Recent Forum Posts
From categories:
Alexmaniac (guest) 04 Aug 2014 12:27
in discussion Hidden / Per page discussions » JPA: Pros and Cons

Using JPA is BDSM.

by Alexmaniac (guest), 04 Aug 2014 12:27
It's me (guest) 14 Jul 2014 13:24
in discussion Hidden / Per page discussions » Too...

Hey, cheer up! There are other flowers in the fields.

by It's me (guest), 14 Jul 2014 13:24
Zulan (guest) 19 Aug 2013 10:04
in discussion Hidden / Per page discussions » ICFPC 2013

We used C++(11). I think it's not too bad for this particular problem. You could probably use lazy languages very well here, but then you have to be very fluent in that language and understand it well enough to specify the algorithm in a way that the language can solve efficiently.
The funny thing is that we did have some serious computing power at our disposal, but didn't use it because we spent all the time on our futile effort to improve the algorithm. So in the end the solver ran almost exclusively on a 5 yr old laptop. I am not sure how much we could have achieved with throwing CPUs at it. I assume the complexity growth is pretty high if the "actual size" increses, so there may be only a small band of problems that can be solved with more CPU. Also if you look at the unofficial stats: labs. skbkontur. ru/icfpc2013/Stats , the high ranking teams still have lots of unsolved 'normal' problems, but get their points from the bonus problems.

by Zulan (guest), 19 Aug 2013 10:04
Sir Digby Chicken Caesar (guest) 17 Aug 2013 15:16
in discussion Hidden / Per page discussions » ICFPC 2013

Ah, yes, if you don't limit search space by problem constraints (at least the set of ops) then it becomes huge earlier.

by Sir Digby Chicken Caesar (guest), 17 Aug 2013 15:16
IcePhoenixIcePhoenix 16 Aug 2013 19:10
in discussion Hidden / Per page discussions » ICFPC 2013

We used C++ (it is subtly mentioned in the write-up), and size limit 7 applies only to the very first super-naive Bruteforcer. Main problems we encountered when raising it were:

  • Lack of laziness
  • Lack of pattern matching

Basically, all our Bruteforcers explored all possible schemas (no reduction-by-equivalence) at once (no laziness). After carefully exploring other optimizations, we raised the size limit to 9 for complete search (all possible schema components) and 11-12 for bounded search (schema components constrained by the problem).

Hope that helps to clear things up =)

by IcePhoenixIcePhoenix, 16 Aug 2013 19:10
IcePhoenixIcePhoenix 16 Aug 2013 18:57
in discussion Hidden / Per page discussions » ICFPC 2013

Hey, thanks for reading! Glad my post was of any use to someone out there =)

Fantastic results, mate, simply fantastic! What PL did you use though? Eager languages tend to complicate a lot of issues =)

PS: I didn't like the task, but that's because I prefer ICFPCs where you submit solvers, not solutions. Makes it a lot more fair imho.

by IcePhoenixIcePhoenix, 16 Aug 2013 18:57
Sir Digby Chicken Caesar (guest) 16 Aug 2013 14:47
in discussion Hidden / Per page discussions » ICFPC 2013

What language did you use? Bruteforcers in Haskell and D and some other langs easily cracked problems of size 12-13. I don't understand how you manage to hit a wall at 7-8.

by Sir Digby Chicken Caesar (guest), 16 Aug 2013 14:47
Zulan (guest) 15 Aug 2013 13:13
in discussion Hidden / Per page discussions » ICFPC 2013

Thanks for the writeup! We actually had a really similar experience. We started with a simple bruteforce and gradually tried to improve it making it more clever. Unfortunately in the end we got lost so badly and the time was running low that we had to use a very early version of the bruteforce with only minor improvements (iterativeDFS, commutative redundancy elimination). This however worked surprisingly well for all except the bonus problems. We were able to solve 80% despite an extremely tight time limit - we stopped each problem if we couldn't solve it in 10 seconds. Unfortunately we also didn't have the time to write a proper threaded scheduling - naive parallel work would have just lead to massive request throttling by the server.

We also couldn't quite finish our adaption to the bonus problem. In the end I think we would have had a great score if we only focused on the parallel implementation, started the solver early and work out the bonus problem. Instead we tried to chase the perfect algorithm that doesn't fail at a given size.

I also found it really surprising how much fun we had, considering the boring task. I must really say that I was disappointed by MSRE despite not really having high hopes. Seriously how can you come up with "try to guess a function" after "Dig through a sand stone computer", "Rescue an Alien by hacking DNA", .. "Play a functional card game", "Dig lambdas in a mine".

by Zulan (guest), 15 Aug 2013 13:13
JohnGluck (guest) 01 Jan 2012 16:30
in discussion Hidden / Per page discussions » Happy New Year 2012

You to, lad. :)

by JohnGluck (guest), 01 Jan 2012 16:30
Yulya (guest) 19 Sep 2011 21:44
in discussion Hidden / Per page discussions » Finally, I'm back!

Glad you had such a good internship Marat! Saw some blog post about your work. Pretty sweet!

by Yulya (guest), 19 Sep 2011 21:44
Incubos (guest) 07 Jul 2011 20:14
in discussion Hidden / Per page discussions » Redmond, WA - 2:10

Only Mac and don't care. ;-)

by Incubos (guest), 07 Jul 2011 20:14
TheBits (guest) 07 Jul 2011 06:28
in discussion Hidden / Per page discussions » Redmond, WA - 2:10

Марат, откройте своё сердце MSVS и TFS. Надо понять философию Microsoft, после чего все программы для разработчиков станут элементарными.
Я бы выбрал Thinkpad. :)

by TheBits (guest), 07 Jul 2011 06:28
vlan (guest) 07 Jul 2011 06:26
in discussion Hidden / Per page discussions » Redmond, WA - 2:10

What's the other VCS you're using? It is hg or git?

Thinkpad because you can use Arch with it! :)

by vlan (guest), 07 Jul 2011 06:26
vlan (guest) 08 Apr 2011 23:31
in discussion Hidden / Per page discussions » JPA: Pros and Cons

"I'm noSQL wizard" :)

by vlan (guest), 08 Apr 2011 23:31

Thanks for the report! :)

by Andrey Vlasovskikh (guest), 24 Jun 2010 01:45
Reit (guest) 18 Feb 2010 19:17
in discussion Hidden / Per page discussions » If you wanna do...

Слава Силе, расконсервировал блог=)
Да, я злостный вредитель и буду чепятать по-русски!

А "мудрость" утащу к себе, я в последнее время мало что успеваю…

by Reit (guest), 18 Feb 2010 19:17
Do comments work in here?
Hidden entity (guest) 29 Dec 2009 18:23
in discussion Hidden / Per page discussions » Some info about this blog

Seems so!

Do comments work in here? by Hidden entity (guest), 29 Dec 2009 18:23
Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License