Feedback Lab 6

[ana-oprescu]

Exercise 1	Comments	Exercise 2	Comments	Exercise 3	Comments	Exercise 4	Comments	Exercise 5	Comments	Exercise 6	Comments	Exercise 7	Comments	Exercise 8	Comments	Total
10	Excellent	5	Results are not being reported with sufficient depth. No means, Stdev, number of runs, figures, etc.	9	Why not starting in 4? Since you know that 2 and 3 are not composites	7	Very good, can be improved by continuing the test even after finding the first false positive, and printing all the found false positives	6	Analysis is ok. In your analysis you are not taking into account that a Carmichael number N fools the Fermat test if you check aˆ(N−1) for some a that is coprime with N. An analysis/test on that direction is needed	6	->issues	6	->issues	8	-> issues	7.8

[TerryvanWalen]

Exercise 2: Check that your implementation is more efficient than expM by running a number of relevant tests and documenting the results.

In the exercise they ask about relevant tests and documenting the results. We are wondering if it is really relevant to calculate the mean and stdev if the differences (between implementations) are in different orders of magnitude for a few runs?? We don't understand what the value is of providing that information when results (same sample runs) are close together and miles form the results of the other implementation.

[ana-oprescu]

Explained :) Dear Hugo, Bert, Can you check Terry's remark? thank you, Ana

…

On Mon, Oct 23, 2017 at 5:46 PM, Terry van Walen ***@***.***> wrote: Exercise: Check that your implementation is more efficient than expM by running a number of relevant tests and documenting the results. In the exercise they ask about relevant tests and documenting the results. We ar wondering if it is really relevant to calculate the mean if the differences are in different orders of magnitude for a few runs?? We don't understand what the value is of providing that information when results are close together and miles form the results of the other implementation. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#11 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABoHG0POJ5j2rYaaYYdAiDOXPrgSgzVmks5svLTfgaJpZM4QCazn> .

[TerryvanWalen]

Exercise 3:
We are wondering why the line is drawn at 3? We also know that 5, 7, 11, 13, 17, 19 are not composites. It's easier to write [2..] than excluding these other numbers but it is not that difficult. It just doesn't make any sense to exclude them from a programming point of view. It costs almost no processing power and you still have to prove for yourself that indeed these are not composites (this is a really simple example). This process of checking it in your head takes to much time in practice instead of simply checking everything.

[ana-oprescu]

Think of the \delta impact on a query for 100k numbers sequentially :)

[hdmeyer]

Hello,

The second question was already answer by Ana :).

About the first one:
The analysis that you've presented is very short/incomplete when comparing with other reports presented (For example: https://github.com/software-engineering-amsterdam/ST2017_WG_2/blob/master/week6/final/ex2.hs). So, we cannot increase the grade of this exercise, since other groups made a major effort in analysing the results.

Also, the justification that you've has just mentioned was not part of your submission.