Assessment Title: Distribution of Software Generated Pseudo Random Numbers
Table 1 summarizes the module assessment component and their respective ratios.
Table 1: Elements of the Module Assessment
Element |
Weight |
A.1 Individual part |
|
A.2 Group Project (Demo and presentation) |
This is the element A.1 with a ratio of 25% of the overall module assessment. It consists in implementing a simple pseudo random number generator in C language using the rand() function of “stdlib.h” and writing test programs to assess its quality. Please notice that in this task we are not seeking generating real random numbers but just interested in investigating the distribution of pseudo random numbers generated by the rand() function. The tests you are going to implement were theoretically suggested by Kendall & Babbington-Smith over a series of papers published in 1938/39.
The rand() function returns a pseudo-random integer between 0 and RAND_MAX, which is defined in stdlib.h. You should NOT replace it with a constant since its value may vary with platforms. You need to rescale this value and fit it between 0 and 10 (0 included, 10 excluded). To generate a single pseudo random number between 0 (included) and 10 IN YOUR code, use the following.
randon_number = (int) (10.0*rand()/(RAND_MAX + 1.0));
Write a single code, randTest.c, that uses the instruction above to generate every single number and implements all the tests given in table 2. Use the parsing command line arguments through the main function (with a single character parameter after the file name) to select and perform a single test in the runtime. E.g., suppose the executable file is named, ranTest.out, if in the runtime one wishes running the first test, they just have to type “ranTest.out 1”, if the second, “ranTest.out 2”, etc.
Table 2: Tests
Test |
Deception |
1. Mean test |
Calculate the mean of 1000 random numbers. The result should be close to 4.5. |
2. Frequency test |
Tabulate the percentage of each digit (0 throughout 9) in 10,000 generated numbers. Each Is expected to be close (roughly) to 10%. |
3. Serial test |
Generate 10,000 pairs of numbers. Tabulate the frequency of each pair, 00, 01,02, ….., 99. This time we would expect roughly 1% in each category. |
4. Poker test |
Generate four digits at a time, and repeat 1,000 times (a thousand sets of 4 digits). Tabulate the sets as: i) all the same (e.g. 4444), ii) 3 digits the same (e.g. 4443, 3444 or 4344), iii) two pairs (e.g. 4334, 4433), iv) one pair (e.g. 4324 or4342), or v) none identical. In theory we would expect frequencies of 1, 36, 27, 432 & 504 respectively for a thousand sets of numbers (i.e., 4,000 digits). You will not necessarily get those exact numbers. |
5. Pocker Test using a file |
Generate 4,000 random digits and save them in a text file (using ASCII coding), where characters are separated by spaces. Repeat the poker test by reading the numbers from the file. |
For this part (Part A.1: individual E-portfolio), please upload: