mitmath / juliacomputation

Repository for Common Ground C25

Julia: Solving Real-World Problems with Computation, Fall 2023

Announcement

Office Hours are Wednesdays 4:30pm-5:30pm in 2-135.

Logistics

MIT's numbering scheme gone nuts: (1.C25/6.C25/12.C25/16.C25/18.C25/22.C25)
This course is part of the Common Ground.

Lectures: Tuesdays & Thursdays 1-2:30 PM in room 2-142

Prerequisites: 6.100A, 18.03, 18.06 or equivalents (meaning some programming, dif eqs, and lin alg)

Instructors: A. Edelman, R. Ferrari, Y. Marzouk, P. Persson (UCB), S. Silvestri, J. Urschel, G. Dalle

Teaching Assistants: Jacob Hansen ([email protected]), Nicholas Klugman ([email protected])

Office Hours: Wednesdays 4:30pm-5:30pm in 2-135

Grading: Homeworks that may be spaced one or two weeks, to be submitted on canvas. No exams.
Lecture Recordings: Available on Canvas under the Panopto Video tab. Should be published the evening after each lecture.
Links: Worth bookmarking.

Piazza	Canvas	Julia	JuliaHub
Discussion	HW submission	Language	GPUs

Description

Focuses on algorithms and techniques for writing and using modern technical software in a job, lab, or research group environment that may consist of interdisciplinary teams, where performance may be critical, and where the software needs to be flexible and adaptable. Topics include automatic differentiation, matrix calculus, scientific machine learning, parallel and GPU computing, and performance optimization with introductory applications to climate science, economics, agent-based modeling, and other areas. Labs and projects focus on performant, readable, composable algorithms and software. Programming will be in Julia. Expects students have some familiarity with Python, Matlab, or R. No Julia experience necessary.

Counts as an elective for CEE students, an advanced subject (18.100 and higher) for Math students, an advanced elective for EECS students, and a computation restricted elective for NSE students. AeroAstro students can petition department to count this class as a professional subject in the computing area. (Professors may be open to petitioning for counting for other programs.)

Class is appropriate for those who enjoy math and wish to see math being used in modern contexts.

While not exactly the same as our past Computational Thinking Class... not entirely different either.

Course Objectives

1. Making mathematics your playground: By the end of the course, students will be able to write interactive Julia code that aids in their understanding of new mathematical systems or concepts.
2. Abstractions: By the end of the course students will be able to use the unique abstractions that exist in the Julia language to write code that can be part of a huge ecosystem. (By contrast many "one-off" homeworks in traditional courses are not able, by their very nature, to reveal the value of abstraction.)
3. Open Source and group collaborations: By the end of the course students will be able to participate in a larger open source project and also will have experienced how programming language can help break down barriers between areas making real the dream of scientific bilinguals that has been promoted at MIT. (See for example former MIT President Reif in the NYT back in 2018.)

Homeworks at a glance

Homework	Assigned	Due	Topic	Solution
HW0	Sep 7	Sep 14	Getting Started
HW1	Sep 14	Sep 21	Automatic Differentiation	HW1 solutions
HW2	Sep 21	Sep 28	Matrix Calculus	HW2 solutions
HW3	Sep 28	Oct 12	Julia Types	HW3 solutions
HW4	Oct 12	Oct 19	Seam Carving	HW4 solutions
HW5	Oct 19	Oct 26	Performance & Parallelism	HW5 solutions
HW6	Oct 26	Nov 2	Challenge Checkin 1
HW7	Nov 2	Nov 9	Survey and Climate ODEs	HW7 solutions
HW8	Nov 9	Nov 16	Challenge Problem Check-In 2
HW9	Nov 16	Nov 28	Matrix representations & atmospheric temperatures	HW9 solutions
HW10	Nov 28	Dec 5	Challenge Problem Final Submission
Challenge	Oct 17	Dec 5	Google Streetview car routing

Each student gets to turn in one homework late without justification. Further late turn ins must be justified. Remember: just because the automated tests succeed doesn't mean your code is 100% correct. It is a necessary, but not sufficient condition.

Homework solutions will be put online after each deadline.

Lectures at a glance

#	Day	Date	Lecturer	Topic	Slides / Notes	Notebooks
0				Julia tutorial	Cheat Sheets
1	R	9/7	Edelman	Intro to Julia		Intro to Julia, Tutorial, Hyperbolic Corgi, Images, Abstraction,
2	T	9/12	Edelman	Automatic Differentiation	autodiff handwritten notes	[AutoDiff jupyter notebook], [autodiff video]
3	R	9/14	Edelman	Automatic Differentiation		Reverse Mode AutoDiff Demo
4	T	9/19	Edelman	Matrix Calculus	Matrix Calc 1	Matrix Jacobians, Finite Differences
5	R	9/21	Edelman	Matrix Calculus	Matrix Calc 2	Linear Transformations, Symmetric Eigenproblems
6	T	9/26	Edelman	Differential Equations Lec 1		Time Stepping (background), ODEs and parameterized types (main topic), Resistors and Stencils (touched on this)
7	R	9/28	Edelman	Differential Equations Lec 2
8	T	10/3	Edelman	Imaging and Convolutions		Image Transformation notebook
9	R	10/5	Edelman	Imaging and Convolutions 2		Seam Carving notebook, Linear Transformations notebook
	T	10/10	Student Holiday
10	R	10/12	Edelman	HPC and GPUs	HPC and GPU Slides
11	T	10/17	Dalle	Package development		Challenge, Good practices
12	R	10/19	Dalle	Performance	Quiz	Package creation, Performance
13	T	10/24	Dalle	Graphs	Quiz	Graphs
14	R	10/26	Dalle	Linear programming	Quiz	Linear programming
15	T	10/31	Ferrari	Greenhouse Effect
16	R	11/2	Ferrari	Equilibrium and transient climate sensitivity		Earth's Temperature Model Mean Surface Temp Modeling
17	T	11/7	Drake	Economic Model of Climate	Slides	Economic Model, Optimization with JUMP
18	R	11/9	Edelman	Snowball Earth & Parallel/GPU computing		Snowball Earth and hysteresis
19	T	11/14	Persson	Mesh Generation	Mesh generation	Computational Geometry
20	R	11/16	Persson	Mesh Generation
21	T	11/21	Edelman	Floating-point Arithmetic
	R	11/23	Thanksgiving
22	T	11/28	Klugman	Fast inverse square root		Notebook
23	R	11/30	Silvestri	Climate Science		Solving the climate system
24	T	12/5	Silvestri	Climate Science
25	R	12/7	Edelman	Discrete and Continuous, are they so very different?
26	T	12/12	Class Party

juliacomputation's Issues

Improvements to HW1a (Pokémon)

• Invite students to use the documentation (? in the REPL / docs tab in Pluto)
• Present intro tutorial before the homework
• Say that just because checks are passing doesn't mean the answer is correct
• Add question numbers
• Explain what an empty struct means + how to compare mutable and immutable structs (== vs ===)
• Clarify inner / outer constructors
• Note on matrix indexing to prevent them from using linear indices
• Give a reference for various kinds of type dispatch syntax (f(x::X) vs f(x::Y) where {Y<:X})
• Specify expected output for attack(::Corgi, ::Pokemon), and state that we do not accept catch-all methods (code should rely on dispatch instead)
• Define what we mean by method ambiguity and explicitly lead students to find and resolve it
• Fix check_defense_philip check
• Explain how boundaries should be handled in the step! function
• Make it clear that the attacker cannot attack itself

Update installation instructions

The link to the installation instructions in Homework 0 is broken. Perhaps it would be nice to have a local page for this class regarding setup, instead of redirecting to the old "Computational thinking"

General remarks about the course

• Give a unifying perspective on the topics we teach and how they come together
• Be clearer on grading / lateness policy / other logistical aspects
• More structured slides