Data Structures & Functional Programming: A Comprehensive Course Overview, Schemes and Mind Maps of Data Structures and Algorithms

Download Data Structures & Functional Programming: A Comprehensive Course Overview and more Schemes and Mind Maps Data Structures and Algorithms in PDF only on Docsity!

http://www.flickr.com/photos/rofi/2097239111/

Data Structures and

Functional Programming

Course Overview

Nate Foster

Cornell University

Spring 2012

Course staff

Instructor: Nate Foster

• Joined Cornell last year from UPenn
• Research area: programming languages
• Functional programmer since 1998

TAs: Shrutarshi Basu (coordinator), Ashir Amer,

Stuart Davis, Gautam Kamath, Katie Meusling,

Greg Zecchinni

Consultants: many

You have a large and veteran staff. Make use of them!

Office hours in Upson 360 Sunday-Thursday from 7-9pm

Additional office hours Thursday from 5-7pm

Course web site

http://www.cs.cornell.edu/Courses/cs

• Course material
• Homework
• Announcements Includes a complete set of course notes  Nearest equivalent to a textbook  But the lectures and sections are definitive Links to lecture notes will go live shortly after lecture Goal is to help, not replace attendance!

Piazza and CMS

 Online discussion forum  Monitored by TAs/consultants  Ask for help, but don’t post solutions CMS  “Course Management System”  Built by Andrew Myers (with help from lots of students)  Assignments and grades posted here

Grading

Breakdown:

• 45% - Problem sets
• 5% - Quizzes (lowest dropped)
• 30% - Preliminary exams (lower exam weighted less)
• 20% - Final exam Will follow the usual CS3110 curve
• Centered around a B/B+

Late policy

You can hand it in until we start grading

• 15% penalty / day
• After we start grading, no credit Save your code and submit early and often
• CMS is your friend
• Be certain you have submitted something, even if it isn’t perfect and you are improving it

If you have a emergency (e.g., medical, family) talk to

Nate before the last second

What this course is about

Programming isn’t hard Programming well is very hard

• Programmers vary greatly
• 10X or more difference in skills We want you to write code that is:
• Reliable, efficient, readable, testable, provable, maintainable… beautiful! Expand your problem-solving skills
• Recognize problems and map them onto the right abstractions and algorithms

Thinking versus typing

“A year at the lab bench saves an hour at the library” Fact: there are an infinite number of incorrect programs Corollary: making random tweaks to your code is unlikely to help

• If you find yourself changing “<“ to “<=“ in the hopes that your code will work, you’re in trouble Lesson: think before you type! http://www.flickr.com/photos/tmartin/32010732/

Rule

Good programmers are lazy

• Never write the same code twice (why?)
• Reuse libraries (why?)
• Keep interfaces small and simple (why?) Pick a language that makes it easy to write the code you need
• Early emphasis on speed is a disaster (why?) Rapid prototyping!

Main goal of CS

Master key linguistic abstractions:

• Procedural abstraction
• Control: iteration, recursion, pattern matching, laziness, exceptions, events
• Encapsulation: closures, ADTs
• Parameterization: higher-order procedures, modules Mostly in service to rule # Transcends individual programming languages

Choice of language

This matters less than you suspect Must be able to learn new languages

• This is relatively easy if you understand programming models and paradigms We will be using OCaml, a dialect of ML Why use yet another language?
• Not to mention an obscure one? Main answer: OCaml programs are easy to reason about

Why?

Awesome OCaml feature: many common errors simply impossible  More precisely, they are caught at compile time  Early failure is very important (why?) Functional language  Programs have a clear semantics  Heavy use of recursion  Lots of higher-order functions  Few side effects Statically typed and type safe  Many bugs caught by compiler

Functional Style

Idea: program without side effects

• Effect of a function is only to return a result value Program is an expression that can be evaluated to produce a value
• For example, evaluating 2+2 yields 4
• Just like mathematical expressions Enables equational reasoning about programs:
• if x equals y, replacing y with x has no effect:
• let x=f(0) in x+x equivalent to f(0)+f(0)

Functional Style

Bind variables to values, don’t mutate existing variables No concept of x=x+1 or x++ These do nothing remotely like x++ let x = x+1 in x let rec x = x+1 in x The former assumes an existing binding for x and creates a new one (no modification of x ) The latter is an invalid expression