Go IIIT-H .Apart from some weak test data, the problems are very nice. Thank you codechef !

Go IIIT-H
.
Apart from some weak test data, the problems are very nice. Thank you codechef !

Yes, Nice problems.I would like to thank CodeChef for the change in the price structure. It will motivate more programmers to participate and improve.Congratulations to all winners.

Yes, Nice problems.
I would like to thank CodeChef for the change in the price structure. It will motivate more programmers to participate and improve.
Congratulations to all winners.

I hope codechef will prove wrong the myth about IITs being the best institute in the country for computer science.

I hope codechef will prove wrong the myth about IITs being the best institute in the country for computer science.

yes IITans might have scored more in the IITJEE exam,but it takes quite a lot to hold on to that performance.Here it is down to developing efficient computational methodologies to solve wonderful real life problems.Its a different ball game altogether.

yes IITans might have scored more in the IITJEE exam,but it takes quite a lot to hold on to that performance.Here it is down to developing efficient computational methodologies to solve wonderful real life problems.Its a different ball game altogether.

According to me a college has very little impact on a person's abilities. So saying a college is performing good/bad almost always has no meaning for me.

According to me a college has very little impact on a person's abilities. So saying a college is performing good/bad almost always has no meaning for me.

Yes. But, unfortunately, that is true only in India

Yes. But, unfortunately, that is true only in India

I think environment also matters. In IIIT, you may have better coding environment. But, here in IITs, programming environment is not that conducive. But, CodeChef is helping in changing that. And in future, persons from IITs may also perform well.

I think environment also matters. In IIIT, you may have better coding environment. But, here in IITs, programming environment is not that conducive. But, CodeChef is helping in changing that. And in future, persons from IITs may also perform well.

Firstly, thanks for an excellent contest. The problems were all extremely nice, and I spent far too many hours in the last 10 days thinking about them.There were some downsides though, and I just want to point them out to make future contests better. I still enjoyed this one immensely, so don't take any of this wrong way.a) The test cases seemed very weak for nearly all of the problems. For example:- it seems a large number of people solved Soccer League with incorrect solutions.- while I had an O(n log n) solution for Overlapping discs, to test the valid of my solution I submitted an O(n log n + H^2) solution , and was extremely surprised to find that it passed. This should never have passed and would have been easy to test.- my solution for Toll Trolls also had a worst case time complexity that wouldn't have passed the time limit.Granted, it is impossible to design worst cases for every algorithm someone could think of, but it feels like the test cases weren't very well thought out for this set of problems.b) There was a very valid clarification request in Motorbike Racing (whether input were integers), that was completely ignored by the admin. Again, this should have been covering in testing, but even if it had slipped through, it needs to be answered ASAP. It was asked on the 1st of February, and then asked about 6 of 7 more times after that – and nobody replied!c) What on earth happened with overlapping discs? The initial problem statement was wrong; the sample output was wrong, and the official solution didn't run inside the time limit! And then another problem had its time limit changed. These are very very fundamental flaws and should have easily been picked up by testers. These sort of basic mistakes are very offputting – Tomasz has already commented somewhere that he waits until part way through the contest to make sure all bugs are ironed out. That doesn't give Codechef a very good look. (I would really like to know what happened with overlapping discs).Overall, I know that small mistakes can always slip through extensive testing, but the kinds of mistakes slipping through (for all contests I have been involved in) aren't small ones. I almost get the feeling testing is non-existent (and nobody has ever really replied to say otherwise) – if you want to run a good contest, you must test thoroughly.On another note, something needs to be done about comments on problems. I already a month ago in the forum that a link to the FAQ be put directly under 'Post a comment', but it appears this was ignored/rejected. Even that may not be sufficient though.As a suggestion, perhaps you could turn on moderation for contest problem comments only? In other words, comments don't show up until an admin has seen them and approved them; any silly comments asking for test cases or just saying 'what have I done' wrong could be trashed without even showing up. The only downside of this approach is that the admins would need to be constantly available to moderate comments, and timezones would cause a problem. So how about [and sorry, being a web developer myself, I can't help lapsing into Drupal-talk ] creating a new user role for 'trusted' members of the community; these people would have the ability to moderate comments themselves. That takes the pressure off the admin, and would mean if any clarifications are required, the problem setter would just need to reply to each approved comment under the problem without having to scan through hundreds. I would be happy moderating comments myself if you agree it would be a beneficial idea.

Firstly, thanks for an excellent contest. The problems were all extremely nice, and I spent far too many hours in the last 10 days thinking about them.

There were some downsides though, and I just want to point them out to make future contests better. I still enjoyed this one immensely, so don't take any of this wrong way.

a) The test cases seemed very weak for nearly all of the problems. For example:

- it seems a large number of people solved Soccer League with incorrect solutions.
- while I had an O(n log n) solution for Overlapping discs, to test the valid of my solution I submitted an O(n log n + H^2) solution , and was extremely surprised to find that it passed. This should never have passed and would have been easy to test.
- my solution for Toll Trolls also had a worst case time complexity that wouldn't have passed the time limit.

Granted, it is impossible to design worst cases for every algorithm someone could think of, but it feels like the test cases weren't very well thought out for this set of problems.

b) There was a very valid clarification request in Motorbike Racing (whether input were integers), that was completely ignored by the admin. Again, this should have been covering in testing, but even if it had slipped through, it needs to be answered ASAP. It was asked on the 1st of February, and then asked about 6 of 7 more times after that – and nobody replied!

c) What on earth happened with overlapping discs? The initial problem statement was wrong; the sample output was wrong, and the official solution didn't run inside the time limit! And then another problem had its time limit changed. These are very very fundamental flaws and should have easily been picked up by testers. These sort of basic mistakes are very offputting – Tomasz has already commented somewhere that he waits until part way through the contest to make sure all bugs are ironed out. That doesn't give Codechef a very good look. (I would really like to know what happened with overlapping discs).

Overall, I know that small mistakes can always slip through extensive testing, but the kinds of mistakes slipping through (for all contests I have been involved in) aren't small ones. I almost get the feeling testing is non-existent (and nobody has ever really replied to say otherwise) – if you want to run a good contest, you must test thoroughly.

On another note, something needs to be done about comments on problems. I already a month ago in the forum that a link to the FAQ be put directly under 'Post a comment', but it appears this was ignored/rejected. Even that may not be sufficient though.

As a suggestion, perhaps you could turn on moderation for contest problem comments only? In other words, comments don't show up until an admin has seen them and approved them; any silly comments asking for test cases or just saying 'what have I done' wrong could be trashed without even showing up. The only downside of this approach is that the admins would need to be constantly available to moderate comments, and timezones would cause a problem. So how about [and sorry, being a web developer myself, I can't help lapsing into Drupal-talk ] creating a new user role for 'trusted' members of the community; these people would have the ability to moderate comments themselves. That takes the pressure off the admin, and would mean if any clarifications are required, the problem setter would just need to reply to each approved comment under the problem without having to scan through hundreds. I would be happy moderating comments myself if you agree it would be a beneficial idea.

Oh, one more thing I forgot to mention. In several problems recently time limits are very confusing due to numerous test cases in a file. For example, If a problem consists of, say 10 test cases, and the time limit is 5 seconds, I would expect a maximal input size is required to run in 0.5 seconds.However, often small test cases + large test cases are combined; so a worst-case 2 seconds per input may suffice due to there not being 10 worst-cases in the same file.This needs to be avoided at all costs. Otherwise it becomes a guessing game as to what sort of test cases you are providing us.

Oh, one more thing I forgot to mention. In several problems recently time limits are very confusing due to numerous test cases in a file. For example, If a problem consists of, say 10 test cases, and the time limit is 5 seconds, I would expect a maximal input size is required to run in 0.5 seconds.

However, often small test cases + large test cases are combined; so a worst-case 2 seconds per input may suffice due to there not being 10 worst-cases in the same file.

This needs to be avoided at all costs. Otherwise it becomes a guessing game as to what sort of test cases you are providing us.

Congrats to all winners… You guys (?) rock!

Congrats to all winners… You guys (?) rock!

I agree completely to what Stephen has to day. I did not solve any of the problems this contest. However, the fact that incorrect solutions have passed, and solutions with terrible complexity have also passed seems really weird to me. And that should not be the case.Whether or not a solution passes should depend more an the Problem Statement than on the Test Data. Please test every problem (and their test cases) extensively before using them for contests. Whats the whole point of organizing the contests if non-deserving solutions are deemed correct? Even I wait for a minimum of three days before I start coding the solution to a problem, since I would feel hyper-frustrated if I saw a problem statement change after writing the code for it. For the test data, you could even allow something like, every contestant who has solved the problem is allowed to add a single test data file for the solved problem. Note that this should be done only in the extreme case if the admins agree that the constraints they have does not allow them to build the best possible set of input data by themselves. What I have suggested, is just a last case measure, which is the least that should be done. Thank You.

I agree completely to what Stephen has to day.

I did not solve any of the problems this contest. However, the fact that incorrect solutions have passed, and solutions with terrible complexity have also passed seems really weird to me.
And that should not be the case.

Whether or not a solution passes should depend more an the Problem Statement than on the Test Data. Please test every problem (and their test cases) extensively before using them for contests.

Whats the whole point of organizing the contests if non-deserving solutions are deemed correct?

Even I wait for a minimum of three days before I start coding the solution to a problem, since I would feel hyper-frustrated if I saw a problem statement change after writing the code for it.

For the test data, you could even allow something like, every contestant who has solved the problem is allowed to add a single test data file for the solved problem. Note that this should be done only in the extreme case if the admins agree that the constraints they have does not allow them to build the best possible set of input data by themselves. What I have suggested, is just a last case measure, which is the least that should be done.

Thank You.

For comments apart, it can be done that if a person has solved a problem in previous contests ( then, he knows how Codechef works ), then his comments will appear instantly …. otherwise his comments will be moderated …

For comments apart, it can be done that if a person has solved a problem in previous contests ( then, he knows how Codechef works ), then his comments will appear instantly …. otherwise his comments will be moderated …

when and how will we receive the prize money??

when and how will we receive the prize money??

Stephen said-”This needs to be avoided at all costs. Otherwise it becomes a guessing game as to what sort of test cases you are providing us.”I am agree with Stephen.My lot of submissions were to test about “test cases”.In problem “Toll Trolls” , my solution can not pass a worst test case.I checked availability of that test case and found that there is nosuch type of test case.i got AC after taking some conditions with many submissions.

Stephen said-
“This needs to be avoided at all costs. Otherwise it becomes
a guessing game as to what sort of test cases you are providing us.”

I am agree with Stephen.
My lot of submissions were to test about “test cases”.
In problem “Toll Trolls” , my solution can not pass a worst test case.
I checked availability of that test case and found that there is no
such type of test case.
i got AC after taking some conditions with many submissions.

hi,congrats 2 all winners What abt the test cases ?

hi,
congrats 2 all winners
What abt the test cases ?

Hey balajiganapathiBasil from the CodeChef team will be sending all the winners an email by the end of today informing them on how they will be receiving their prize money. So make sure you check your email and congratulation on coming 19th!

Hey balajiganapathi
Basil from the CodeChef team will be sending all the winners an email by the end of today informing them on how they will be receiving their prize money.
So make sure you check your email and congratulation on coming 19th!

Since the January contest we have decided to not publish the test data. The successful solutions, however, are visible.

Since the January contest we have decided to not publish the test data. The successful solutions, however, are visible.

Congrats to all the winners. Unfortunately, I did not have enough time to compete through the contest. But, even in the best scenario, I could have solved only 2 more. Hats off to all those who were able to make through all the problems. The motorbike racing was really hard. But on the other hand, I would like to know if codechef is making any progress on the week long contest as suggested by Ajay last month. But please note that such week-long contests would demand lower difficulty(of the order of the late 2009 problems) and clarity of problem statements right from the start of the contest. It will be a disaster to have a week long contest and have the problem statement/test cases wrong to begin with. I hope such initiatives will be considered by codechef.

Congrats to all the winners. Unfortunately, I did not have enough time to compete through the contest. But, even in the best scenario, I could have solved only 2 more. Hats off to all those who were able to make through all the problems. The motorbike racing was really hard.

But on the other hand, I would like to know if codechef is making any progress on the week long contest as suggested by Ajay last month. But please note that such week-long contests would demand lower difficulty(of the order of the late 2009 problems) and clarity of problem statements right from the start of the contest. It will be a disaster to have a week long contest and have the problem statement/test cases wrong to begin with. I hope such initiatives will be considered by codechef.

college matters only in starting to know what the programming contests are are how to do in these. But after that college doesn't matter anymore. Also by looking price structure we are very exited to compete monthly competition. previouIsly we had mind set that we will never win. But know i too can say I can win!!

college matters only in starting to know what the programming contests are are how to do in these. But after that college doesn't matter anymore. Also by looking price structure we are very exited to compete monthly competition. previouIsly we had mind set that we will never win. But know i too can say I can win!!

I too thought that contest is very long. If it can be 5-7 days then it is fine. We can work on it. But for 10-11 days its very long and we have to study our other subjects also, so we cann't spend our whole time in it only,

I too thought that contest is very long. If it can be 5-7 days then it is fine. We can work on it. But for 10-11 days its very long and we have to study our other subjects also, so we cann't spend our whole time in it only,

Pieguy rocks. he is just too good. See his submissions. Just one submission for every problem. He solved all toughest problem in one submission and I know he always comes up with the best algorithm. But still how is it possible.How do you do it. You really rock pieguy.

Pieguy rocks. he is just too good. See his submissions. Just one submission for every problem. He solved all toughest problem in one submission and I know he always comes up with the best algorithm. But still how is it possible.How do you do it. You really rock pieguy.

May be before submitting he test his solution on some of his own generated test cases.

May be before submitting he test his solution on some of his own generated test cases.

I think its the other way round. It gives a lot of flexibility if you have other things to do. You can choose on what days you will be denoting time to these problems. Or you can just finish up with your submissions in 1 week and leave.

I think its the other way round. It gives a lot of flexibility if you have other things to do. You can choose on what days you will be denoting time to these problems. Or you can just finish up with your submissions in 1 week and leave.

Agreed. 5-7 days isn't enough for people that are busy with other things. You don't need to work on the problems 24/7 for 10 days if you don't want to.

Agreed. 5-7 days isn't enough for people that are busy with other things. You don't need to work on the problems 24/7 for 10 days if you don't want to.

Also, the contest can start on the 1st Friday of the month, so that we get 2 full week-ends to solve.

Also, the contest can start on the 1st Friday of the month, so that we get 2 full week-ends to solve.

The IITs get equal magnitude of analytically thinking brains as IIITs if not more.But I guess IITans spend lesser time to procedural programming and programming contests than the IIITans,may be because IITans get to keep themselves busy in Internships,Projects and all a little bit more than the IIITans.

The IITs get equal magnitude of analytically thinking brains as IIITs if not more.But I guess IITans spend lesser time to procedural programming and programming contests than the IIITans,may be because IITans get to keep themselves busy in Internships,Projects and all a little bit more than the IIITans.

obvious buddy.do u submit without even running once?

obvious buddy.do u submit without even running once?

Submitting without running ,LOL . So you think not testing on some of our own generated test cases is equivalent to submitting without running.Well I test my solution only on the cases provided by the question.I know that's very bad idea and i may fetch too many wrong submission but I don't care as there's no negative penalty. What I actually meant is exhaustive testing which I am sure you don't do for most of the problems.

Submitting without running ,LOL . So you think not testing on some of our own generated test cases is equivalent to submitting without running.Well I test my solution only on the cases provided by the question.I know that's very bad idea and i may fetch too many wrong submission but I don't care as there's no negative penalty.

What I actually meant is exhaustive testing which I am sure you don't do for most of the problems.

thats true. but still one submission is just too great to me. He can think about every corner case , generates data , tests it. Moreover the algo is so efficient that he has not got a TLE. thats pure ability and I admire it

thats true. but still one submission is just too great to me. He can think about every corner case , generates data , tests it. Moreover the algo is so efficient that he has not got a TLE. thats pure ability and I admire it

I appreciate the praise, however I should note that I made 3 submissions on Toll Trolls, all of which were incorrect (the last one passed because of weak test cases). For the 3 hard problems, it was extremely easy to write both random testcase generators and brute force checkers, so I figured there was no reason not to do offline testing.

I appreciate the praise, however I should note that I made 3 submissions on Toll Trolls, all of which were incorrect (the last one passed because of weak test cases). For the 3 hard problems, it was extremely easy to write both random testcase generators and brute force checkers, so I figured there was no reason not to do offline testing.