Doubt in LIGHTHSE test cases.

Many people in this problem have scored 85/100 points…It means that they have solved the larger set but failed in the smaller set which is itself a subset of the larger set…Can anyone explain how can this happen if the test cases are set properly…(Means smaller subtask is also a part of the larger subtask).Although if what i doubt is correct i would suggest the codechef team to append the smaller subtask of this problem into the larger subtask and rejudging the solutions then…

Might be the case as {(x,0),(0,y)}-{(0,0)} for sub task 1.

I’ve always guessed that contestants do this to get a higher ranking in the contest. The idea is the following: They have code that can solve all of the test cases, but they intentionally make it fail on all but the hardest ones. Therefore, their score is as low as possible while knowing which problems they have completely solved. It’s likely that there will be contestants that have solved fewer problems but have a higher score, and these contestants might become satisfied with their standing and stop solving problems. Then right before the contest ends, the users submit their fully working problems and potentially beat people they might not have. This is just my speculation, but it seems reasonable.

In my opinion, small test cases focus on correctness, while large test cases focus on performance, along with some overflow/out-of-bound memory access issues which only exploited in large test cases. There are corner cases which should be covered by good small test suite, but some of them may appear rarely in large cases, and repeat testing is unnecessary.

We can’t discuss anything regarding the test cases or questions of ongoing contest. Everyone should try himself to find out the problem in his/her code (especially corner cases). Things regarding the test cases and your personal problems should be asked in the comments section below the question from the admin, author or tester.

Also, people (eg @nadimayaz)should refrain themselves from giving hints or some test cases to other users. It is unfair to provide help or test cases during contest.

2 Likes

I think in this question.If I take coordinates of islands as (-4,7)(3,8)(4,-7)(3,-8)(-5,-2).Then there is one possibility of answer is 2 lighthouses required and location is at 3rd NW and 5th SW. But I think software is not checking this condition.is it so??

i don’t think this strategy useful in such long contest. Looking through previous leaderboard, top rank coders always solve all problems. Only the score of the challange/tie-breaking problem decides the final rank. So it is better to optimize for it. This strategy may be used in short contests, but faces time penalty issue, which is critical in some cases.

I don’t think it is really useful either, but you can’t perform any worse in long contests by doing it. In short contests, this would kill your score due to submission penalties and late final submissions. Anyway, that’s just my guess as to why people do it.

I think it’s more of a score board discussion / strategy guessing / test case generation in general, instead of specific test case discussion.