Julia: Solving Real-World Problems with Computation, Fall 2026
(course material work in progress)
|
|
Real-world problemsWe will take applications such as climate change and show how you can participate in the big open source community looking to find solutions to challenging problems with exposure to github and parallel computing. |
 |
Corgi in the washing machineYou will learn mathematical ideas by immersion into the mathematical process, performing experiments, seeing the connections, and seeing just how much fun math can be. |
 |
Revolutionary interactivityOur course material is built using real code, and instead of a book, we have a series of interactive notebooks. On our website, you can play with sliders, buttons and images to interact with our simulations. You can even go further, and modify and run any code on our website! |
 |
Learning Julia In literature it’s not enough to just know the technicalities of grammar. In music it’s not enough to learn the scales. The goal is to communicate experiences and emotions. For a computer scientist, it’s not enough to write a working program, the program should be written with beautiful high level abstractions that speak to your audience. This class will show you how. |
 |
Fall 2026 course page, under construction.
- Instructor: Alan Edelman
- Cross-listed as: 1.C25/6.C25/12.C25/16.C25/18.C25/22.C25
- Time: Mondays and Wednesdays, 2:30 PM to 4:00 PM
- Room: 45-102
- MIT listing: Department of Mathematics course list
Teaching Experiment
We are working with two core beliefs:
- Humans like to learn, at least when it is not too stressful.
- LLMs can help us leapfrog our learning.
We will use LLMs to learn something beyond our prerequisites. Perhaps some of you will learn general relativity, a sophisticated climate model, Lie algebra, or how a car does regenerative braking, but the key requirements are:
- The material is something you conventionally would not have the background for, and you are on the honor system here.
- You learn it, create a Julia notebook (LLMs allowed!), and teach it to all of us, seminar style.
For the rest of the class, we want you engaged and asking questions.
For a current example of this approach, see the Navier–Stokes in the news Pluto notebook.
Grading
Projects may be completed individually or in small groups, with each student responsible for understanding and presenting their part.
- Homework: 25%
- Two class presentations: 25%
- Class participation and questions: 15%
- Final Project Notebook: 35%
Homework 1
Homework 1: Getting Started and Multiple Dispatch was released Monday, September 14, 2026 and is due Monday, September 21, 2026 at 11:59pm Eastern time. A readable static HTML version will be generated automatically.
This assignment combines the setup and introductory Julia exercises from the former HW0 with the types and multiple-dispatch exercises from the former HW1.
Homework 2
Homework 2: Choose a reach project is the first step toward the final project. Choose an ambitious topic, explain why it is a reach for you, begin asking questions and learning, and create a first Julia notebook with an interactive element when appropriate. LLMs are allowed, but students may also work without them. This assignment will be refined as the projects develop. A readable static HTML version will be generated automatically.
The first homework gets everyone set up and introduces the Julia ideas needed for the projects later in the semester. The project proposal will follow separately.
Homework 3
Homework 3: Automatic Differentiation introduces dual
numbers, forward- and reverse-mode differentiation, and Jacobians. It is due
Monday, October 19, 2026 at 11:59 PM EDT. A readable static HTML
version
will be generated automatically.
Homework 4
Homework 4: How Are We Learning? asks you to reflect honestly on how the course is working, investigate one gap in your understanding, and propose concrete improvements to the course and your own learning. It is due Monday, November 2, 2026 at 11:59 PM EST.
Homework 5
Homework 5: Understanding Something Deeply builds on Homework 4: choose one part of your presentation to explain deeply and make a serious start. This prepares you for a 10–12 minute end-of-semester presentation, with additional time for questions. The homework is due Monday, November 23, 2026 at 11:59 PM EST.
Calendar
Dates follow the MIT Registrar calendar.
Lecture 1: Communicate with Computation
Lecture 2: Julia and Navier-Stokes
Lecture 3: Julia speed, random variables as types
Lecture 4: John Urschel on floating point formats
Lecture 5: Automatic differentiation
Lecture 6: Reverse mode automatic differentiation
Lecture 7: Ranking teams with linear algebra
This notebook expects accompanying team_data.csv and game_data.csv files.
Lecture 8: Applied math in the age of AI and TOP500 supercomputers
October 14: General Circulation Models — Logan Reich
October 19: Option pricing — Kathleen Zhang and Krishna Parvataneni
October 19: Kepler: Harmony to Chaos — Favio Lesnjak
October 21: Fallout by Height — Thomas Shoales
October 26: The Phase Problem in Crystallography — Mary Feliz and Drew Hussar
October 26: How a balanced ocean makes eddies — Rochelle Griffith
October 28: How many GPUs should I rent? — Jhanny Jimenez
October 28: Is That My Hand? — Yusef Hosein and Lexcea King
November 2: TCP Congestion Control — An Dinh
November 2: Why Creditors Accept Less — Maximo Del Pino
November 4: Ball Trajectories in Baseball — Nathan Zhu
Previous Offering