BubbleSort – Some simple Performance measurements
The goal for this exercise is to measure the performance of Bubble Sort. The
longer-term goal is to compare this performance to other, alternative algorithms
(such as selection sort, QuickSort, MergeSort, HeapSort, etc) so that you will
have a clear understanding of why the Big O notation is useful.
Much like you did with the two searching methods, you're now going to
instrument your sorting method so that you can measure it's performance. Copy
the code you wrote in the previous exercise, and name it
BubbleSortPerfMeasured. It should take the extra, out parameter numSwaps,
much like the other PerMeasured methods do. When this BubbleSort starts, it
should initialize that variable to be zero. It should increment (increase by one)
that parameter each time it swaps any two elements. You should also provide
a second out parameter, that gets incremented each time it compares any two
elements (regardless of whether it swaps them or not). We’ll be collecting up
both these sets of information because the number of comparisons is a good
measure of how much work the algorithm is doing overall, and the number of
swaps is a more specific indicator of exactly how much work this algorithm is
doing (especially if you want to compare the efficiency of this algorithm to a
similar sorting algorithm, such as Selection Sort)
So BubbleSortPerfMeasured was given the following array, and was
asked to sort it into ascending order, it would set numSwaps to be 1 (since
you'd need to swap the 6 & the 10 once, in order to end up with a sorted array),
and numComparisons to be 1.
Index: 0
Value: 10
Index: 1
Value: 6
Alternately, if the array looked like this:
Index: 0
Value: 6
Index: 1
Value: 10
you'd still have numComparisons being 1, but numSwaps would be 0,
since it had never swapped anything.
Once you’ve done that, you should test the function (once you've written
it), using the code that’s provided to you in the NUnitTests_BubbleSort_Measured
class, which you can run using the NUnit support in the provided starter project.
You should feel free to supplement that code with your own test cases if you
wish, but you are neither required nor expected to.
What you need to do for this exercise:
1. Implement the BubbleSortPerfMeasured method, within the
SearchingAndSorting class.
a. Make sure that your method can handle being passed a null parameter
for the array. In this case, you can simply zero out the comparison /
swap counters and then return.
b. Note that you don't (technically) need to complete the "BubbleSort"
exercise in this same lesson – you can jump straight to this exercise.
However, many people find it easier to do that exercise first, then copyand-paste that code into this exercise.
2. Once you’ve done this, all the tests in the NUnitTests_BubbleSort_Measured
class should pass. This class is located in the starter solution, in the
PCE_ForTests project, in the file named PCE_Test.cs.
a. You may need to set another project as the start up project (such as
the PCE_Test_Runner project), by right-clicking on that other project,
and selecting the “Set As Startup Project” menu option. Within the test
runner project, you may (or may not) need to change the code in the
RunTests.cs file, particularly at the top of the Main method.
b. Keep an eye on the results of the ‘EqualsFuzzy’ test – if the test is
failing because you’re getting a wildly different number of comparisons,
and yet you think that the number of comparisons is correct, then talk
to the instructor, and/or post a question to the Google Group.