📘 Large-Scale Integer Programming

A USTC course on optimization and reinforcement learning methods

📍 Summer 2025 · USTC East Campus Room 5205

👨‍🏫 Instructor: Prof. Lan Lu, Prof. Zhi Wang

🧑‍💻 Teaching Assistant: Long Chen (lchen1307@mail.ustc.edu.cn)


🔗 Quick Navigation

Overview · Slides · Quizzes · Assignments


🧭 Course Overview

This course introduces essential algorithmic techniques for solving large-scale problems in integer programming and reinforcement learning, taught by Prof. Lan Lu and Prof. Zhi Wang, respectively.

It covers modeling, LP relaxation, duality, and exact algorithms such as branch-and-bound and cutting planes, along with decomposition methods like column generation and Benders decomposition.

On the learning side, the course explores both classical reinforcement learning—such as dynamic programming and temporal-difference methods—and modern approaches including deep RL and recent developments involving large language models (LLMs).

Through a combination of theory and implementation, students will learn to build, analyze, and apply advanced optimization and learning models to real-world problems. Grades will be based on attendance, in-class quizzes, and homework assignments.


📚 Course Topics and Slides

CourseCatagoryTopicSlides
1IPFoundations of IP: Modeling and FormulationDownload
2IPFormulation Strategies and Logical ConstraintsDownload
3IPCompactness, Reformulation, and Geometry in IPDownload
4IPComputational MILP: Branch and BoundDownload
5IPDuality and Value Functions in MILPDownload
6IPRelaxations and Lagrangian DualityDownload
7IPBranch-and-Bound Search StrategiesDownload
8IPSeparation, Optimization, and Cutting-PlaneDownload
9IPCutting Plane GenerationDownload
10IPColumn Generation and ApplicationsDownload
11IPConstraint Decomposition in Integer ProgrammingDownload
12IPVariable Decomposition and Benders’ DecompositionDownload
13IPHeuristics and Neighborhood SearchDownload
14RLDynamic ProgrammingDownload
15RLMonte Carlo and Temporal-Difference LearningDownload
16RLIntroduction to Deep Reinforcement LearningDownload
17RLPolicy GradientsDownload
18RLAdvanced Policy GradientsDownload
19RLActor-Critic AlgorithmsDownload
20RLValue Function MethodsDownload
21RLDeep Q-LearningDownload

New slides and readings will be available by 7 PM each day.


📝 In-class Quizzes

Short quizzes will be conducted during lectures to reinforce key concepts and assess understanding.

DateQuiz TopicFormatPDFSolution
July 10Strong FormulationProofDownloadNone
July 11-1Parameter OptimizationCalculationDownloadNone
July 11-2Lagrangian RelaxationProofDownloadNone

📥 Assignments

There will be a few assignments:

HWTopicDue
NoneNoneNone

Assignments should be submitted as PDF or ZIP via the submission form before the deadline.


📌 For updates, resources, and slides, please check the top navigation bar regularly.