Promblems of Lunchtimes

I don’t understand the purpose of holding Lunchtimes if the questions are not to be within the scope of the IOI syllabus.

Though I didn’t participate in the December lunchtime, I did go through the problems. Why are there questions on Lucas Theorem and FFT when they are not a part of the IOI syllabus. Having a contest with IOI style grading doesn’t make it a good form of preparation for IOI. The questions need to be of similar kind as well.

If that can’t be ensured, then why not remove this lunchtime.


alt text

@dpraveen I think this is clear enough that only BASIC modular operations are in IOI syllabus. Basic meaning addition, subtraction, and multiplication.


Another problem during December LunchTime was the weak Test Cases for Delivery Man:

Any random solution for “Delivery Man” was sufficient to get 10 points.

The test cases should give AC only for solutions with valid algorithms unlike the ones in LunchTime where any random algorithm was enough to get 10 points in the question “Delivery Man”.

Even an O(N) solution of simply traversing through the arrays, taking Max(A[i],b[i]) and adding it to the total sum, which is a completely wrong algorithm – Was Passing for 10 points , and failing only for 2 test cases.


Where is it mentioned that Lucas’ Theorem is not in IOI syllabus?


@dpraveen on page 4 while not explicitly stated prerequisite knowledge only asks for basic modular arithmetic. Also modular division and inverse elements are explicitly excluded which are much more common than lucas theorem. So it can be concluded that lucas theorem won’t be required to solve IOI problems.


And what is the problem with that? I consider 10p to be ok for a wrong solution. If there are wrong solution with 100p there is a problem.

The fact that a completely wrong solution passes for 8 out of 10 testcases is unacceptable.
It can be really misleading to have a completely Wrong Approach pass for 8 out of the 10 testcases.
Users can be misled to think that they are missing out some edge cases for the actual 100 point AC ,
when in fact – THEY ARE NOT EVEN CLOSE!


@alexvaleanu : “i consider 10p to be okay for a wrong solution”… what do you mean by that? codechef should go passing around points for free to every solution? if a solution is wrong, it shouldn’t get even a single point.


There is a difference between a wrong solution and a solution which was aimed at only one subtask… If a solution is completely right for a smaller subtask, even though it couldn’t pass other subtasks with higher constraints, only then should it get partial points… the issue here was that even LOGICALLY INCORRECT solutions were passing, due to weak test cases… so what if they only got 10 pts… the fact is that each and every point matters in contests like this…and wrong solutions should not be able to pass any subtask, be it small or big… really disappointed in your attitude…

1 Like

@animesh_f You do realize that passing 8/10 test cases is useless if there are groups of tests?

@gvaibhav21 10p is almost nothing so I don’t see the harm if giving them for a source which does something related to the problem.


You aren’t getting the point.

If any solution passes for 8 test cases out of 10 – be it in different groups – one is bound to think that their solution is close to the Correct 100 point Answer .

The only problem is that such weak test cases can easily mislead many users which is not supposed to be done.
If the aim was giving 10 points to everyone – Why not just make those random solutions pass for that ONE 10 PT Subtask and give WA for the rest? Why make it AC for 8 Test Cases ?
It can be really misleading. I hope you have got the point.