab
Classify problems and algorithms, count search spaces and put heuristics to work.
Blind Search Lab
Breadth- and depth-first search, Dijkstra, uniform-cost and bidirectional search.
Informed Search Lab
Minimum spanning trees, hill climbing, beam search, A* and contraction hierarchies.
Simulated Annealing Lab
Acceptance probability, initial temperature and cooling schedules on functions and combinatorial problems.
Tabu Search Lab
Tabu structures, aspiration and tenure, and their effect on hard combinatorial problems.
Genetic Algorithm Lab
Selection, crossover and mutation for binary, real-valued and permutation encodings, plus multi-objective and adaptive GAs.
Particle Swarm Optimization Lab
Velocity updates, parameters and topologies for continuous, binary and permutation problems.
Ant Colony Optimization Lab
Stigmergy, transition rules and pheromone updates, and the main ACO variants.
Intro to ML Lab
Learning as optimization, the learning paradigms, and how ML helps solve combinatorial problems.
Unsupervised Learning Lab
Similarity measures, k-means, DBSCAN and self-organizing maps, including a SOM that solves a 194-city TSP.
Supervised Learning Lab
Neural networks, GCNs, attention, pointer networks and amortized optimization.
Reinforcement Learning Lab
MDPs, Q-learning, actor-critic on CartPole, PPO and multi-armed bandits.
Search and optimization libraries in Python
Python libraries for mathematical programming, graph search, metaheuristics and machine learning.
Benchmarks and datasets
Test functions, combinatorial benchmarks such as TSPLIB, and the datasets used in the book.
Exercises and solutions
End-of-chapter exercises with worked solutions as notebooks.
What readers say
Praise for the book
Elegantly covers modern optimization algorithms in depth and breadth. A great resource.
A treasure trove. Perfect for anyone looking to navigate the intricate world of optimization in real-world scenarios.
An exceptional book on the most prominent optimization algorithms.
This little gem introduces the reader to the algorithms and how you should apply them to real-world problems. A must-read!
Citation
Cite the book
Alaa Khamis, Optimization Algorithms: AI techniques for design, planning, and control problems. Manning Publications, ISBN 978-1633438835, 2024.
Found a mistake? See the errata.
@book{Khamis2024,
title = {Optimization Algorithms: AI techniques for design, planning, and control problems},
author = {Alaa Khamis},
year = {2024},
isbn = {978-1633438835},
publisher = {Manning Publications}
}