2018-11-10 17:30 - Events
We hosted a hub for the Battle Dev contest
Tuesday evening there was the Battle Dev, a programming contest with more than 3000 contestants, 2 hours of challenge on 6 programming problems.
To give the opportunity for INSA Lyon students to participate in a friendly atmosphere and achieve good results, INSAlgo’s board decided to host a hub. With pizzas and drinks, our contestants indeed scored high.
INSA de Lyon was ranked 3rd out of 132 schools.
Here are the rankings of some of our contestants (out of 3914, among whom 3009 solved at least one exercise):
- Benoît, 7th
- Paul, 20th
- me, 49th
- Sylvain, 50th
- Arthur, 63rd
- Mathis, 100th
- Alexis, 109th
Critical look at the contest
Even if we like to participate in that kind of event, the quality of this contest is sometimes disappointing. I can easily forgive the long submissions because of the heavy load that the servers undergo, but there are things that I find more troublesome.
Test sets that don’t filter out inefficient solutions
The statement of problem 5 was requesting a polynomial solution (the problem was to find whether it is possible for a tutor to have a course with all heir students, given that every student indicates their ability on two time intervals). After spending a lot of time searching a polynomial solution, I learnt from Paul that a backtracking solution was accepted (though it is, in the worst case, exponential, i.e not polynomial at all). I then implemented the backtracking solution and it was accepted, but later than those who didn’t try to fulfill the statement’s requirements. Almost all the top 100 is affected by this issue.
Moreover, the limit in the statement for the number of students is N <= 1000, but in the test cases it was never reaching 1000. I implemented a recursive function in Python with recursion depth N and it worked, though the recursion limit in Python is very often less than 1000.
The ranking of schools disadvantages schools with large participation
The ranking of schools is calculated from two criteria: the average number of exercises solved by the students, and the number of students in the top 100.
Therefore the winning school is Dublin City University, with only a few participations of contestants who are accustomed to this kind of contest and ranked in top 100. A school like INSA de Lyon, despite having dozens of well-ranked contestants, including multiple contestants ranked better than the best students from DCU, was ranked 3rd because the broad participations lowered the average number of problems solved per contestant.
To win, a school needs to restrict its number of contestants to those who are accustomed to this kind of contest, instead of communicating to all its students. This goes against our will of opening competitive programming to everybody and let people learn problem solving in a playful way.
Anyway, our hub was a success and we will do that again!