[77] Data-Driven Mathematical Optimization in Pyomo (Jeffrey C Kantor)

Join our Meetup group:
www.meetup.com/data-umbrella
Jeffrey C Kantor: Data-Driven Mathematical Optimization in Pyomo
Resources
- Pyomo on GitHub: github.com/Pyomo/pyomo
- Book (Data-Driven Mathematical Optimization in Python): mobook.github.io/MO-book/intr...
- Slides: docs.google.com/presentation/...
About the Event
The seminar introduces Pyomo, an open-source modelling language for mathematical optimization problems in Python. Pyomo is powerful, flexible, open-source, easy to learn, and compatible with other Python packages. It allows users to formulate and solve various optimization problems, including linear programming, nonlinear programming, mixed-integer programming, and many others. This seminar will introduce a collection of fifty Jupyter notebooks demonstrating a wide range of data-driven applications for optimization in Python.
Timestamps
00:00 Data Umbrella introduction
03:39 Introduce Jeffrey, the speaker
04:54 Jeffrey begins
05:33 What is Pyomo?
06:32 Some team members behind Pyomo: Krzysztof Postek, Alessandro Zocca, Joaquim Gromicho
07:28 What is mathematical optimization? compared to machine learning?
07:55 Data Science / Machine Learning / Optimization
10:00 Types of objectives: Physical, Financial, Information
11:38 Types of decision variables: continuous, discrete, true/false
13:53 Types of constraints
15:42 NEOS family tree of optimization problems
18:17 Why Pyomo? (PYthon Optimization Modeling Objects p-y-o-m-o) (history and features of pyomo)
24:15 An example of going from a business problem to a solution using Pyomo: how much of product X and Y to produce to maximize profitability?
27:28 Convert a mathematical model to a pyomo model
29:30 Pyomo model + Solver .... Solution
30:58 Overview of the Pyomo workflow
33:01 Applications of Pyomo
33:16 Disjunctive programming ... "either" / "or" decisions
36:04 GDP Transformation (Generalized Disjunctive Programming)
39:20 Example problem: Strip Packing (pack shapes into economical arrangements, such as shelves, boxes)
41:00 Math model with disjunctions
42:53 Pyomo parameters and sets ... "Data Driven"
44:31 Indexing constraints
45:23 Strip packing example solution
46:21 Cryptocurrency Arbitrage
48:44 Pooling and blending ..... Nonconvex programming
51:29 online book "Data-Driven Mathematical Optimization in Python": mobook.github.io/MO-book/intr...
52:24 Q&A
53:24 Q: Amazon use these techniques for their packaging?
54:35 Q: Can this be linked to quantum computing?
56:05 Q: Can you recommend a good framework book on optimization?
58:15 Q: What are some of the challenging problems you have solved in industry?
01:01:45 Q: How was the performance of Pyomo comparison with Jump?
01:03:50 Supply chains / optimization
About the Speaker
Jeffrey Kantor is a Professor of Chemical and Biomolecular Engineering at the University of Notre Dame. Professor Kantor does research and teaches in the broad area of systems control and optimization, applying these concepts to many different applications ranging from chemical processes, energy systems, finance, and the control of complex natural watersheds. Professor Kantor received a Bachelor’s degree in Chemical Engineering from the University of Minnesota, a Masters's and a PhD degree from Princeton University, and a post-doc in the Chemistry Department at the University of Tel Aviv before joining Notre Dame. He has previously held administrative appointments at Notre Dame, including Dean of the Graduate School, Vice President for Research, Vice President and Associate Provost at Notre Dame, and visiting positions at Imperial College and Princeton University.
- LinkedIn: / jeffrey-kantor-7a1ab3a
- GitHub: github.com/jckantor
Obituary:
- Notre Dame: news.nd.edu/news/in-memoriam-...
- Legacy: www.legacy.com/us/obituaries/...
- Voyageurs: www.voyageurs.org/kantor
#python #optimization #optimizationtechniques

[@KheraShanu]

having less views doesn't mean the work you do has less impact, thanks for this. Really awesome video. Prof was really good at explaining this stuff!

[@DataUmbrella]

Glad you enjoyed it.