Comments on: November Contest Results http://blog.codechef.com/2009/11/11/november-contest-results/ Practice. Compete. Discuss. Sun, 12 Feb 2012 15:11:00 +0000 hourly 1 http://wordpress.org/?v=3.0.1 By: Stephen Merriman http://blog.codechef.com/2009/11/11/november-contest-results/#comment-597 Stephen Merriman Fri, 13 Nov 2009 04:28:46 +0000 http://blog.codechef.com/?p=611#comment-597 I'm not really sure if multiple challenge problems would help, because unless they were very well balanced it would render one of them pointless. For example, if you combined the Help the DJ problem with one of the challenge problems in the earlier contest, then everyone would score so closely in Help the DJ that all you would need is a semi-decent program here, and the other challenge problem would have a bigger range in scores and thus be the sole deciding factor.<br><br>Having a couple of very hard problems would be the best idea like ajaysomani mentioned below, though of course that is very hard to do from Codechef's side of things :)<br><br>I think the challenge problem wasn't the best this time - while it was an interesting problem (and all credit to the top scorers), when you are aiming to improve by 1 solitary point over 20 test cases, a lot of luck is involved; my best improvement was when I had a bug in my code :)<br><br>Still, an enjoyable contest as always. (Definitely need to make sure the time limits are sorted next time though - you'll notice a heap of the solutions, including mine, involved submitting a massive number of solutions to reverse engineer the time limit for each test case. Had to do the same thing last contest too.) I'm not really sure if multiple challenge problems would help, because unless they were very well balanced it would render one of them pointless. For example, if you combined the Help the DJ problem with one of the challenge problems in the earlier contest, then everyone would score so closely in Help the DJ that all you would need is a semi-decent program here, and the other challenge problem would have a bigger range in scores and thus be the sole deciding factor.

Having a couple of very hard problems would be the best idea like ajaysomani mentioned below, though of course that is very hard to do from Codechef's side of things :)

I think the challenge problem wasn't the best this time – while it was an interesting problem (and all credit to the top scorers), when you are aiming to improve by 1 solitary point over 20 test cases, a lot of luck is involved; my best improvement was when I had a bug in my code :)

Still, an enjoyable contest as always. (Definitely need to make sure the time limits are sorted next time though – you'll notice a heap of the solutions, including mine, involved submitting a massive number of solutions to reverse engineer the time limit for each test case. Had to do the same thing last contest too.)

]]> By: Stephen Merriman http://blog.codechef.com/2009/11/11/november-contest-results/#comment-2099 Stephen Merriman Fri, 13 Nov 2009 04:28:00 +0000 http://blog.codechef.com/?p=611#comment-2099 I'm not really sure if multiple challenge problems would help, because unless they were very well balanced it would render one of them pointless. For example, if you combined the Help the DJ problem with one of the challenge problems in the earlier contest, then everyone would score so closely in Help the DJ that all you would need is a semi-decent program here, and the other challenge problem would have a bigger range in scores and thus be the sole deciding factor.Having a couple of very hard problems would be the best idea like ajaysomani mentioned below, though of course that is very hard to do from Codechef's side of things :)I think the challenge problem wasn't the best this time - while it was an interesting problem (and all credit to the top scorers), when you are aiming to improve by 1 solitary point over 20 test cases, a lot of luck is involved; my best improvement was when I had a bug in my code :)Still, an enjoyable contest as always. (Definitely need to make sure the time limits are sorted next time though - you'll notice a heap of the solutions, including mine, involved submitting a massive number of solutions to reverse engineer the time limit for each test case. Had to do the same thing last contest too.) I'm not really sure if multiple challenge problems would help, because unless they were very well balanced it would render one of them pointless. For example, if you combined the Help the DJ problem with one of the challenge problems in the earlier contest, then everyone would score so closely in Help the DJ that all you would need is a semi-decent program here, and the other challenge problem would have a bigger range in scores and thus be the sole deciding factor.Having a couple of very hard problems would be the best idea like ajaysomani mentioned below, though of course that is very hard to do from Codechef's side of things :) I think the challenge problem wasn't the best this time – while it was an interesting problem (and all credit to the top scorers), when you are aiming to improve by 1 solitary point over 20 test cases, a lot of luck is involved; my best improvement was when I had a bug in my code :) Still, an enjoyable contest as always. (Definitely need to make sure the time limits are sorted next time though – you'll notice a heap of the solutions, including mine, involved submitting a massive number of solutions to reverse engineer the time limit for each test case. Had to do the same thing last contest too.)

]]> By: ajaysomani http://blog.codechef.com/2009/11/11/november-contest-results/#comment-596 ajaysomani Thu, 12 Nov 2009 02:30:52 +0000 http://blog.codechef.com/?p=611#comment-596 I'd prefer the idea of having one or two hard problems in the contest, to make problems other then tie breaker have an impact on the rankings. But those problems shouldnt be as hard that nobody solves them (Although this is very unlikely considering the quality of participation from some of the US guys)<br>June and September contest are perfect examples. I'd prefer the idea of having one or two hard problems in the contest, to make problems other then tie breaker have an impact on the rankings. But those problems shouldnt be as hard that nobody solves them (Although this is very unlikely considering the quality of participation from some of the US guys)
June and September contest are perfect examples.

]]> By: ajaysomani http://blog.codechef.com/2009/11/11/november-contest-results/#comment-2098 ajaysomani Thu, 12 Nov 2009 02:30:00 +0000 http://blog.codechef.com/?p=611#comment-2098 I'd prefer the idea of having one or two hard problems in the contest, to make problems other then tie breaker have an impact on the rankings. But those problems shouldnt be as hard that nobody solves them (Although this is very unlikely considering the quality of participation from some of the US guys)June and September contest are perfect examples. I'd prefer the idea of having one or two hard problems in the contest, to make problems other then tie breaker have an impact on the rankings. But those problems shouldnt be as hard that nobody solves them (Although this is very unlikely considering the quality of participation from some of the US guys)June and September contest are perfect examples.

]]> By: gauravagarwal http://blog.codechef.com/2009/11/11/november-contest-results/#comment-595 gauravagarwal Wed, 11 Nov 2009 21:07:46 +0000 http://blog.codechef.com/?p=611#comment-595 Congratulations to all the winners.<br><br>As Balakrishnan suggested the problem set might be a combination of problems of varying difficultly with 1 or 2 problems a bit tricky/harder<br><br>Moreover, I support the idea of only 1 challenge problem per contest. It allows you to focus, trying different strategies and getting better and better solutions. Congratulations to all the winners.

As Balakrishnan suggested the problem set might be a combination of problems of varying difficultly with 1 or 2 problems a bit tricky/harder

Moreover, I support the idea of only 1 challenge problem per contest. It allows you to focus, trying different strategies and getting better and better solutions.

]]> By: gauravagarwal http://blog.codechef.com/2009/11/11/november-contest-results/#comment-2097 gauravagarwal Wed, 11 Nov 2009 21:07:00 +0000 http://blog.codechef.com/?p=611#comment-2097 Congratulations to all the winners.As Balakrishnan suggested the problem set might be a combination of problems of varying difficultly with 1 or 2 problems a bit tricky/harderMoreover, I support the idea of only 1 challenge problem per contest. It allows you to focus, trying different strategies and getting better and better solutions. Congratulations to all the winners.As Balakrishnan suggested the problem set might be a combination of problems of varying difficultly with 1 or 2 problems a bit tricky/harderMoreover, I support the idea of only 1 challenge problem per contest. It allows you to focus, trying different strategies and getting better and better solutions.

]]> By: Balakrishnan http://blog.codechef.com/2009/11/11/november-contest-results/#comment-594 Balakrishnan Wed, 11 Nov 2009 20:52:46 +0000 http://blog.codechef.com/?p=611#comment-594 Congratulations to Josh and Ajay.<br>Thanks again for the wonderful contest problems. I really enjoyed the "Help the Judge" problem. It was really fun coding tests for a 2-SAT problem.<br>I don't know how true this is for the rest of the crowd participating in codechef. But I am really learning an awful lot from these contests.<br><br>Just as a side-note : Having only 1 challenge problem causes a high variance in the rank list. As you can see the difference between the top-10 people is just 0.004. A way to eliminate this kind of problem is to either have multiple challenge problems(may be two) or to make the non-challenge problems much tougher(I do not support this idea). Congratulations to Josh and Ajay.
Thanks again for the wonderful contest problems. I really enjoyed the “Help the Judge” problem. It was really fun coding tests for a 2-SAT problem.
I don't know how true this is for the rest of the crowd participating in codechef. But I am really learning an awful lot from these contests.

Just as a side-note : Having only 1 challenge problem causes a high variance in the rank list. As you can see the difference between the top-10 people is just 0.004. A way to eliminate this kind of problem is to either have multiple challenge problems(may be two) or to make the non-challenge problems much tougher(I do not support this idea).

]]> By: Balakrishnan http://blog.codechef.com/2009/11/11/november-contest-results/#comment-2096 Balakrishnan Wed, 11 Nov 2009 20:52:00 +0000 http://blog.codechef.com/?p=611#comment-2096 Congratulations to Josh and Ajay.Thanks again for the wonderful contest problems. I really enjoyed the "Help the Judge" problem. It was really fun coding tests for a 2-SAT problem.I don't know how true this is for the rest of the crowd participating in codechef. But I am really learning an awful lot from these contests.Just as a side-note : Having only 1 challenge problem causes a high variance in the rank list. As you can see the difference between the top-10 people is just 0.004. A way to eliminate this kind of problem is to either have multiple challenge problems(may be two) or to make the non-challenge problems much tougher(I do not support this idea). Congratulations to Josh and Ajay.Thanks again for the wonderful contest problems. I really enjoyed the “Help the Judge” problem. It was really fun coding tests for a 2-SAT problem.I don't know how true this is for the rest of the crowd participating in codechef. But I am really learning an awful lot from these contests.Just as a side-note : Having only 1 challenge problem causes a high variance in the rank list. As you can see the difference between the top-10 people is just 0.004. A way to eliminate this kind of problem is to either have multiple challenge problems(may be two) or to make the non-challenge problems much tougher(I do not support this idea).

]]>