In this series of exercises, you will write a script that performs some measurements that justify the need to sort a vector.
Background:
• Since this need is only apparent when processing large amounts of data, we will use a vector of at least 100,000 elements.
• MATLAB has two built-in functions that collectively provide stopwatch capability: tic and toc. The function tic starts the stopwatch, and toc returns the number of seconds since the last tic.
• However, since the time difference is measured to the resolution of milliseconds, we will have to repeat the experiments a number of times to achieve accurate measurements.
• Since this involves a for loop that itself will cost time, we must also measure the overhead of the for loop and subtract that from the computational cost of the operation we are timing.
a. Begin your script by deciding on the vector size, N, and the number of repetitions, reps, and making a vector, vec, of random integers between 0 and N.
b. Write an empty function, junk, that consumes a vector of size N and a number and returns a number.
c. Use tic and toc to compute the time for a for loop that does nothing but pass vec and round(N*rand(1,1)) to junk. Adjust the value of reps until this loop takes about a second to run. Store this cost in the variable overhead.
d. Write a function, linearSearch, that consumes a vector and a number and performs a linear search on the vector looking for that number. It should return either the position of that number in the vector or an empty vector if the number is not present.
e. Add a loop to your script like the loop in step c, except calling linearSearch. Subtract overhead from the computed time and save it as linearSearchTime.
f. Now sort the vector vec using the MATLAB sort(...) function and store it in a new vector, sortedVec.
g. Write the function binarySearch that consumes a vector and a number and uses the algorithm in section 16.1.1 to find the position of the number in the vector returning its position or the empty vector. h. Now, repeat step e, calling your new binarySearch function, saving the time as binarySearchTime.
i. Plot the times of the linear and binary searches for a range of values of N. What do you learn from these plots?