This week in competitive programming

Last week had a lot of very different rounds.

On Thursday, the long Russia Day weekend was kicked off by TopCoder SRM 624 (problems, results, top 5 on the left, my screencast). My medium problem failed because of the following reason: it asked to find the answer modulo one billion plus nine, whereas I mechanically returned the answer modulo one billion plus seven, since the latter modulo is the standard choice in such problems. That's not a good excuse, of course, and moreover, I actually noticed that the modulo was different when reading the problem statement, but forgot about that when I started coding. Does it mean that coding the "boring" part of the solution happens almost subconsciously?..

This problem didn't stop Gennady who won with a 100+ point gap. Congratulations!

On Friday, Codeforces ran a slightly unusual round - Zepto Code Rush 2014 (problems, results, top 5 on the left). The round had prizes, but somehow that did not attract more contestants than usual: 2148 contestants submitted at least one problem, compared to 2491 in the last Div1/2 round (it happened on Sunday, so that might account for higher turnout). I've managed to do a very stupid thing again: the hardest problem from this round was a slightly easier version of a problem from Petr Mitrichev Contest 11, so I immediately knew how to solve it and could even find working code for the more difficult version. Naturally, I was the first to submit a solution, and it failed the system test - the Codeforces problem allowed items to cost 0, while my problem didn't, and the solution didn't handle those correctly.

Nikolay (KAN) didn't solve this problem but solved all the rest, added five succesful challenges, and won the round - great job!

On Saturday, 25 people won a trip to Los Angeles in Google Code Jam 2014 Round 3 (problems, results, top 5 on the left). The advancers were decided on the two most difficult problems C and D. Problem C was tricky algorithmically, but easy to code; problem D was more straightforward but had a lot of small details and required a lot of time to code. In the end nobody solved problem D in full, so people solving problem C were at the top. Congratulations to Egor on winning this round!

Early on Sunday, the 25 contestants who will go to Berlin were decided in Yandex.Algorithm 2014 Round 3 (results, top 5 on the left).

People going to the final round were decided based on the sum of three rounds, here are overall elimination round results, top 5 on the left. Evgeny (eatmore) won a very convincing first place in the overall results - congratulations! It's interesting that just 7 contestants got in top 25 in both Yandex.Algorithm and Google Code Jam: eatmore, tourist, hos.lyric, dzhulgakov, vepifanov, sdya, Jedi_Knight. Code Jam had a lot more competitors overall, but one might expect that most of the best competitors would participate in both rounds, so the randomness in such selection systems is still very big.

And finally, Internet Problem Solving Contest 2014 rounded up the week (problems, results, top 5 on the left). This is a very exciting competition that has created a format of its own. It happens just once a year and doesn't have any prizes, but it still attracts a lot of contestants, including those who don't participate in any other competitions, like Reid Barton. The reason is very simple: every year it has a very diverse set of problems, and the problems are never boring, but are always interesting and satisfying to solve. It's hard to describe what that means exactly, so I'll just provide links to different types of problems from this year's contest: a very difficult problem that could appear at a "normal" ACM-style contest, a problem requiring to analyze the hashing code in the C++ and Java standard libraries, a problem that asks you to recognize facial expressions - but only on the surface; I'm pretty sure no team who solved this problem actually analyzed the faces. Can you see the trick?

Congratulations to Reid, Tomek and John on an amazing victory!

Thanks for reading, and see you next week (well, in 4 days)!