Scribd
Upload
7 views2 pages

Lec 3

This lecture discusses recursion in computer science, using the Fibonacci sequence as a primary example. It introduces a method for finding a hidden number using both iterative and recursive approaches, highlighting the inefficiency of a linear search compared to a binary search. The binary search method significantly reduces the number of steps required to find the number, demonstrating the importance of efficient algorithm design.

Uploaded by

sam.kid170
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
7 views2 pages

Lec 3

This lecture discusses recursion in computer science, using the Fibonacci sequence as a primary example. It introduces a method for finding a hidden number using both iterative and recursive approaches, highlighting the inefficiency of a linear search compared to a binary search. The binary search method significantly reduces the number of steps required to find the number, demonstrating the importance of efficient algorithm design.

Uploaded by

sam.kid170
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Addis Ababa University, Amist Kilo July 6, 2011

Algorithms and Programming for High Schoolers

Lecture 3

Recursion: Recursion in computer science solving a problem by solving simpler instantiations


of the same problem. As a classic example, consider the Fibonacci sequence 1, 1, 3, 5, 8, 13, . . .. This
sequence is defined by the 0th and 1st Fibonacci numbers both being 1, and subsequent Fibonacci
numbers being the sum of the previous two.
That is, if Fi represents the ith Fibonacci number,
(
1 if i = 0 or i = 1
Fi =
Fi−1 + Fi−2 otherwise

Now here is an example of using recursion to calculate the nth Fibonacci number. Note that
the fibonacci function calls itself on smaller, i.e. simpler, inputs.

def fibonacci(n):
if n<2:
return 1
return fibonacci(n-1) + fibonacci(n-2)

Let’s give another example. Suppose our friend Bob has a number between 1 and 100 and isn’t
telling us what it is. However, he’s implemented a bool function isGreaterThan(n) for us which
returns True if his number is greater than n and False otherwise. Now, let’s implement a function
searchForBobsNumber().

def searchForBobsNumber():
x = 1
while isGreaterThan(x):
x += 1
return x

We could also implement this function using a recursive helper function.

# We assume the answer is in the range [a,b]


def searchForBobsNumberHelper(a, b):
if isGreaterThan(a):
return searchForBobsNumberHelper(a+1, b)
else:
return a

def searchForBobsNumber():
return searchForBobsNumber(1, 100)

1
Binary Search: If Bob’s number is 100, potentially the while loop in the above implemen-
tation would run for 100 steps before finding Bob’s number. The implementation would be even
slower if Bob’s number could be in the range from 1 to one billion, and unbearably slow if in the
range from 1 to one trillion. One way of remedying this is using a technique known as binary search.
Suppose Bob’s number is between 1 and m and we first ask whether his number is greater than
bm/2c. Then, no matter what his number is, we will eliminate roughly half of the possibilities, and
we can recursively search in the range that is left. That is, consider the following faster version of
searchForBobsNumberHelper(a,b) (remember that int division in Python rounds down to the
nearest int):

# returns True if n equals Bob’s number, and False otherwise


def isEqualTo(n):
return isGreaterThan(n-1) and not isGreaterThan(n)

# We assume the answer is in the range [a,b]


def searchForBobsNumberHelper(a, b):
mid = (a+b)/2
if isEqualTo(mid):
return mid
elif isGreaterThan(mid):
return searchForBobsNumberHelper(mid+1, b)
else:
return searchForBobsNumberHelper(a, mid)

Suppose Bob’s number can be in the range [1, m], and for simplicity, let’s assume m is a power
of 2 (if it isn’t, we’ll just pretend his number can be in the range [1, q] where q is the smallest power
of 2 greater than or equal to m). Since we cut down the number of possibilities in half each time,
and we will get the answer right away if there is only one possibility left, the number of times we
actually need to ask Bob if his number is greater is the smallest number k such that m/2k = 1,
which gives k = log2 m.
Compare this with the while loop implementation in the previous section, which could take m
steps. This is a big difference for large m; note that the log base 2 of one trillion is only about 40.
Many of you are probably familiar with the term “Gigahertz” corresponding to the speed of the
CPU in your computer. This, roughly, corresponds to the number of instructions your computer
can execute in one second, with 1GHz meaning one billion instructions per second (this isn’t quite
what it means, but it’s a good enough approximation for this discussion). Thus, the while loop
implementation for m being one trillion on a 1GHz machine would take at least 1000 seconds, which
is a bit over 16 minutes. In fact, it would take even more time since some instructions take longer,
and some steps which are just one line in Python actually correspond to multiple instructions when
you translate to machine language. Meanwhile, the binary search implementation would execute
almost instantaneously. From now on in the class, this will be our focus when studying algorithms:
finding ways of implementing functions such that they run quickly. Note that the while loop and
binary search implementations in this section both give the right answer, but the latter is vastly
superior when it comes to running time.

You might also like