Блог пользователя Vedkribhu

Автор Vedkribhu, история, 6 лет назад, По-английски

Should I upsolve problems with rating much higher than my current rating? Example my rating is 1600 should I upsolve 2200+ rated problems?

  • Проголосовать: нравится
  • +13
  • Проголосовать: не нравится

»
6 лет назад, скрыть # |
 
Проголосовать: нравится -14 Проголосовать: не нравится

YES! upsolving higher rated probs teaches several new concepts. So for learning, +500-700 would be great. You might not be able to solve on your own, but can learn a lot from editorials, others' solns.

»
6 лет назад, скрыть # |
Rev. 3  
Проголосовать: нравится +8 Проголосовать: не нравится

Yes, it is a good idea, but must also account for time — usually 2200 problems are E problems on Div. 2 contests, and you must first be able to solve A, B, C, and D and still have enough time for E in the end.

TLDR: Yes and no, depending on your situation.

Edit — unless you don't want to solve all of A, B, C, D, then you don't have to account for time as much. It's just that most people prefer to start from A and work to the end.

»
6 лет назад, скрыть # |
 
Проголосовать: нравится +9 Проголосовать: не нравится

In my opinion, not that high. Maybe 1900-2000 rated problems should be ok for you.

»
6 лет назад, скрыть # |
 
Проголосовать: нравится +8 Проголосовать: не нравится

I think it's better to solve problems a bit harder than your rating but not a lot harder. If you can solve 2200+ rated problems, then go ahead with upsolving. But otherwise, it's better to do about 2000 rated problems for a challenge.

You could also takes some hints from the editorial or tags for helping to solve the problem.

»
6 лет назад, скрыть # |
 
Проголосовать: нравится +45 Проголосовать: не нравится

Some problems have a high rating because they require specific technical knowledge. Some advanced tree problems are rated 3000 but aren't so difficult if you have a background understanding of standard techniques like heavy-light decomposition or centroid decomposition. Or a problem might be rated 2000 only because it requires using max flow, but isn't very logically complex if you have max flow available as a black box. For such problems it's completely fine to spend a couple of days understanding the established theory and adding it to your toolbox.

Other problems have a high rating while being purely ad hoc. Even if you read the editorial, you'll probably not be able to understand why the approach used should be in any way natural or intuitive. You'll just be stuck with the feeling of "how did they come up with that?" It's probably better to do easier ad hoc problems as stepping stones with the goal of working your way up to harder and harder problems.

»
6 лет назад, скрыть # |
 
Проголосовать: нравится +4 Проголосовать: не нравится

Idk if it's the best way but I usually upsolve questions only upto 200 rating above my rating or if I gave a lot of time to it during contest and really liked the question. Also I think the ratings of the questions in div3 are a bit inflated (eg. Cyclic Shifts Sorting which has 2400 rating), so I also keep that in mind while upsolving.