Held: Tuesday, 21 April 2015

Back to Outline 44 - Association Lists. On to Outline 46 - Binary Search Lab.

Summary

We consider the general problem of searching and binary search, one of the most efficient algorithms for searching.

Related Pages

• Reading: Searching
• Lab: Binary Search
• Morning EBoard 45 (Source a.m. 45)
• Afternoon EBoard 45 (Source p.m. 45)

Overview

• The problem of searching.
• Analyzing algorithmic efficiency.
• Making searching more efficient using divide-and-conquer.
• A demo: Destructive binary search.
• Considering the parameters to binary search.

Administrivia

• Continue partners!
• Review sessions Thursday.
• For office hours this week, check http://rebelsky.youcanbook.me
• 5-10 min admin, 5-10 min project, 15-20 min sorting, remainder lab.
• When you send me email questions of the form "this doesn't work",
  • Send me all the code I will need in order to run examples, including any utilities you are using.
  • Send me the expression that you are using that's not working.
  • Explain to me what's happening.

Upcoming Work

• Project images due today.
• Projects due next Tuesday.
• Lab writeup: http://bit.ly/151-2015S-lab44
• Reading for Wednesday: Review Search Algorithms

Extra Credit Opportunities

Academic

• Convocation April 22nd, Erica Lehrer '92 (Post-Conflict Memory, Ethnography)
• CS Talks Wednesday:

Peer Support (Morning Section)

• KY's radio show, "We Think We're Funny", 9-10pm Mondays.
• Julia's radio show, "The Hot Box". Wednesday night/Thursday morning 1:00-2:00 a.m.

Peer Support (Afternoon Section)

• Contra Dance April 24 in Main Hall.
• Financial Literacy workshop Monday night 7-8 in JRC 226.

Miscellaneous

• Town hall April 23, noon or 7:30 pm, "How we have conversations."

Other Good Things (no extra credit)

• Contra Dance April 24 in Main Hall.

Project Tips

Image-making techniques

• Drawings as values, with or without recursion
• image-variant, `image-transform!
  • Also (image-redo! image (lambda (col row color) ...)
  • image-transform! and image-redo! behave differently with selections.
• image-compute
• Gimp tools, with or without recursion
• Turtle graphics

I've added a variety of project tips that I thought we'd look at quickly.

• Recoloring a turtle's drawing. (Needs improvement.)
• Selecting a polygon.

You can also find a few others on the tips page, such as a procedure to copy and paste.

Common Problems and Algorithms

• As we discussed early in the semester, a key aspect of computer science is the design of algorithms, formalized processes that provide solutions to problems.
• There are a number of common problems for which computer scientists have developed common solutions.
• We'll visit two problems over the next few days: searching and sorting.
• As we develop algorithms, we'll consider intuitive ways that one might come up with the algorithms.

Searching

• Goal: Find a value in a collection.
• Typically, the collection is linear: A vector or list.
• Sometimes, the collection is also unordered. That is, there is no known arrangement to the list. For example, the books on the MathLan book shelves are not in an arrangement that would make it easy to search for a book with a particular title or by a particular author.
• For unordered collections, the typical search is sequential::N search, look at each element in turn.
  • If we distribute things between multiple computers, we can spend the same total computing effort, but significantly less time.
• Sometimes, the collection is ordered. That is, the collection is organized by the primary key in which we search.
  
  • For example, a phone book is sorted by name.
• However, we can also use something known as binary search:
  • Look in the middle of the collection.
  • If the middle is too small, anything smaller is also too small, so discard and try again.
  • If the middle is too large, anything larger is also too large, so discard and try again.
  • If the middle is just right, you're done.

Demonstration: Destructive Binary Search

• We'll do a quick demonstration of binary search

Exploring the Search API

Suppose we have data for various students on campus: Last Name (string), First Name (string), Graduation Year (integer), Box Number (integer), and Phone number (string). We might search by any of the four criteria (and by other criteria) and we might therefore order in various ways.

(define people
  (vector 
    ("Aanderson" "Aan"    2017 4114 "x4410")
    ("Brown"     "Bruin"  2016 8123 "x9000")
    ("Doe"       "J"      2018 9999 "none")
    ("Smith"     "Kieran" 2015 4112 "x9231")
    ("Taylor"    "Mic"    2017 1234 "x0001")))

• Right now, we have things sorted by last name (and, surprisingly, by first name), so we might search by last name (or by first name).
• If we wanted to search by year, we'd order by year (although that's not likely to give particularly useful results).
• If we wanted to search by box number, we'd order by box number.
• Our search algorithm needs to know how to find the key in an entry
  • Last name: car
  • First name: cadr
  • Year: caddr
  • Box number: cadddr
• Our search algorithm needs to know how the keys are ordered
  • Traditionally, strings are ordered alphabetically
  • Traditionally, numbers are ordered numerically, from smallest to largest
• So our binary search algorithm needs four parameters:
  • The vector to search
  • The key to search for
  • The instructions for getting a key from each element
  • The mechanism used to order the elements
• All of this is a bit more complicated if we have a compound key, such as "last name plus first name plus phone number"

Lab

• Do the lab
• Be prepared to reflect.